Southern Cross University

ePublications@SCU
Theses

2010

The retention of software development employees in the IT industry in Taiwan

Chin-Yao Tseng
Southern Cross University

Publication details
Tseng, CY 2010, 'The retention of software development employees in the IT industry in Taiwan', DBA thesis, Southern Cross University, Lismore, NSW. Copyright CY Tseng 2010

ePublications@SCU is an electronic repository administered by Southern Cross University Library. Its goal is to capture and preserve the intellectual output of Southern Cross University authors and researchers, and to increase visibility and impact through open access to researchers around the world. For further information please contact epubs@scu.edu.au.

The Retention of Software Development Employees in the IT Industry in Taiwan

August 2010

Chin-Yao Tseng Master of Business Administration (Griffith University, Australia) Bachelor of Business Management (Griffith University, Australia)

Submitted to The Graduate College of Management Southern Cross University, Australia In Fulfilment of the Requirements for the Degree of Doctor of Business Administration

DECLARATION OF ORIGINALITY
I, Chin-Yao Tseng, certify that the work presented in this thesis is, to the best of my knowledge and belief, in whole or in part, has not been presented for a Degree at this or any other University. I also certify that, to the best of my knowledge, any help received in preparing this thesis and all sources used have been acknowledged in this thesis.

Signed

Chin-Yao Tseng

Date: August 2010

ii

ACKNOWLEDGEMENT
Upon the completion of my study journey, there are many people I want to acknowledge in appreciation for your support all the way. Without your ongoing assistance and patience, I would have not accomplished my study within my estimated time period.

First of all, I would like to acknowledge my supervisor, Associate Professor Michelle Wallace at Southern Cross University in Australia. Because of your richness in knowledge, experience, mentoring, support and understanding of the learning curve, I have gained expertise in the HR field, particularly in the retention area. Thank you so much for all the effort you put in to my knowledge development.

Secondly, I would like to express my appreciation to Associate Professor Don McMurray who supported me for the methodology part of the thesis. He always leads me to think outside of the box and makes things easy. In particular, he always encourages me to triangulate the argument from different aspects.

Thirdly, I would like to say thank you to the DBA staff of the Graduate College of Management, including Professor Philip Neck, Associate Professor Peter Miller, Sue White, Susan Riordan and Ashley Cooper, who always offered excellent service and administrative support whenever I needed it. I would also like to say thank you to Di Clarke who spared her precious time to refine my English grammar and language usage for this thesis.

iii

Fourthly, I would like to say thank you to my friends and previous colleagues, who supported me in completing the pilot survey and in introducing possible companies which might want to participate in the survey. I would also like to say thank you to the Human Resource Bank in Taiwan for conducting the survey when I encountered difficulties in the field arrangement.

Last but not least, I would also like to say thank you to my parents, my father (Rong-Zone Tseng) and my mother (Shuei-Lian Huang), who supported me in every aspect of my study, activities and careers. They always dedicate their best to their children for a prosperous future. My accomplishment definitely makes them proud of themselves.

iv

ABSTRACT
This thesis is about retention of software development employees in the IT industry in Taiwan. Retention involves the management of voluntary staff turnover so that effective staff are encouraged to remain in the employ of the organization. The core concept of this thesis is to manage voluntary turnover of these software development employees and provide a tentative set of retention guidelines appropriate to IT companies in Taiwan.

The rationale for this research is that the loss of software development employees not only involves the costs of a departing employee, but also involves loss of knowledge, social capital and company memory which may cause IT companies to fail. This is because IT companies tacit knowledge gained from experience and know-how may not be passed on from senior workers to junior workers. Consequently, the competitive advantages of IT companies may not be sustained and they could be forced to leave the business.

In addition, there are two research background problems identified for this study, which are a decreasing core workforce resulting from the aging population, and insufficient software development employees resulting from Taiwans education system. Both factors make the study significant, as software development employees are a key source of profit, skill and sustainable competitive advantage for IT companies. Therefore, the aim of this thesis is to identify what retention factors are important in retaining software development employees in the IT industry in Taiwan.

A quantitative approach was used for this study and an empirical retention model has been established involving six retention factors. These six retention factors were identified as

independent variables and their relationships with the dependent variable, which was the software development employees decision to stay, were examined. These six retention factors are: job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility.

The target population of this study was full-time software development employees who work in the R&D departments of HsinChu Science Park (HCSP) in Taiwan. There were three different types of software development employees identified for this study, namely software engineers, project leaders and assistant managers. A web-based online survey was used.

The findings of this study indicate that two unique and four common retention factors could be related to software engineers and assistant managers decisions to stay. One unique and five common retention factors could be related to project leaders decisions to stay. In addition, these three different types of software development employees are in different career stages. Therefore, tentative guidelines for retention policy and practice in IT companies could be developed based on a mixture of the career stages and unique retention factors of each different type of software development employee.

The recommendations of this study suggest that most software engineers and some junior project leaders could be in the career stage of Exploration, in which they could seek a comparable salary and could need technical skills development. Most project leaders and some senior software engineers could be in the career stage of Establishment, in which they could seek to pursue dual career development. This stage may activate job-hopping attitudes. Finally, assistant managers could also be in the career stage of Establishment, in which they could seek for more freedom in decision making and different pay level and
vi

pay content.

Keywords: IT, Retention, Software Development Employees, Taiwan.

vii

A double-blind, peer-referred paper relating to the research in this thesis has been accepted and was presented at the following conference.

Tseng, C. Y. & Wallace, M. (2009). The retention of software development employees in the IT industry in Taiwan. Paper presented to the 23rd Annual Australian and New Zealand Academy of Management, Melbourne, 1-4 December.

viii

TABLE OF CONTENTS

Declaration of Originality.ii Acknowledgements...iii Abstract.............................v Table of Contents..iv List of Figures..xv List of Tables......xvi Abbreviations...xviii

Chapter 1: Introduction
1.1 Introduction.1 1.2 Background to the Research..2 1.3 Research Problem, Hypotheses, Objectives, and Contributions4 1.4 Justification for the Research8 1.5 Methodology...11 1.6 Outline of this Thesis.14 1.7 Definitions...15 1.8 Delimitations of Scope and Key Assumptions.17 1.9 Conclusion..18

ix

Chapter 2: Literature Review

2.1 Introduction20 2.1.1 Overview of the Taiwanese IT industry..21 2.2 Labour Market...25 2.2.1 Workforce Crisis...28 2.2.2 Software Development Employees .32 2.3 Retention.35 2.3.1 Job Embeddedness Theory..36 2.3.2 Unfolding Model...40 2.3.3 Motivation Theory41 2.4 Career Satisfaction and Actualization..44 2.5 Retention and Career Actualisation in the IT Industry in Taiwan50 2.6 Research Problem...61 2.6.1 The Problem of a Decreasing Core Workforce...62 2.6.2 The Problem of Insufficient Software Development Employees...62 2.7 Research Question..64 2.8 Theoretical Framework..65 2.9 Conclusion...67

Chapter 3: Methodology
3.1 Introduction69 3.2 Justification for the Research Paradigms....70
x

3.2.1 Justification for the Positivist Research.75 3.2.2 The Process of Quantitative Methodology.77 3.3 Research Design.79 3.3.1 Methods of Data Collection.79 3.3.1.1 Sample of Target Population.85 3.3.1.2 Sampling Frame.85 3.3.1.3 Sampling Plan.87 3.3.1.4 Risk Management...89 3.3.1.4.1 Pre-risk management89 3.3.1.4.2 Post-risk management..91 3.3.1.5 Sample Size.94 3.3.1.6 Probability and Non-Probability sampling..95 3.3.2 Questionnaire Design.97 3.3.2.1 Development of the Measurement Scale...99 3.3.2.2 Development of the Questionnaire Items101 3.4 Validity and Reliability Measurement.104 3.5 Statistical Methods for Data Analysis..106 3.6 Steps in the Data Collection Process....108 3.6.1 Step 1: Pilot Study...109 3.6.2 Step 2: Main Study..110 3.7 Ethical Considerations..111 3.8 Limitations.112 3.9 Conclusion..113

xi

Chapter 4: Data Analysis and Results

4.1 Introduction...115 4.1.1 Overview of Research Focus and Survey Response.116 4.2 Data Analysis Procedure.......116 4.3 Data Evaluation and Testing118 4.3.1 Respondents and Their Demographics......118 4.3.2 Descriptive Data Analysis...121 4.3.3 Principal Components Analysis (PCA).124 4.3.4 Multiple Linear Regression (MLR)...131 4.3.5 Post-Hoc Analysis....143 4.3.5.1 T- Tests....144 4.3.6 ANOVA Tests.147 4.4 Summary of Data Analysis...149 4.4.1 Summary of the Empirical Research Hypotheses....149 4.4.2 Summary of the Research Question..152 4.5 Conclusion.154

Chapter 5: Conclusion and Implications

5.1 Introduction...156 5.2 Conclusions about the Empirical Research Hypotheses160 5.3 Conclusions about the Research Question..171 5.3.1 Conclusions for Correlation Tests.172 5.3.2 Conclusions for Standard MLR Tests..173
xii

5.3.3 Conclusions for T-Tests.179 5.3.4 Conclusions for ANOVA Test181 5.4 Conclusions about the Research Problem..182 5.5 Contributions for Theory.189 5.6 Contributions for Policy and Practice191 5.7 Limitations of this Research194 5.8 Implications for Further Research.195 5.9 Conclusion.196

REFERENCES...198

xiii

APPENDICES Appendix A: Appendix B: Appendix C: Appendix D: Appendix E: Information Sheet for HR Managers....218 Informed Consent Form for HR Managers..220 Research Questionnaire .....221 Survey Email ...227 Human Research Ethics Committee (HREC) Notification...228 Appendix F: Appendix G: Appendix H: Appendix I: Appendix J: Appendix K: Appendix L: Appendix M: Appendix N: Appendix O: Appendix P: Appendix Q: Appendix R: Measurement Scale Information....229 KMO and Bartletts Test.231 Communality Test........232 Scree Plot Test..234 Eigenvalues of Total Variance235 Factor Components.237 Validity and Reliability Test...239 Normality Test.249 Transformation Test251 Linearity and Homoscedasticity of Residuals...255 Correlation Dada Analysis for Multicolinearity...257 T-Tests..262 ANOVA Tests...267

xiv

LIST OF FIGURES
Figure 1.1: Figure 1.2: Structure of Chapter 1...1 Empirical retention model of software development employees.6 Figure 1.3: Figure 1.4: Figure 2.1: Figure 3.1: Figure 3.2: Figure 3.3: Figure 4.1: Figure 4.2: Figure 5.1: Figure 5.2: Block diagram for the research justification...10 Structure of this Thesis......14 Structure of Chapter 2...21 Structure of Chapter 3...70 Flow of Deductive process.79 Criteria of scale measurement for data analysis...111 Structure of Chapter 4.115 Data analysis procedure..118 Structure of Chapter 5....160 Correlation results for the retention of all aggregated employees..167 Figure 5.3: Correlation results for the retention of software engineers...168 Figure 5.4: Correlation results for the retention of project leaders...169 Figure 5.5: Correlation results for the retention of assistant managers...170

xv

LIST OF TABLES
Table 3.1: Table 3.2: Table 3.3: Summary of research paradigms.75 Advantages and Disadvantages of the four surveys..80 Summary of advantages and disadvantages of sampling designs96 Table 3.4: Table 3.5: Table 4.1: Table 4.2: Table 4.3: Table 4.4: Questionnaire structure with measurement scale....101 The purposes of pilot survey...109 Survey respondents by demographics (n=415)............119 Summary of Gender in comparison to Age...121 Summary of Gender in comparison to Salary..122 Summary of Gender in comparison to Academic Qualification.122 Table 4.5: Summary of Gender in comparison to Working Period in Current Company123 Table 4.6: Summary of Gender in comparison to Working Period in IT industry124 Table 4.7: Correlation results for software development employees.135 Table 4.8: Summary of R and ANOVA for all aggregated employees..136 Table 4.9: Table 4.10: Table 4.11: Table 4.12: Table 4.13: Regression coefficients for all aggregated employees..137 Summary of R and ANOVA for software engineers....138 Regression coefficients for software engineers.139 Summary of R and ANOVA for project leaders......140 Regression coefficients for project leaders...141
xvi

Table 4.14: Table 4.15: Table 4.16:

Summary of R and ANOVA for assistant managers...142 Regression coefficients for assistant managers....143 Summary of the t-tests for two age groups comparison..146

Table 4.17: Table 4.18:

Summary of the ANOVA test for three groups...149 Conclusions of the results relating to the hypotheses for four groups..150

Table 5.1:

Correlation results for four groups of software development employees.172

Table 5.2:

Standard MLR results for four groups of software development employees.174

Table 5.3: Table 5.4: Table 5.5:

Summary of t-tests results for age groups...180 Summary of ANOVA results.181 Summary of retention factors of software development employees.185

Table 5.6:

Summary of career stages and generations for software development employees..192

xvii

ABBREVIATIONS
ACS B2B CEDEFOP CEPD DBA DEV DF DGBAS ECN EU FA FDM FR GDP HCSP HR HREC HRM HTML ICT IT JA KMO Australian Computer Society Business to Business European Centre for the Development of Vocational Training Council for Economic Planning and Development Doctor of Business Administration Development Degree of Freedom Directorate-General Budget, Accounting and Statistics Engineering Change Notification European Union Factor Analysis Freedom in Decision Making Fair Remuneration Gross Domestic Product HsinChu Science Park Human Resource Human Resource Ethical Committee Human Resource Management Hyper Text Markup Language Information and Communications Technology Information Technology Job Appreciation Kaiser-Meyer-Olkin

xviii

MLR MOEA NCHA OECD PC PCA R&D STEM SCU SPSS WLB WF

Multiple Linear Regressions Ministry of Economic Affairs New Challenging Work Organisation for Economic Cooperation Development Personal Computer Principle Component Analysis Research and Development Scientific, Technical, Engineering and Mathematic Southern Cross University Statistical Package for Social Science Work-Life Balance Workplace Flexibility

xix

Chapter 1: Introduction
1.1 Introduction
This thesis is about retention of software development employees in the IT industry in Taiwan. This chapter presents an outline of this thesis and begins with an overview of the background problems of a decreasing core workforce and insufficient software development employees in the IT industry in Taiwan. Both factors are obstacles to sustaining a competitive advantage for IT companies in Taiwan and signal a research problem that needs to be addressed. This chapter also discusses research hypotheses, objectives, contributions, justification for the research focus, related methodology, outline of this thesis, definitions, delimitations of scope and key assumptions, and finally a conclusion is included. Figure 1.1 outlines the structure of this chapter.

Figure 1.1: Structure of Chapter 1 1.1 Introduction 1.2 Background to the Research 1.3 Research Problem, Hypotheses, Objectives, and Contributions 1.4 Justification for the Research 1.5 Methodology 1.6 Outline of this Thesis 1.7 Definitions 1.8 Delimitations of Scope and Key Assumptions 1.9 Conclusion
Source: Developed for this study.

1.2 Background to the Research

The global shortage of highly skilled workers, who can be regarded as software development employees in the post-industrial economy, has created a serious economic issue, that is somewhat separate from the recent global financial downturn (Korane, 2009; Vavra, 2009). It is expected that the current skills gap will enlarge, and global competition for talent will become more severe under a continuous revolution of information and communications technology (ICT) (Drucker, 2007; Gordon, 2009; Walker & LaRocco, 2002; Zheng, Soosay, & Hyland, 2008). This is because global demographic trends show that the pool of new highly skilled workers is dropping drastically, especially for the scientific, technical, engineering and mathematically based jobs (STEM) (Dychtwald, Erickson, & Morison, 2006; Gordon, 2009). This indicates that software development employees are in high demand in todays knowledge-based economy and tight labour market, as business growth for IT companies relies heavily on their productivity (Brandel, 2007; Drucker, 1999, 2007). Although training is a prominent approach to resolving the shortage of software development employees, retention of existing talent becomes more essential and critical (Doh, Stumpf, Tymon, & Haid, 2008; Horwitz, Heng, & Quazi, 2003; Mayfield & Mayfield, 2008; Punia & Sharma, 2008).

Based on the aspects of global demographic change, the workforce population is expected to be mainly filled by Generation X (those aged between 33 to 44 years) and Millennials (Generation Y and Net Generation, those aged below 32 years) (Dychtwald, et al., 2006; Gordon, 2009). However, the core workforce population is decreasing because the speed of core workers leaving exceeds the speed of young workers entering the workforce (Dychtwald, et al., 2006; Gordon, 2009b). This problem of a decreasing core workforce has also created serious concerns and challenges for the IT industry in Taiwan. Therefore, it is suggested that retention of existing software development employees can sustain a
2

competitive advantage for IT companies as companies knowledge and know-how can be passed on from senior workers to junior workers (Anonymous, 2007; Clarke, 2009).

From the point of view of software development employees, their productivity has offered a sustainable competitive advantage for IT companies by continual innovation of products and services that contain optimum output of quality (Davila, Epstein, & Shelton, 2006; Drucker, 2007; Huang & Lin, 2006). As innovation is the heart of the knowledge-based economy (Walker, 2007), it differentiates between companies retaining market leadership and barely surviving (Davila, et al., 2006). However, Taiwans education system has failed to supply enough software development employees for the IT industry, which has undergone an economic transformation with a shift from quantity to quality (Chen & Liu, 2003). This is because rote memorization in Mandarin is emphasized in Taiwans education system, and the English language is dominant on the Internet, so this confines the learning process (Masuyama & Vandenbrink, 2003). Therefore, retention of existing software development employees has become critical for the success of IT companies in Taiwan as their value determines the future of the organization (Abell & Oxbrow, 2001; Drucker, 2007).

In order to resolve the background problems of a decreasing core workforce and insufficient software development employees in the IT industry in Taiwan, this research explores the relationship between retention factors and software development employees decisions to stay. This research also provides a set of tentative retention factors as a guide; informed by research results, appropriate to IT companies, and effective in retaining software development employees in the IT industry in Taiwan.

1.3 Research Problem, Hypotheses, Objectives, and Contributions

Based on the discussion of scholarly evidence above, the core workforce in Taiwan is expected to decrease by the aging population (CEPD, 2008). In addition, Taiwans education system has not supplied enough software development employees for the IT industry in Taiwan (CEPD, 2006; Masuyama & Vandenbrink, 2003). Both factors have caused the shortage of software development employees in the IT industry in Taiwan. Therefore, the research problem addressed in this research is:

What retention factors are important in retaining software development employees in the IT industry in Taiwan?

Essentially, I argue that retention of existing software development employees could make a difference for IT companies in Taiwan as they represent a sustainable competitive advantage. From the findings, I also argue that a diverse set of tentative retention factors should be tailored to meet the needs of different types of software development employees according to their career stages and generational characteristics. It is deemed that retention of these software development employees can be effectively achieved when each of them is satisfied with their career achievement.

Informed by scholarly evidence on financial compensation (Dychtwald, et al., 2006), training and development (Hager, 2004), promotion (Chen, Chang, & Yeh, 2006), recognition (Messmer, 2006), challenging work (Lock, 2003), innovation and creativity (Jaskyte & Kisieliene, 2006), leadership style (McShane & Travaglione, 2007), autonomy (Davenport, 2005), work-life balance (Hytter, 2007), and job satisfaction (De Cieri & Kramar, 2005), a theoretical framework of retention model has been established from the literature and discussed in Chapter 2. In addition, six retention factors were summarised
4

from the Principal Component Analysis (PCA) in Chapter 4 and used to answer the research problem. They were renamed as job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility. These six retention factors were identified as the independent variables and their relationships with the dependent variable which was the software development employees decisions to stay, was examined.

The reason for choosing software development employees was because they are a key source of profit, skill, and sustainable competitive advantage (Drucker, 1993), executing daily operations of programming for the projects of customers. In addition, it is hard to recruit for this group of software development employees in todays labour market (Gordon, 2009; Hoffman, 2009). Hoch, Roeding, Purkert, and Linder (2000, p.68) stress that The scarcity of software professionals is the major barrier to growth in the software industry and one of the key challenges of software leaders. Thus, this group of employees is a highly appropriate focus. Figure 1.2 provides an empirical retention model of software development employees for this research.

Figure 1.2: Empirical retention model of software development employees

H1: Job Appreciation

H2: Fair Remuneration

H3: Freedom in Decision Making Decision to stay H4: New Challenging Work

H5: Development

H6: Workplace Flexibility

Source: Developed for this research.

In summary, in order to answer the research problem, and to examine the empirical retention model of software development employees, six empirical research hypotheses have been constructed as follows:

Empirical Hypothesis 1: Job appreciation is positively related to the software development employees decisions to stay in the IT industry in Taiwan. Empirical Hypothesis 2: Fair remuneration is positively related to the software development employees decisions to stay in the IT industry in Taiwan. Empirical Hypothesis 3: Freedom in decision making is positively related to the software development employees decisions to stay in the IT industry in Taiwan. Empirical Hypothesis 4: New challenging work is positively related to the software development employees decisions to stay in the IT industry in Taiwan.
6

Empirical Hypothesis 5: Development is positively related to the software development employees decisions to stay in the IT industry in Taiwan. Empirical Hypothesis 6: Workplace flexibility is positively related to the software development employees decisions to stay in the IT industry in Taiwan.

Based on the shortage of software development employees in the IT industry in Taiwan, retention of existing software development employees is one solution proposed in this study. This research examined the relationship between six retention factors and software development employees decisions to stay. In addition, this research also provided a set of tentative retention factors for each different type of software development employees for effective human resource management. Therefore, the purpose of this research has been to achieve the following two objectives.

1.

To identify what retention factors are important in retaining software development employees in the IT industry in Taiwan.

2.

To develop a set of tentative retention factors as a guideline; informed by the research results, and appropriate to the IT companies in Taiwan, in order for them to effectively retain their software development employees.

The contributions of this research will be presented at Chapter 5, Section 5.5 and 5.6. In summary, Section 5.5 makes four theoretical contributions to new knowledge: a software development employees retention model, the importance of retention factors to each different type of software development employees, and their comparisons between two age groups, and the closure of a knowledge gap about retention. Section 5.6 makes three sets of contributions to policy and practice as each of the three sets of tentative retention guidelines is tailored to meet the needs of different types of software development
7

employees according to their career stages and generational characteristics.

1.4 Justification for the Research

As stated above, a war for talent will continue to hit the growth of IT companies through the impact of the decreasing core workforce and insufficient software development employees. This indicates that IT companies in Taiwan need to find ways to sustain their growth as well as their competitive advantage. In addition, attracting, motivating and retaining software development employees have become more critical under todays global competition and fast pace of technological revolution (Horwitz, et al., 2003; Schulz, Camp, & Waltman, 2008). This is because the success of IT companies relies heavily on the human capital of software development employees, who can match the use of advanced technologies to the companies needs (Mata, Fuerst, & Barney, 1995; Ross, Beath, & Goodhue, 1996; Schulz, et al., 2008). Therefore, although a training approach is important to close the skills gap, retention of existing software development employees is also important and is the focus of this research.

Work content which involves long working hours and an inflexible working life, results in job-hopping proclivities in the IT industry and thus a high level of voluntary turnover (Khatri, Chong, & Budhwar, 2001; Schulz, et al., 2008). This is regarded as an unfriendly workplace in the IT industry, one which could influence software development employees to leave an organization despite an attractive salary and stock bonus (Chen & Huang, 2006). Although fair remuneration is a popular tool to attract software development employees, other retention factors, such as career appreciation and development, are strong determinants in their decisions to stay or leave (Agarwal & Ferratt, 2001; Eyring, 2008; Foote, 1998; Schulz, et al., 2008; Thapanachai, 2006).

In addition, as the values of each generation are different, their needs should be tailored to the diversities of their career stage (Craig & Hall, 2005; Nankervis, Compton, & Baird, 2005; Tapscott, 2009; Trower, 2008). It is suggested that employees whose career anchors are compatible with their jobs have higher career satisfaction levels and lower intentions to leave an organization than those who have incompatible career anchors with their jobs (DeLong, 1982; Igbaria, Greenhaus, & Parasuraman, 1991; Quesenberry, 2006; Schein, 1978).

Furthermore, retention of existing software development employees is the most cost-effective approach to offset the loss of knowledge, skills and competencies of IT companies (Anonymous, 2001; Schulz, et al., 2008; Somaya & Williamson, 2008). Therefore, it is suggested that IT companies can establish a friendly workplace with a cost-effective retention model to sustain a long-term relationship with their software development employees. This is because a different set of tentative retention factors has been developed based on the career satisfaction of each different type of software development employees. Figure 1.3 below provides a block diagram for the research justification.

Figure 1.3: Block diagram for the research justification Workforce Supply Qualified Workers Supply

Core workers decreasing

Lack of encouraging creativity within education system Lack of Internet access with
English capability

Mature workers increasing

The Shortage of Qualified Workers

Unfriendly Workplace in
the Taiwanese IT Industry

1. Long working hours 2. Inflexible working life

Job-hopping proclivities increased & voluntary turnover rate increased Gap of Retention Influence Importance of Retention Factors - Job Appreciation - Fair Remuneration - Freedom in Decision Making - New Challenging Work - Development - Workplace Flexibility

Friendly Workplace in the Taiwanese IT Industry Minimized Gap of Retention Influence Career Satisfaction & Stay Longer in the Organization
Source: developed for this research. 10

1.5 Methodology
As the research approach is dependent upon a researchers paradigm, a quantitative approach was used for this research. It is deemed that a positivist paradigm is the suitable approach to answer the research question because the theoretical framework was deduced from the existing theories (Guba & Lincoln, 1991). In addition, the research examined the relationship between the retention factors and software development employees decisions to stay, which existed externally and objectively (Easterby, Thorpe, & Lowe, 1991). Finally, the researcher alone collected the data and analysed it.

A web-based online survey was adopted for this research for the following reasons. Firstly, it offers the cheapest cost and fastest speed from the options of survey methods (Zikmund, 2003). Secondly, it involves higher anonymity of participants so they can feel free to answer all questions (Sue & Ritter, 2007). Thirdly, there is no interviewer bias during the survey (Sue & Ritter, 2007). Finally, it is an appropriate method for software development employees because they are computer literate (Neuman, 2006).

Selection of the target population was focused on full-time software development employees who work at research and development (R&D) departments located in HsinChu Science Park (HCSP) which is the oldest and largest science park in Taiwan. The sampling plan was designed to approach the selected IT companies and invite them to participate in the survey. However, the Human Resource Bank was authorized to conduct the survey because some difficulties occurred in the field work. These difficulties are discussed in the section of risk management.

Questionnaire design involved the requirements of accuracy and relevance of information (Zikmund, 2003). A five-point Likert-type interval scale was used to measure how strongly
11

instruments agreed or disagreed with the responses from participants (Cavana, Delahaye, & Sekaran, 2001). Finally, the research complied with voluntary participation, anonymity, and confidentiality throughout the entire process of data collection. This is a code of ethical conduct for social research supervised by the Human Research Ethics Committee (HREC) at Southern Cross University (SCU).

The data collection process involved two steps including a pilot study and a main study. A pilot study is a pre-test study using a small sample to test out all aspects of the survey in which comments made could help revise the instrument used in the main study (Gorard, 2003). It is suggested that a small scale trial run of a large survey be conducted for several purposes, such as testing questionnaire wording, layout, sequencing, and completion time (Veal, 2005).

As a disproportionate subsample size (software engineers: 313; project leaders: 72; and assistant managers: 30) was collected, the results for the comparisons between three groups will be interpreted in more tentative terms throughout the entire thesis. De Vaus (2004, p. 176) stresses that A strong correlation in a small sample may be statistically non-significant, while a much weaker correlation in a large sample may be statistically significant.

The choice of data analysis techniques is determined based on the research objectives that answer the research question. The research question was to assess the retention factors associated with software development employees decision to stay. Tabachnick and Fidell (2001, p. 17) stress that If the major purpose of analysis is to assess the associations among two or more variables, some form of correlation/ regression or chi square is appropriate. Hair, Black, Babin, and Anderson (2010, p. 23) also stress that With the
12

objective and conceptual model specified, the researcher has only to choose the appropriate multivariate technique based on the measurement characteristics of the dependent and independent variables. Thus, regression will be used to examine the retention factors in association with software development employees decision to stay as both variables were measured with interval scale and satisfied assumption of normality (Manning & Munro, 2007). In addition, the moderate value between those retention factors achieved p < 0.05 (Manning & Munro, 2007).

Other statistical techniques such as t-test and ANOVA were further used to answer demographic questions. T-tests were used to answer the difference between age groups (referring especially to the two age groups of 20 to 29 years and 30 to 44 years as they are the majority of respondents). The justification for this is the suggested retention factors linked to age group, which is measured with dichotomous nominal scale and interval scale and to satisfy assumption of normality (Manning & Munro, 2007). This is because the research was interested post hoc in identifying the retention factors of software development employees in relation to their career stage of exploration and establishment.

ANOVA was used to explore post hoc the difference between three groups of software development employees. The justification for this is because the research was interested in identifying the differences in retention factors between different types of software development employees.

The findings for this thesis indicate that a set of tentative retention factors could be developed based on two unique retention factors of software engineers. A set of tentative retention factors could be developed based on one unique retention factor of project leaders. Finally, a set of tentative retention factors could be developed based on two unique
13

retention factors of assistant managers.

1.6 Outline of this Thesis

This thesis is constructed of five chapters. Figure 1.4 outlines the structure of this thesis.

Figure 1.4: Structure of this Thesis Chapter 1: Chapter 2: Chapter 3: Chapter 4: Chapter 5:
Source: Developed for this study.

Introduction Literature Review Methodology Data Analysis Conclusion and Implications

Chapter One: Introduction provides an overview of the background to this research, research problem, research hypotheses, research objectives, research contributions, justification for the research, methodology, and scope of this research.

Chapter Two: Literature Review provides scholarly evidence of knowledge that reviews the concept of retention from the broad aspect of a tight labour market to the importance of retention theory, and follows on with personal career satisfaction. The conceptual model of software development employees retention is then established, and the theoretical framework with five main hypotheses is developed for this study.

Chapter Three: Methodology discusses the justification of the research paradigm, the choice of survey method, target population, sampling frame, sampling plan, risk
14

management, questionnaire design and its measurement scale, and steps in the data collection process. It also introduces the Statistical Package for Social Science (SPSS) test for data analysis.

Chapter Four: Data Analysis provides descriptive data analysis and followed by Principal Component Analysis (PCA), and normality test. The results of statistical analysis include standard Multiple Linear Regression (MLR), t-tests, and ANOVA test.

Chapter Five: Discussion and Conclusion provides the conclusions for the empirical research hypotheses, research question, and research problem. The implications / contributions for theory, policy, and practice are also included, as well as the limitations of this research and implications for further research.

1.7 Definitions
The key definitions of words or terms used for this research are provided below.

Assistant Managers: refers to software development managers who account for the success of customers projects and supervision of project leaders and software engineers.

Baby Boomer: people who were born between January 1946 and December 1964 (19 years); aged between 45 to 63 years old; TV generation (Tapscott, 2009).

Career Achievement: refers to an individuals career satisfaction level.

Career Anchor: refers to self-perceptions that influence an individuals career choice and career decisions (Schein, 1978).
15

Career Stage: refers to age related career development that involves an individuals response to occupational opportunities and life-role participation (Super, 1990).

Core Workforce in Taiwan: workforce aged between 14 to 44 years old (CEPD, 2008).

Generation X: people who were born between January 1965 and December 1976 (12 years); aged between 33 to 44 years old; computer generation (Tapscott, 2009).

Job-Hopping Proclivities: refers to the capability of software development employees to keep seeking opportunities for better financial compensation and higher levels of promotion.

Knowledge-Based Economy: refers to IT companies sustainable competitive advantage that relies on human intellectual capital (Abell & Oxbrow, 2001).

Mature Workers: workforce aged over 55 years old (Dychtwald, et al., 2006).

Mid-Career Workers: workforce aged between 36 to 54 years old (Dychtwald, et al., 2006).

Millennial (Generation Y or Net Generation): people who were born between January 1977 and December 1997 (21 years); aged between 12 to 32 years old; Internet generation (Tapscott, 2009).

Project Leaders: refers to software development employees who are responsible for the management of projects and supervision of software engineers.
16

Retention: refers to the act of retaining valuable software development employees that an organization wants to keep.

Software Engineers: refers to bottom-line software development employees who perform firmware or software programming for the projects of customers.

Voluntary Turnover: refers to the choice of a software development employee to either stay or leave an organization.

Young Workers: workforce aged under 35 years old (Dychtwald, et al., 2006).

1.8 Delimitations of Scope and Key Assumptions

The target geographic location for this study focuses on HsinChu Science Park which delimits the scope of knowledge about retention of the entire software development employees in Taiwan because Taiwan has many industrial science parks. As will be for the discussed in Chapter 5 this could be a platform for further research, and more importantly, the geographic location could expand considerably to not only focus on Taiwan, but also focus on China. This is because many IT companies in Taiwan have established their offices or factories in China for better global competition as multinational companies (MNCs). This group of software development employees is diversified and has a broader view about career achievement than those who prefer to work locally.

The research methodology used a quantitative approach, which also delimits the scope of knowledge as the findings were limited to the relationships between the identified variables. More importantly, the findings about significant factors were unable to understand more deeply the reasons behind these workers intentions to stay within the
17

organization as well as for not significant factors. Therefore, a qualitative approach could be undertaken for further research and findings could be explained more deeply by interpreting the investigations that are literal creation in the inquiry process.

The focus on software development employees delimits the scale of knowledge about retention from a broad aspect of sustainable competitive advantage for IT companies as a whole. This is because the success of some IT companies in Taiwan is achieved by other key employees such as hardware employees and sales and marketing people. The focus of key employees can be reconsidered according to the needs of research in the future.

In addition, the IT industrys culture is more open to western practices than Taiwanese national culture in this case. Bae and Lawler (2000) in Korea and Wu (2004) in Taiwan discuss that there are changes from traditional culture that impact on work culture and countries like Taiwan now accepts strategic HRM approaches including those relating to retention. Therefore, Taiwan as a culture was not included as a moderating effect.

1.9 Conclusion
This chapter is a roadmap for this thesis and begins with a background to the research, followed by research problem, research hypotheses, research objectives, and a summary of contributions for theory, policy, and practice. Justification for this research is to minimize the gap of retention influence by satisfying different types of software development employees career achievements. A quantitative approach was used for this research and the outline of this thesis is provided as well as the definitions of words and terms used in this thesis.

18

Based on the provision of this roadmap, this thesis can proceed, and begins by reviewing the literature in Chapter 2 which justifies the need for IT companies in Taiwan to retain software development employees.

19

Chapter 2: Literature Review

2.1 Introduction
Chapter 1 focused on a summary of the research background, problem, hypotheses, objectives, contributions, justification for the research, methodology, definitions and delimitations of scope, key assumptions for this study, and provided an outline of this thesis. This chapter reviews the scholarly evidence and develops a theoretical retention framework for this study.

This chapter consists of nine sections and begins by reviewing the background of the Taiwanese IT industry. Section Two analyses the labour market by discussing global demographic trends and the importance of software development employees. Section Three analyses the concept of retention by discussing the theories of job embeddedness, unfolding model and motivation. Section Four explains career satisfaction and actualization by addressing the compatibility of a persons career anchor and job setting options within the organisation over time at different career stages. Section Five explains retention and career actualization in the IT industry in Taiwan by addressing financial compensation, career development, self-achievement, job satisfaction, and the environmental factors that will determine the employees decision to stay in the organisation. Section Six identifies the research problem by discussing the two research background problems of a decreasing core workforce and insufficient software development employees. Section Seven provides research question developed from the scholarly evidence. Section Eight provides the theoretical framework, which includes five main research hypotheses with ten variables. Finally, a conclusion is provided in Section Nine. Figure 2.1 outlines the structure of this chapter.

20

Figure 2.1: Structure of Chapter 2 2.1 Introduction 2.2 Labour Market 2.3 Retention 2.4 Career Satisfaction and Actualization 2.5 Retention and Career Actualization in the IT industry in Taiwan 2.6 Research Problem 2.7 Research Question 2.8 Theoretical Framework 2.9 Conclusion
Source: Developed for this study.

2.1.1 Overview of the Taiwanese IT industry

According to the Ministry of Economic Affairs (MOEA) in Taiwan, Taiwans economy has succeeded in the growth of the foreign trade surplus (MOEA, 2008). In the projection of the Directorate General of Budget, Accounting and Statistics (DGBAS) under Executive Yuan, Taiwans economy will rebound more strongly than many advanced countries after the recent global financial downturn (DGBAS, 2009). It is projected that Taiwans GDP will reach 3.92% in 2010 (DGBAS, 2009), relying mainly on its global competiveness in outstanding technological innovation and R&D capability (MOEA, 2008). In addition, R&D professional-related jobs are anticipated to create the large number of job vacancies in the labour market (MOEA, 2008). This indicates that IT companies in Taiwan have to find ways to retain those valuable software development employees, in order to support the
21

growth of the organisation and thus sustain a competitive advantage in the global market.

Taiwans economic strength is rooted in its industrial base which has made a successful paradigm shift from labour-intensive industries to capital and technology industries, and is now in the growth of knowledge-based industries. According to notification from HsinChu Science Park (HCSP), it began with a manufacturing boom in basic-level exported goods and progressed to technology-intensive industries such as computers and their peripheral components (HCSP, 2009). Taiwan today is recognized as one of the worlds largest producers of computer-related products, and boasts an extensive and well-established network of industrial zones (HCSP, 2009). It is also growing a number of high-technology industry clusters which are supported by high-quality human resources and fully integrated supply-chain networks (HCSP, 2009).

Taiwan is also recognized as having strong industrial research and development capabilities. That is, national expenditure on R&D as a percentage of GDP rose from 2.44% in 2004 to 2.52% in 2005. Taiwan scholars have published 15,661 scientific dissertations, the 18th highest in the world, and held 5,118 U.S. patents, the fourth largest quantity in the world after the U.S., Japan, and Germany (HCSP, 2009). This shows that Taiwan has gained a high competitive advantage against competitors all over the world. Taiwan is also home to the largest venture capital industry in Asia. It has been recorded that by the end of 2005, there were 231 venture capital firms actually operating in Taiwan, raising a total of NT$189.3 billion (which equates to AUD$6.76 billion) for 10,295 investment projects, accounting for support of 368 listed companies (HCSP, 2009).

Furthermore, Taiwan has a highly integrated industrial base with a mature business to business (B2B) infrastructure in which the E-commerce trade networks are able to link
22

more than 20,000 Taiwanese exporters globally (HCSP, 2009). This facilitates all companies with a friendly working environment to effectively access all valuable information and connections quickly and efficiently. These are all the business advantages of Taiwan, which makes investment in the country so attractive to global companies, resulting in transformation of the industries rapidly and effectively.

Taiwan has obviously been under pressure to transform its economic structure and to adjust its industrial policies to respond to a fast-changing global environment (HCSP, 2009). This is because a new economic reality has quickly merged with knowledge and innovation, which makes the global competition more severe. Therefore, in order to ensure Taiwan has gained a sustainable competitive advantage in competing globally, more highly specialized technical and managerial personnel are required (HSCP, 2009).

From a human resource perspective, since Taiwan has pursued an export-led technology innovation strategy for the IT industry, software development employees have become increasingly demanding in relation to working conditions (Hobday, 1994; Hu & Mathews, 2005; Hu, Zheng, & Lamond, 2007). These working conditions could involve long working hours and high pressure, and consequently lead to turnover intention (Moore, 2000). Therefore, retention of software development employees is one solution proposed in this research that helps IT companies to gain a better competitive position in the global market.

In terms of cultural value, Confucianism is the cornerstone of cultural tradition in Taiwan (Huang, 2001). The basic tenets of Confucian philosophy are harmony and consensus that emphasizes a conflict-free, group-based system of social relations. The principle of Confucian value promotes respect for work, discipline, thrift, protecting face, ordering
23

relationships by status, duty to family and economic egalitarianism (Huang, 2001). It is believed that the value of Confucianism has a significant influence on peoples behaviour and ideas that shapes the policies of human resource management across many industries in Taiwan.

However, in the IT industry in Taiwan, many scholars have found that the influence of traditional culture on Confucian value has moved from group contributions to individual contributions (Chen, 1997; Farh, 1995; Huang, 2001; Wu, 2004). This results in the shift of managerial efficiency of the Taiwanese IT industry that has adapted to Western style of HRM practice. The new change of HRM practice encourages individual employees to be involved in the goal-setting and decision-making process rather than encourages on relationship and life-time employment (Wu, 2004). In addition, the Western style of HRM practice is more significant in Taiwan when corporate business is bigger in size and complexity (Leung, 1995).

As the IT industry in Taiwan is an export-oriented business, the challenges of HRM practice have been significantly influenced by both internal and external business environment within global competition. This results in a trend towards dominance of Western style of HRM practice with little Confucian value that formed and exercised in the IT industry in Taiwan (Wu, 2004; Yeh, 1991). There is a growing body of literature that discusses a convergence of management styles practices in Asian contexts that move them somewhat more towards Western styles with some regional influences (Chatterjee & Nankervis, 2007; Lui, Lau, & Ngo, 2004; Rowley, Benson, & Warner, 2004). In addition, this study aims to identify what retention factors are important in retaining software development employees in the IT industry in Taiwan. As a result, proposed variables may have considered based on Western style of HRM practice as cultural factor may not be as a
24

strong moderating factor as others in this study. Therefore, the use of a Western paradigm regarding retention is justified.

2.2 Labour Market

Although the recent financial downturn has shown that investment in the IT industry has slowed down somewhat, a war for talent is expected to return as the global economy recovers. This is because the IT industry is fundamental to economic growth (Anonymous, 2003; Kavanagh, 2008). According to the Organisation for Economic Cooperation and Development (OECD), labour shortages and skills gaps are big concerns and challenges that will continue to impact on global economies performance (OECD, 2008). This is because IT-related services and software development projects will generally grow, along with new broadband infrastructures, as they are an essential part of spending and are partly recession-proof (OECD, 2008). In addition, non-OECD countries such as BRICs (e.g. Brazil, Russia, India, and China) are particularly active in sustaining the global economic growth today and are expected to exceed the OECD countries (OECD, 2008). This indicates that competition for software development employees affects both advanced countries and emerging countries, as there is a large skills shortage in the global labour market.

According to Taylor (2008), a skill shortage exists when employers are unable to attract, recruit and retain sufficient staff with the specific skills and capabilities required at current levels of remuneration. These conditions are based on the following supply and demand factors. Supply factors are quality and quantity of the education and training system; occupational professional preferences of workers; access to on-going skill development; regional mobility; demographic trends; overall labour participation; migration; organisational culture and employment conditions; and remuneration (Taylor, 2008, p.13).
25

Demand factors are the level of demand for the industrial products and services; infrastructure and capital investment; technological change and innovation; industrial relations law; and the location where the goods and services are produced (Taylor, 2008, p.13). Apparently, the current skilled workforce is insufficient to support the demands of business growth in the IT industry around the world.

It has been identified that scientific, technical, engineering and mathematically based jobs (STEM) are hard to fill in todays global labour market (Gordon, 2009a). In particular, there are nine, most in demand, skilled jobs in the IT industry, namely programming / application development, help desk / technical support, project management, networking, business intelligence, security, web 2.0, and telecommunications (Hoffman, 2009). Although the recent financial downturn has forced IT companies to cut their expenditure, these types of skilled jobs remain in strong demand to sustain business growth in the IT industry (Antonopoulos, 2009; Arellano, 2007; Hoffman, 2009; Jowitt, 2009).

In the United States in 2009, there are still more than 3 million skilled job vacancies, and more importantly, it is expected that 12 million to 24 million skilled jobs will fall vacant from 2010 to 2020 (Gordon, 2009a). According to the European Centre for the Development of Vocational Training report (CEDEFOP), about 7.5 million engineering jobs will be created by 2015 in the pan-European countries (EU-25+), as the employment patterns shift towards service and knowledge-intensive jobs (CEDEFOP, 2008). For example, the recent demand for software development employees in networking security has created an increasing number of unfilled jobs in the labour market. Similarly, according to the Australian Computer Society (ACS), an estimated 25,000 IT jobs are predicted to be vacant by 2020 in Australia if there are no immediate policy changes (ACS, 2008).
26

In the emerging countries, China is projected to reach a shortfall of 300,000 software development employees, and a shortfall of 450,000 software development employees has been projected by India (Kavanagh, 2008). Both countries are expected to create a significant skills shortage in the global labour market because their original roles in providing talent to the rest of the world have shifted to support the growth of their domestic economy (Gopinath, 2007). More importantly, both countries have undertaken an approach to buy out IT companies from advanced countries to get access to the high-level software development employees (Gopinath, 2007). Consequently, global demand for software development employees is believed to result in a severe war for talent, as the future success of a business is dependent on the battle for talent (Anonymous, 2008).

According to a CIO Executive Council survey in North America, close to two-thirds of top-tier IT executives agree that they are experiencing a skills crisis for critical project challenges and are considering going abroad to look for software development employees (Arellano, 2007). This is because the side-effects of the skills shortage have limited an organisations productivity, efficiency and ability to innovate (Jowitt, 2009). In addition, it has already been shown that there is slowed growth in IT-dependent industries, increased outsourcing overseas, and a decreased supply of technology goods and services (Nelson & Todd, 2004). Consequently, this has significantly impacted on the key dimensions of business success in relation to business growth, speed to market, quality of products and services, and customer relationships (Jowitt, 2009).

As noted above, a severe competition for software development employees will not be only between IT companies but also between countries. In addition, China has offered a higher financial package in comparison to the rest of the world to attract this group of software development employees (The Economist, 2007). Consequently, this results in a
27

skills shortage because of higher staff turnover and rising wage costs (The Economist, 2007). This is a big issue because work ethics could be collapsed by a competitive financial package creating a norm in the work environment where the period of employment becomes as short as two years or even less (The Economist, 2007). Therefore, it has been suggested that identifying retention factors for existing software development employees is the only way to retain a long-term employment relationship and thus sustain a growth of business (Martin, 2008; Nelson & Todd, 2004; The Economist, 2007).

2.2.1 Workforce Crisis

Global demographic changes are the major reasons causing a large number of skills shortages, although other factors may be involved, including changing skill needs within occupations and labour participation rates (Gordon, 2009b; Martin, 2008; Taylor, 2008). The disproportionate size of the retiring group of the baby boomer generation, an increasing long life expectancy, and decreasing birthrates of the young cohort are expected to double the percentage of workers over fifty-five during the next five to ten years (Dychtwald, Erickson, & Morison, 2006). This indicates that the trend towards an aging population is a big challenge under a strong projection of global economic growth, as it has created difficulties in attracting and retaining staff in the IT industry (Drucker, 1999; Martin, 2008). A low birthrate in developed countries and massive retirement has caused a drastic reduction in the pool of new highly skilled workers (Dychtwald, et al., 2006; Gordon, 2009a, 2009b).

Statistics indicate that by 2050, about 9% of the ICT workforce will be retiring under current global demographic changes (Garmise, 2006). In addition, nearly 20% of the European working population will go into retirement within the next five years (McQuade, Sjoer, Fabian, Nascimento, & Schroeder, 2007), along with a 50% drop in the employment
28

rate of those aged 55 to 64 (Jouhette & Romans, 2006). Loomis (2006, p.47) also suggests that by 2012, almost one in five workers will be fifty-five years old and older. In the emerging countries, China is also facing a similar situation in that about 33% of the workforce will be over the age of 50 within the next five to ten years (Hawser, 2008). Thus, it is evident that an aging workforce is inevitable and the IT industry is expected to lose a large number of experienced workers over a short period of time.

As an aging workforce will increasingly result in a lesser supply of skilled workers, it was expected to significantly affect the IT industry starting in 2008 (Dychtwald, et al., 2006). That is, the number of young skilled workers is insufficient to fill the vacant jobs that are created by the massive retirement of the baby boomer generation employees who are now reaching retirement age (Dychtwarld, et al., 2006; Garmise, 2006). In addition, the workforce growth rate of the age segment 35-44, which are the primary executive development years, is actually declining (Dychtwarld, et al., 2006). This indicates that the skills shortage is going to impact on a wider range of employment levels, which appears to involve the bottom up to the management level in the IT industry.

From the point of view of workforce availability, it is a crisis that there is an insufficient number of skilled workers on a variety of employment levels and more importantly, companies tacit knowledge gained from experience could be lost (Dychtwarld, et al., 2006; Garmise, 2006). For example, retired people are unable to successfully pass on their knowledge and skills to the next generation of employees who replace them (Clarke, 2009; McQuade, et al., 2007). The loss of retired people could be mainly associated with the loss of an expert employee with knowledge of products and processes of the company; the loss of customer and supplier contacts, relationships and established trust; the loss of understanding of the informal personal networks; and many other losses including
29

knowledge of competitors and their products (McQuade et al., 2007, p.763). Consequently, the loss of companies knowledge could have serious impacts on business development and sustainable competitive advantage for IT companies in the global market.

Since there is a trend towards an aging workforce in the IT industry, IT companies must respond to the demographic changes with a proper employment practice for the replacement of retired baby boomer generation employees (Schramm & Burke, 2004; Young, 2006). In addition, a multi-generational workforce is a result of the aging population, and management practices must appeal to, and accommodate, the expanding variety of workers needs and preferences (Dychtwald, et al., 2006; McVey & McVey, 2005). Therefore, three career cohorts, namely mature workers, mid-career workers and young workers, have been created to enable effective management (Dychtwald, et al., 2006).

The mature workers cohort refers to people who are 55 years old and older in the workforce (Dychtwald, et al., 2006). This group of people mainly comprises the generation of the baby boomers, who are attracted to a culture that values their experiences, knowledge, skills and capabilities (Armstrong-Stassen, 2006; Dychtwald, Erickson, & Morison, 2004; Dychtwald, et al., 2006). In addition, 70% of mature workers are willing to engage in full-time or part-time work, as long as there is a better balance of work and life than they experienced in their earlier working years (Loomis, 2006). This indicates that mature workers can be effectively managed by increasing their autonomy and involvement under a flexible retirement work policy and a mature worker-friendly organisation (Buyens, Van Dijk, Dewilde, & De Vos, 2009; Claes & Heymans, 2008; Dychtwald, et al., 2006).

30

The mid-career workers cohort refers to people who are aged between 36 and 54 years old in the workforce (Dychtwald, et al., 2006). This group of people spans two generations and mainly comprises Generation Xers, who value family life the most, followed by workplace flexibility and speed of career progression (Trower, 2008). In addition, Generation Xers are marked by skepticism and self-reliance in the workplace (Trower, 2008). As this cohort has been working for a long period of time, they are highly exhausted and are generally regarded as a sandwiched worker that has been squeezed by a commitment to children and parents (Dychtwald et al., 2006). Therefore, it is suggested that mid-career workers can be effectively managed by offering a wide variety of work arrangements and benefits that enable work-life balance.

The young workers cohort refers to people who are under 35 years old in the workforce (Dychtwald, et al., 2006). This group of people mainly comprises the generation of Millennials (Generation Y or Net Generation), who value the freedom to make decisions, opportunities to learn and grow, teamwork and enjoyment in the workplace (Dychtwald, et al., 2006). This cohort is particularly well-educated, independent, functional and diverse, but less loyal to the organisation, and generally change jobs in less than three years (Dychtwald, et al., 2006). This indicates that job-hopping is a major issue for the management of young workers, so a focus on job involvement can be promoted to fulfil their desire for independence, learning and rapid growth (Dychtwald, et al., 2006; Garmise, 2006).

As an aging population has created a multi-generational workforce, IT companies must respond to todays diverse workforce with proper HR policies and practices in order to attract, retain and develop the new generation of diverse workers for the growth of business (Dychtwald, et al., 2006; Martin, 2008; McVey & McVey, 2005). Therefore, it is
31

important to identify retention factors for different age groups of employees, as these can help IT companies to effectively motivate and retain their different groups of software development employees.

2.2.2 Software Development Employees

Software development employees are one group of knowledge workers, and their productivities are key factors to the economic success of companies (Davenport, 2005; Drucker, 2007; Garmise, 2006). This is because todays work content largely requires new knowledge production and innovation of human capital, which is the quality output of their ideas and productivities (Drucker, 2000; Garmise, 2006). In addition, software development employees own the means of production which means that they carry incremental knowledge in their brains and thus can be highly mobile and independent (Drucker, 1998, 2006). This indicates that the ascent of knowledge work makes software development employees more valuable and attractive to the organisation than they need the organisation. In this respect, there is a challenge in that IT companies must find ways to attract, develop and retain the software development employees, in order to sustain their competitive advantage in the global competition.

The importance of software development employees and their knowledge is being recognized by the new economy. This is because their knowledge is the only asset of the company which cannot be easily replicated, and is a source of profit, skill and sustainable competitive advantage (Drucker, 1993; Jones & Miller, 2007). For example, new entrants can easily reproduce and clone the existing market products, but cannot easily transfer tacit knowledge which is synthesized through learning and experience in the organisation (Drucker, 1993; Garmise, 2006). Tacit knowledge is the expertise, experience and capability of software development employees, integrated with process and company
32

memory, so it is always contextual and skillful (Abell & Oxbrow, 2001; Jones & Miller, 2007). Through the generation, dissemination, application, and reuse of tacit knowledge, companies can create innovative ideas for continuous new product development and thus sustain a competitive advantage (Garmise, 2006; Kreiner, 2002; Ng & Li, 2003).

The application of incremental knowledge and information has accelerated the innovation process, which provides opportunities for companies to operate in new markets, to change ways of working, and to deliver different products and services (Abell & Oxbrow, 2001; Garmise, 2006). It is believed that only a continuous process of systematic innovation and improvement can protect companies from the pressures of global competition (Drucker, 2007; Jones & Miller, 2007). Nelson and Winter (1982) have also argued that the transformation of incremental knowledge into the daily basis of routine work can effectively prevent damage from the impacts of poaching staff, reverse engineering or industrial espionage. This indicates that the only secure form of competitive advantage for companies is to actively embody tacit knowledge into the process of continuous incremental innovation (Hammer, Leonard, & Davenport, 2004; Jones & Miller, 2007). Consequently, the innovative use of tacit knowledge results in profit (Jones & Miller, 2007).

Since the global economy has shifted towards knowledge-intensive jobs, software development employees are a growing category of the workforce (Drucker, 2007; Policastro, 2007). This group of employees is regarded as the most expensive type of workers that organisations employ because they possess the most economic value of incremental knowledge development (Drucker, 2007; Garmise, 2006). In addition, this group of employees is highly educated and is capable of using knowledge, theory and concepts rather than using physical force or manual skill (Policastro, 2007). It appears that
33

the productivity of software development employees is measured by the contributions of performance in comparison to the productivity of manual workers which is measured by the output of manufacturing products (Drucker, 2007). Consequently, software development employees are viewed as assets of companies which need to be developed and grown, whereas manual workers are viewed as a cost to companies which needs to be controlled and reduced (Drucker, 2007).

In general, software development employees have a high degree of expertise, education, or experience, and the primary purpose of their jobs involves the creation, distribution, application and reuse of knowledge (Davenport, 2005). In addition, software development employees are responsible for sparking innovation and growth of an organisation by investing in new products and services (Davenport, 2005; Drucker, 2007). However, software development employees are critical to the success of organisations, their unique characteristics of personality present challenges that HR managers have to deal with in the workplace. That is, a high degree of autonomy and independence are two common attributes of software development employees (Davenport, 2005; Drucker, 2007).

According to Drucker (2007), the work content of software development employees naturally involves a high degree of autonomy, which imposes responsibility for their productivity on their own management of personal development and growth. This indicates that software development employees do not like work content that is highly involved with specifying detailed steps and flow of knowledge-intensive processes (Davenport, 2005). In fact, software development employees require a fair process in decision-making, which can influence their commitments to the participation and involvement of job tasks (Davenport, 2005). This is because software development employees are thinkers and they live in ways that use their cognitive ability to define
34

difficult problems and identify intelligent solutions (Bostock, 1996; Davenport, 2005; Drucker, 1992). Indeed, software development employees are proud of their domain knowledge and are desirous for recognition from others (Davenport, 2005).

As software development employees highly value their domain knowledge, they are independent and reluctant to share it without rewards or guarantees of continued employment (Davenport, 2005). This is a problem, in that IT companies could fail if employees domain knowledge is not open to share with one another, as the success of software development relies on the three key ingredients of domain knowledge, deadlines and dialog (Mah, 2009). This indicates that collaboration and communication are also essential parts of the productivity of software development employees (Bostock, 1996; Mah, 2009). Rubens (2008) has stressed that the future economy is about knowledge and relationships, because social networks can enable people to explore ideas, develop new concepts and learn in ways that will be revolutionary. Therefore, the success of a business is deemed to rely on the effective management of software development employees, in which managers can motivate them to share their domain knowledge by giving them autonomy of job design (Davenport, 2005; Mah, 2009).

2.3 Retention
Retention of software development employees is important to sustain a competitive advantage for IT companies in todays global market (Drucker, 2007; Jones & Miller, 2007). Studies on the benefits of retaining software development employees have significantly indicated that IT companies can gain a higher operating performance, higher returns on assets and higher returns on capital employed (Cascio, 2002). In contrast, the loss of software development employees is expensive and may thus be detrimental to IT companies (Glebbeck & Bax, 2004; Khatri, Chong, & Budhwar, 2001; Reiche, 2009). The
35

effect of losing software development employees has significantly impacted on the performance of IT companies from the two dimensions of tangible and intangible perspectives (Frank, Finnegan, & Taylor, 2004; Tanova & Holtom, 2008).

The tangible perspective includes the administrative expenses of recruiting, selecting and training and development of new staff, whereas the intangible perspective includes the loss of employee morale, social capital and company memory (Morrell, Load-Clarke, & Wilkinson, 2004). It is noted that the total costs of replacing an employee amounts to 150% to 175% of the salary costs of the departing IT staff (Nelson & Todd, 2004). The intangible loss of knowledge, experience, and know-how of companies is more significant than the tangible costs of replacing staff because it could lower productivity and performance of companies and thus result in loss of profits (Frank, et al., 2004; Tanova & Holtom, 2008). Therefore, IT companies must understand the serious impacts of the loss of software development employees and find a way to retain them. This is the focus of this study.

2.3.1 Job Embeddedness Theory

Job embeddedness is a new construct of retention theory that explains why people stay in, or leave an organisation using non-work and non-attitudinal factors (Mitchell, Holtom, & Lee, 2001). The concept of non-work factors is related to personal and family activities off the job, as a conflict between work life and family role could lead employees to consider leaving an organisation (Mitchell, Holtom, & Lee, 2001). The concept of non-attitudinal factors is related to personal attachment to their co-workers, network groups and leisure activities both on and off the job (Mitchell, Holtom, & Lee, 2001). It is noted that leaving an organisation requires sacrifice or giving up these connections and other benefits that exist at the current organisation (Mitchell, Holtom, & Lee, 2001). Therefore, job
36

embeddedness is a broad constellation of influence on employees decisions to stay or leave an organisation (Mitchell, Holtom, & Lee, 2001; Mitchell, Holtom, Lee, & Erez, 2001).

The importance of job embeddedness helps organisations to understand the scenario that even satisfied employees could still leave an organisation for their broad personal interests (e.g., family role at different career stages). This theory is different from the traditional turnover theories of work-related attitudes, which only focus on job satisfaction and organisational commitment (Holtom & Inderrieden, 2006; Holtom & O'Neill, 2004; Lee, Mitchell, Sablynski, Burton, & Holtom, 2004). It is noted that attitudinal constructs such as job satisfaction only play a relatively small role in employee retention and leaving (Griffeth, Hom, & Gaertner, 2000; Hom & Griffeth, 1995). Mallol, Holtom, and Lee (2007) and Tanova and Holtom (2008) also show that job embeddedness has more influence on employees decisions to stay rather than their attitudes towards work or actual opportunities in the labour market.

The foundation of job embeddedness theory is developed from embedded figures theories and field theory (Lewin, 1951). Embedded figures are those in which people are immersed in their field and connected through many links within their backgrounds and environment (Mitchell, Holtom, Lee, et al., 2001). This views an individual as a part of a complex web of relationships and attachments (Mitchell, Holtom, Lee, et al., 2001). The more extensive the web is, the stronger the influence the web will have on an individual who is considering making changes. This is because that change will affect many other features of the individuals life (Mitchell, Holtom, Lee, et al., 2001; Tanova & Holtom, 2008). Consequently, this keeps employees from leaving an organisation as they are entrenched in their jobs and webs and become part of the surroundings (Mitchell, Holtom, Lee, et al.,
37

2001).

According to Mitchell, et al. (2001), job embeddedness consists of three dimensions: (1) links to other people or activities, (2) self-perception of fit with the job, organisation and community, and (3) perceived sacrifices associated with changing jobs. These aspects are important both on (organisation) and off (community) the job. Thus, employees decisions to stay can be explained using six dimensions, which are links, fit, and sacrifice associated with an individuals organisation and with his or her community. As the focus of this study is the retention of software development employees within an organisation, the emphasis of organisational dimension is discussed and provided.

Links to the organisation refer to relationships that an employee has with co-workers, superiors and mentors within an organisation. Mitchell, et al. (2001) stress that many organisations now use strategic teams to increase employees network bonds, and use a mentor system to increase the attachment of their female employees. They further stress that the relationships could extend to cover an employee and his or her family in social, psychological and financial areas that embed them (Mitchell, et al., 2001). Other studies also show that people are more likely to be loyal to fellow and network peers (Amaram, 2005; Guidice, Heames, & Wang, 2009). Thus, it is deemed that the more links between an employee and the web, the more he or she is bound to the job and the organisation.

Fit with the organisation refers to an employees perceived compatibility or comfort with an organisation (Holtom & Inderrieden, 2006). This involves a self-perception of fit that is especially important during the early stages of adjustment to a job (Mitchell, et al., 2001). It is suggested that an employees values, career goals and plans for the future must fit with the organisational culture and the demands of the immediate job (e.g., job knowledge,
38

skills and abilities) (Holtom & Inderrieden, 2006). This is because employees will continue to assess how their work contents have matched their career goals. Some scholars show that an organisation can retain employees by managing their career aspirations and helping them to find a way to meet their long-term career goals (Harman, Lee, Mitchell, Felps, & Owens, 2007; Holtom, Mitchell, & Lee, 2006). Thus, it is deemed that the better the organisational fit with the career goal of an employee, the more likely he or she is to stay.

Sacrifice associated with leaving the organisation refers to the perceived costs of material or psychological benefits that may be forfeited by leaving a job (Holtom, et al., 2006). These benefits could involve profit sharing, stock options, career advancement opportunities, flexible work arrangements, child care support, colleagues, interesting projects or perks. Some of these benefits are significant to employees who have families, especially females with small children at home (Holtom, et al., 2006; Mitchell, et al., 2001). Thus, the more an employee gives up when leaving, the more difficult it is to sever employment with the organisation (Holtom & Inderrieden, 2006; Shaw, Delery, Jenkins, & Gutpa, 1998).

Prior studies have shown that financial factors have the highest correlation with the sacrifice dimension of job embeddedness when employees are considering quitting a job (Allen, Shore, & Griffeth, 2003; Bergiel, Nguyen, Clenney, & Taylor, 2009). In addition, supervisor support is also significantly correlated with the employees decisions to stay as it can help better person-organisation fit and links to colleagues (Bergiel, et al., 2009; Holtom, et al., 2006; Stinglhamber & Vandenberghe, 2003). Some studies also show that the benefits of child care support or family-friendly hours are strongly linked to the embeddeding of female employees within an organisation (Holtom, et al., 2006; Tanova &
39

Holtom, 2008). These all show that a job embeddedness approach can effectively retain employees as they would sacrifice current benefits when considering leaving an organisation. Therefore, it helps the study in identifying retention factors that retain and motivate employees with an organisation.

2.3.2 Unfolding Model

The unfolding model is a theory of voluntary employee turnover that explains the decision and behaviour of people who voluntarily leave an organisation (Lee & Mitchell, 1994). This theory is explained by four decision paths through the process of an individuals shocks, scripts, image violation (e.g. values, goals and plans for goal attainment), dissatisfaction, job search, evaluation of alternatives and offers in hand (Lee & Maurer, 1997; Lee & Mitchell, 1994). In addition, decision path 1, 2, and 3 is triggered by a shock event, and decision path 4 (4a and 4b) is triggered by a continuous evaluation of images which are accumulated from job dissatisfaction (Lee & Mitchell, 1994). Therefore, this model provides an effective way to understand why people may stay or leave within an organisation (Holtom, Mitchell, Lee, & Inderrieden, 2005; Maertz & Campion, 1998).

A shock event could be positive, neutral or negative; expected or unexpected; and internal or external to the persons life at different career stages. The incompatibility of a persons beliefs and images has been found to be mainly associated with work-related shock (e.g. promotion and work stress) in comparison to non work-related shock (e.g. marriage and new born baby) over a persons career life (Holtom, et al., 2005; Lee, Gerhart, Weller, & Trevor, 2008). In addition, lack of person-organisation fit plays an important role in leading the decision to stay or leave made by an individual (Lee & Mitchell, 1994). This is because people have different career goals and personal interests at different times or stages of their career life (Lee & Maurer, 1997; Tanova & Holtom, 2008). Therefore, it is
40

suggested that people can be somehow satisfied through work arrangements based on a dimension of their work content and their career life.

Prior studies have shown that an unfolding model can help an organisation to prevent a high level of voluntary turnover from anticipated events (Lee & Maurer, 1997; Lee, Mitchell, Holtom, McDaniel, & Hill, 1999; Morrell, Loan-Clarke, Arnold, & Wilkinson, 2008). For example, a work-life balance policy can keep young married employees who otherwise may leave an organisation because of new born-baby. This refers to decision path 1 (Lee & Maurer, 1997; Lee, et al., 1999). In addition, when unemployment is low and jobs are plentiful, managers have to respond quickly to prevent employees who may leave via decision path 3, which refers to unsolicited job offers or inquiries from the labour market (Lee, et al., 1999). It is also suggested that intervention should quickly respond to dissatisfied employees who may perceive injustices in the allocation of training opportunities and promotion (Morrell, et al., 2008). Therefore, it is deemed that an unfolding model can effectively manage voluntary staff turnover through intervention, and thus motivate employees to stay longer with an organisation (Lee, et al., 1999; Morrell, et al., 2008).

2.3.3 Motivation Theory

Motivation sustains a complexity of psychological behaviour in which people can be induced to engage in a particular direction (e.g. goal-oriented aim), intensity (e.g. personal efforts), and persistence (e.g. personal behaviour) of voluntary behaviour (Stone, 2005). In addition, the needs of people are different, so a diversity of motivational techniques should be applied for the different needs of people. For example, the cohort of mature workers may prefer to have more leisure life time in comparison to the cohort of young workers who may like to have more opportunities to experience a new career (Dychtwald, et al,
41

2006; Stone, 2005). It is suggested that the more managers understand effective motivation across different cohorts of generations, the more job satisfaction can be achieved in the workplace.

According to Nelson and Todd (2004), compensation, flexible work arrangements, and non-monetary compensation are important factors in managing the retention of software development employees and in mitigating the effects of voluntary turnover. It is believed that money plays a key role in motivating employees because it basically enables them to purchase food, clothes, and also provides them with the means to pursue leisure activities (Chiu, Luk, & Tang, 2002). It can also serve as a symbol of achievement, recognition and status because it represents the value of their contribution to the organisation, and their success in the society (Chiu, et al., 2002; Stone, 2005). This is highly associated with the process theory, in that equity theory shows the value of money as a motivator.

However, in motivation theory, such as Maslows needs hierarchy theory and Herzbergs two-factor theory, money tends to be in the lower order needs and hygiene factors respectively, and cannot actually motivate employees but can prevent job dissatisfaction. Therefore, along with monetary compensation such as remuneration and bonuses, job satisfaction as a key motivator can be achieved through non-monetary compensation such as work environment, feedback and rewards, social relationships, personal growth opportunities, and self-achievement (Nelson & Todd, 2004; Stone, 2005).

In order to better understand how motivation factors influence the behaviour of employees and encourage them to stay with an organisation, content theories and process theory of motivation are discussed and provided. Content theories are Maslows needs hierarchy theory and Herzbergs two-factor theory. Process theory is Equity theory.
42

Maslows needs hierarchy theory Maslows needs hierarchy theory is one of the most popular motivation theories that help organisations understand the needs of employees and thus influence their behaviours and attitudes at work. Consequently, employees are satisfied to remain with an organisation (Kaliprasad, 2006; Ramlall, 2004). Maslow believes that employees are motivated to satisfy five basic types of needs: physiological (e.g., basic wage), safety (e.g., job security), social (e.g., friendly supervisor), esteem (e.g., recognition from supervisor and colleagues) and self-actualisation (e.g., challenging work and participation in decision making) (Stone, 2005). It is deemed that higher order needs (e.g., self-esteem and self-actualisation) have a stronger influence on long-term effects in comparison to lower order needs (e.g., physiological, safety and social) (Kaliprasad, 2006; Ramlall, 2004; Stone, 2005).

Herzbergs two-factor theory Herzbergs two-factor theory is the first motivation theory on the content of the job itself, which has been widely used in management circles and has had a big impact on job design (Ramlall, 2004; Stone, 2005). It is also called motivation-hygiene theory, whereby motivators are job-related, and hygiene factors are non-job-related. Job-related motivators are higher order needs for self-achievement, recognition, intrinsic interest in the work, advancement and growth (Ramlall, 2004; Stone, 2005). Non-job-related hygiene factors are lower order needs such as company policy, pay, co-worker relations and supervisory styles (Ramlall, 2004; Stone, 2005). Ramlall (2004) stresses that motivation can be increased through job redesign (e.g., job enrichment) to allow for increased challenge and responsibility, opportunities for advancement, and personal growth and recognition. It is noted that job content is truly a motivation factor and relates to producing job satisfaction and higher performance, whereas hygiene factors merely prevent employees from job dissatisfaction (Stone, 2005).
43

Equity theory Equity theory is based on fair treatment between the outcomes employees receive, and the efforts they input compared with similar levels of others. These are social exchange relationships, and they explain how employees evaluate their outcome-input ratio compared to others (Adams, 1965). According to Carrell and Dittrich (1978), employees firstly develop beliefs about what constitutes a fair and equitable return for their contributions to their jobs. Secondly, employees tend to compare what they perceive to be the exchange they have from their organisation (Carrell & Dittrich, 1978; Ramlall, 2004). Finally, when they believe that their treatment is not equitable, relative to the exchange they perceived compared to others, they will be motivated to take actions they deem appropriate (Carrell & Dittrich, 1978; Ramlall, 2004). It is noted that once employees weigh up the fairness of their rewards, they will be motivated to do more or less work (Fitzgerald, 2006; Stone, 2005).

2.4 Career Satisfaction and Actualisation

Career satisfaction is the self-achievement of employees whose internal career needs (e.g. self-concepts and career values) match job setting options within the workplace over time at different career stages (Igbaria, Greenhaus, & Parasuraman, 1991; Lee, 2002; Schein, 1987). This is the career choice and career decisions of employees throughout the movement of a persons career life and is regarded as a career anchor (Schein, 1996) or career orientation (DeLong, 1982). The effect of psychological contracts has also influenced the career satisfaction of employees who have become more experienced in respect to their length of employment relationship within the organisation (Agarwal & Ferratt, 2000; Igbaria, et al., 1991). In addition, the career values of employees have significantly shifted towards a balance of personal goals (e.g. technical competence) and family life at all ages, genders and stages (Lee, 2002; Quesenberry, 2006; Schein, 1987).
44

Therefore, it is suggested that a long-term employment relationship can be established based on the compatibility of an individual career anchor within the job setting options of the organisation over time at different career stages.

Career anchors There are ten categories of career anchors that can influence individual career choice and career decisions as they reflect most peoples self-concepts, basic values, motives and needs (DeLong, 1982; Mgaya, Uzoka, Kitindi, & Shemi, 2009; Quesenberry, 2006; Schein, 1996). They are: 1) managerial competence; 2) technical competence; 3) autonomy; 4) creativity; 5) service; 6) challenge; 7) lifestyle integration; 8) identity; 9) job stability; and 10) geographical security. For example, managerial competence is related to desire for promotion and upward mobility. Technical competence is related to who desire for training and development. Autonomy is related to desire to be more independent in the decision making.

As a persons career life moves on, a diversity of career anchors is expected to be dynamically formulated over time in accordance with that persons job attitudes and workplace experiences (Agarwal & Ferratt, 2000; Mgaya, et al., 2009). In the IT industry, there are five typical career anchors that play an important role in influencing the voluntary turnover of staff, including managerial competence, technical competence, autonomy, job stability and geographical security (Agarwal & Ferratt, 2000). In particular, the first two of the five career anchors are significant because some software development employees are only interested in pursuing their careers in the progression of technical competence, not to pursue management competence and roles (Lee, & Maurer, 1997; Nelson & Todd, 2004). Therefore, it is important to match the career anchors to individual career goals as the activities of training and development in technical positions are quite
45

different from the activities of training and development in managerial positions (Igbaria, et al., 1991; Lee & Maurer, 1997).

Some studies also show that job stability is consistently and significantly related to the career anchor of software development employees as it relates to financial rewards as an exchange for work (Crepeau, Crook, Goslar, & McMurtrey, 1992; Crook, Crepeau, & McMurtrey, 1991; Igbaria & McCloskey, 1996; Ramakrishna & Potosky, 2001/2002; Sumner & Yager, 2004). Autonomy also indicates that software development employees seek work systems that are free from organisational constraints and enable them to achieve their goals (Crepeau, et al., 1992; Crook, et al., 1991; Ramakrishna & Potosky, 2001/2002; Sumner, Yager, & Franke, 2005). Variety also indicates that software development employees are naturally attributed to challenging work, as it allows them to re-skill, adapt and upgrade their skill sets (Crepeau, et al., 1992; Sumner, et al., 2005). Identity is also significant, as software development employees are significantly motivated by the recognition of their status and prestige in their career development, which may include a set of critical skills that enhances their market value (Crepeau, et al., 1992; Sumner, et al., 2005).

Carlson, Derr, and Wadsworth (2003) also show that people increasingly want work that is interesting and challenging, while also supporting a meaningful personal and family life. Consequently, a balance of work and family life becomes a significant factor of career anchor that affects software development employees, particularly for female employees (Carlson, et al., 2003; Igbaria & McCloskey, 1996). However, there are some studies that are contrary to one another. For example, creativity is found to be significant from Sumner, et al. (2005), but not significant from Crepeau, et al. (1992). This could be that a combination of career anchors is naturally constructed for a given employee and is varied
46

in terms of temporal characteristics (Quesenberry & Trauth, 2007). Therefore, it is deemed that software development employees decisions to stay could be influenced by a combination of career anchors that affects them at different times of their career life.

Career stages Similarly, there are three career stages identified for this study, as it is believed that they provide an understanding of a persons behaviour towards job attitudes because each career stage comprises a variety of specific activities and psychological adjustments (Agarwal & Ferratt, 2000). For example, newcomers may seek to find ways to fit into the organisation, whereas senior members of the organisation may seek to achieve a higher level of promotion (Chen, Chang, & Yeh, 2006). The three career stages are of exploration, establishment and maintenance. Details are shown as follows.

1. Career stage of exploration: workers are those aged around 20 to 29 years old; first join an organisation; seek to develop the necessary skills and abilities required for the job; and establish a good relationship with their superior and peers (Agarwal & Ferratt, 2000; Chen, et al., 2006). 2. Career stage of establishment: workers are those aged around 30 to 44 years old; are able to work independently; and seek higher pay and positive reputations (Chen, et al., 2006; Niles & Harris-Bowlsbey, 2005). In particular, career achievement is of the utmost importance and may require frequent opportunities for promotion (Chen, et al., 2006). 3. Career stage of maintenance: workers are those aged around 45 to 64 years old; attempt to hold what they have accomplished; and may be in the transition of retirement from work (Chen, et al., 2006; Niles & Harris-Bowlsbey, 2005).

47

Prior studies have shown that a career stage model is a useful tool in understanding the important work attitudes that individuals have in accordance with their age and various working experiences (Allen & Meyer, 1993; Carlson & Rotondo, 2001; Conway, 2004). For example, the development of technical competence is significantly related to employees decisions to stay during their early stages of career life (Agarwal, De, & Ferratt, 2002; Conway, 2004; Koh, Lee, Yen, & Havelka, 2004). Job stability, growth and family responsibility may shift to become the focus of employees decisions to stay at their middle stages of career life (Conway, 2004; Hess & Jepsen, 2009). Finally, job involvement and employability may become the focus of employees decisions to stay at their late stages of career life (Conway, 2004; Hess & Jepsen, 2009).

It is noted that when individuals become capable and experienced in the labour market, organisational rewards such as promotion, pay and other benefits may be varied for them as incentive tools. This is because once individuals gain highly transferable competence, they would not be organisationally bound but instead are highly marketable (Sullivan, Carden, & Martin, 1998). Consequently, employees decisions to stay with an organisation tends to be primarily influenced by both transferability of competence and level of internal work values (Hsu, Chen, Jiang, & Klein, 2003; Sullivan, et al., 1998).

Transferability of competence is related to an individuals knowledge, skills and abilities that are marketable or organisation-specific (Baker & Aldrich, 1996; Bird, 1996; Waterman, Waterman, & Collard, 1994). Level of internal work values is related to an individuals career goals that they attempt to achieve through their career life (Sverko & Vizek-Vidovic, 1995). This is also regarded as an individuals career anchor (Hsu, et al., 2003). In this regard, career satisfaction of individuals can be successfully achieved through the motivation of individuals career development which links to their career goals and career
48

stages (Agarwal & Ferratt, 2000; Chen, Chang, & Yeh, 2004; Conway, 2004; Lee, 2002; Mahatanankoon, 2007).

As career satisfaction varies across a diversity of career stages and personal goals, it is suggested that the needs of an employees career development should be periodically reviewed on a basis of three years tenure of employment in IT the industry (Agarwal & Ferratt, 2000). This is because three years working experience can make newcomers capable and independent as well as change the needs of senior members, and thus facilitate them to review their psychological contracts with their organisations (Agarwal & Ferratt, 2000). Consequently, job-hopping attitudes could be activated if employees are not satisfied with their professional identity and status or career progressions.

It is also suggested that employees career success should be related to their self-management towards desired career progress (Kuijpers, Schyns, & Scheerens, 2006). It is regarded as active career-actualization which is related to the realization of personal goals and values in ones career vis--vis the facilitation and constraining conditions of the work situation (Kuijpers, et al, 2006). They further distinguish between intrinsic and extrinsic career success. They stress that intrinsic career success is subjectively compared with the persons own appreciation of his or her career actualization (achievement, recognition and career satisfaction) whereas extrinsic career success relates to external appreciation (salary and occupational status) (Kuijpers, et al., 2006).

However, the individuals attitudes towards the career development program should be measured beforehand because every individual is diverse in their need to pursue different career paths goals. Therefore, if both employees and organisations are consensual upon the employees career development program, it is believed that a high level of performance
49

and commitment can be consequently expected from the employees and a longer tenure will also be expected. Thus, a win-win situation can be created by both parties.

2.5 Retention and Career Actualisation in the IT Industry in Taiwan

As discussed above, an effective retention strategy is firmly related to the career satisfaction of an employee whose career anchors match the job setting options of the organisation over time at different career stages. In addition, a periodical review of an employees career development is an effective way to reduce the psychological effects because a diversity of reward systems can be applied to motivate them towards personal career actualization. It is believed that financial compensation, career development, self-achievement, environmental factors, and job satisfaction will determine the employees decision to stay in the organisation. Therefore, these components will be considered as elements of the immediate discipline for the study. The following five major factors have been identified in the literature as being likely to be relevant to retention.

Financial Compensation Financial compensation, including base pay and incentive pay, plays a complex role in employees attraction, retention, and motivation. Milkovich and Newman (2005) stress that pay elements are one of the most effective rewards that can motivate employees to perform well and to adapt and master new technology and work process. In particular, incentive pay (e.g. in the form of stock options) is an important factor to attract and retain software development employees (Dobson, 2007; Wang, 2007). Some studies also note that more software development employees perceive a high level of internal equity (in comparison to other employees of the same work in the same company) and external equity (in comparison to other employees of the same work in other company) for the decision to stay within an organisation (Agarwal & Ferratt, 1999; Pare, Tremblay, & Latrick, 2001).
50

This indicates that although software development employees receive an above average pay than other industries, a fair and comparable pay structure is still needed to attract, retain, and motivate them (Dobson, 2007; Fong & Shaffer, 2003; Gee & Burke, 2001).

In motivation theory such as Maslows needs hierarchy theory, a better pay system can motivate employees to perform well as their lower-level needs are already being satisfied (Kaliprasad 2006; Stone, 2005). In contrast to pay satisfaction, Herzbergs two-factor theory has noted that money cannot actually satisfy employees but can prevent job dissatisfaction as hygiene factor. Furthermore, equity theory has explained that employees want to receive fair treatment, and they are motivated once they weigh up the fairness of their rewards compared to others input and outcome (Kaliprasad 2006; Stone, 2005). It is deemed that money is not everything, but it is one of the primary reasons people work across their life span. More importantly, a fair and comparable payment is a way to retain qualified software development employees in the IT industry.

The retention theory of job embeddedness argues that the more employees are involved in the organisation, the more chance they will stay because it involves sacrifice of valuable connections and other benefits that exist in present organisation (Mitchell, Holtom, & Lee, 2001). Although financial compensation can only achieve a short-term objective goal, it serves as a symbol of achievement, recognition and status. In addition, it represents the value of employees contributions to the organisation, and their success in society (Stone, 2005). Therefore, financial compensation will be proposed as one important variable for the study.

51

Career Development Career development is mainly related to training, development, and advancement opportunities of employees across their life span. Through different career stages, motivations may be reinforced differently into different career orientations of employees. For instance, technical skills training may suit technical-oriented employees but may not suit managerial-oriented employees (Lee & Maurer, 1997; Nelson & Todd, 2004). In particular, software development employees at the career stage of Exploration may pursue technical competence rather than advanced promotion which may be pursued by the different career stage of Establishment (Chen, Chang, & Yeh, 2006). Software development employees usually earn high salaries, self-improvement becomes a focus of retention efforts, leading to increases in their job satisfaction (Acton & Golden, 2003; Shah, Sterrett, Chesser, & Wilmore, 2001). In addition, training and development has recently been seen as an opportunity not only to enhance employability but also to prepare for career progression, as employees are beginning to view rewards from a more holistic perspective (Salopek, 2008; Thomson, 2007). It is deemed that software development employees could be satisfied with their jobs when they are developing their career prospects with ongoing skills development.

Advanced progression, such as promotion, plays an important role in retaining software development employees, especially for those who are in the career stage of Establishment and career orientation of managerial competence. It is noted that the speed of promotion could lead employees to think that they have value in the organisation and are more suited for work in their present organisation than in others (Bernhardt & Scoones, 1993; Huang, Lin, & Chuang, 2006). In addition, promotion brings employees some benefits such as increased salaries, empowerment, autonomy, recognition and respect from others (Huang, et al., 2006). Although Lazear (1986) argues that promotion could make employees more
52

attractive to other organisations, many scholars argue that if there is not enough opportunity for qualified employees to move upward, they would be dissatisfied and consider leaving (Booth & Hammer, 2007; Branham, 2000; Huang, et al., 2006; Trevor, Gerhart, & Boudreau, 1997). Thus, it is deemed that promotion could signify a software development employees abilities, which implies not only a motivation factor but also a subtle means of evaluating the employment relationship (Huang, Lin, & Chuang, 2006).

In Maslows theory, the concept of career development (training and development and promotion) can be a motivator to satisfy software development employees higher-order needs of self-esteem. Employees with opportunities for training and development may decide to stay longer in the organisation (Smith & Rupp, 2002). In Herzbergs two-factor theory, career development can satisfy software development employees growth needs and increase engagement.

In the unfolding model of voluntary turnover, lack of training and development opportunities and promotion may have prompted employees to think that they are not valuable assets in their current organisation (Lee & Mitchell, 1994). Thus, they may consider leaving their organisation. In contrast, the retention theory of job embeddedness suggests that employees would think more about sacrifice of social connections and other benefits when it comes to missed opportunities for training and development and promotion (Mitchell, Holtom, & Lee, 2001). Therefore, there are two important variables in career development that will be proposed for the study, which includes training and development, and promotion.

Gagg (2005) has stressed that there are increasing signs that many companies seem to be reluctant to spend money on training and development. That is because there is a question
53

of what sort of training program is the most effective to benefit both individuals and organisations. However, many companies have created social networks for alternative unplanned training and development courses in the organisation. This has provided an important role in fostering a learning climate, which demonstrates not only a simple transfer of information, but also yields new knowledge (Hatala, 2006; Rhee, 2004; Wenger & Snyder, 2000). Nevertheless, social networking will not be proposed as a variable for the study, because of the need to limit the length of the survey instrument (and thus increase the probability of an adequate response rate). A higher number of variables could make the survey overlong.

Self-Achievement Self-achievement is related to intrinsic career success and a persons own appreciation of his/ her career actualisation. Woodruffe (1999) stresses that the sense of recognition and feeling appreciated motivates people to stay within their organisations. Many scholars support that an appropriate recognition in the workplace can motivate employees to perform well and satisfy the aspiration of employees (Castilla & Ruiz, 2008; Gee & Burke, 2001; Holt, Love, & Nesan, 2000; Wirtz, Heracleous, & Pangarkar, 2008). This is because people naturally desire to be appreciated and feel that their efforts are rewarded in the eyes of peers and management (Baldoni, 2005; Gee & Burke, 2001). In contrast, lack of recognition may lead employees to think that their efforts are useless and they may consider leaving their organisation (Baldoni 2005). Although Bhatnagar (2007) argues that recognition may be effective for first three years of employees tenure, it is still widely used to encourage hard work and reduction of turnover intention (Kelley, Blackman, & Hurst, 2007; Slagter, 2007).

54

Recognition can be formed in many ways and tied into different motivations of employees in relation to their career stages and orientations. For example, there may be a difference in desire for recognition in creativity oriented employees depending on the career stage. Thus, a trophy with a small amount of incentive money may encourage their continuous hard work in the Exploration stage. In contrast, managerial oriented employees may think more on promotion progression in their career stage of establishment. It is believed that recognition is a powerful tool to encourage hard work and public reward system with a small amount of money will not only motivate the employees, but also develop a positive organisational culture (Messmer, 2006). Therefore, recognition is regarded as an important variable for this study.

Challenging work is one significant factor that naturally attracts and retains software development employees in the IT industry as this is a strong motivator to influence their attitudes to work and participate (Lock, 2003; Milliman, Czaplewski, & Ferguson, 2003; Smith & Rupp, 2002). Holland, Hecker, and Steen (2002) also point out that software development employees highly value technical challenges for self-development because professional identity makes them proud of themselves and appreciated by others. It could be that the nature of software development employees is extremely related to both intrinsic (e.g. challenging work) and extrinsic (e.g. high pay) career success. It is believed that software development employees are highly associated with self-achievement, so they always work longer hours to advance their knowledge, skills and ability, and to feel different (Wynekoop & Walz, 2000).

As work content in the IT industry involves challenges, the desire to achieve becomes a powerful incentive to motivate software development employees at work (Colin, Hanges, & Locke, 2004; Kochanski & Ledford, 2001). In addition, challenging work could be a
55

meaningful and interesting work for mature workers as they may feel they are useful in their organisation (Slagter, 2007). However, challenging work, on the other hand, could put pressure on some people because different people have different career goals at their different career stages. This is because software development employees are vulnerable to work exhaustion as the nature of their work environment involves long working hours and high pressure (Moore, 2000). Despite this latter factor, challenging work is regarded as an important variable for this study as it is an integral part of a software development employees job.

An innovative and creative culture is believed to be highly related to the intrinsic motivation that attracts employees to stay in the organisation (Jaskyte & Kisieliene, 2006). This is because innovative and creative work is one of the significant factors, which can satisfy software development employees. Prior studies have shown that software development employees are captivated by doing something spectacular that has never been done before in their work (Amabile, 1997; DeMers, 2002). However, the innovative environment must be supported from the highest-level of management through to middle-level supervisors who directly foster creativity when communicating with the bottom line of the employees. This is because creativity involves risk-taking and can be effectively diminished during a downsizing of an organisation (Amabile & Conti, 1999).

In addition, Barry (2007) stresses that only about 36% of the leaders surveyed actually agree that encouraging risk-taking and innovation within their team was important and only 42% rate themselves as effective. This indicates that software development employees perceptions of a creative environment could lead to turnover intentions (Mayfield & Mayfield, 2008), as it depends on management support from the top down to the supervisors.
56

As innovation and creativity could involve risk in the project development, the level of risk-taking would vary according to different career stages of software development employees. It is believed that people in the career stage of Exploration would be willing to try something new in their life in comparison to those who are in career stage of establishment. This is because they are naturally in the different segments of Generations (e.g., Generation X and Y). Therefore, innovation and creativity is regarded as an important variable for this study.

In Maslows theory, self-achievement (recognition, challenging work, and innovation and creativity) is regarded as employees highest-order needs, which is in contrast to their basic physical needs (Stone, 2005). Appreciation is believed to be a significant motivator to retain software development employees (Luftman, 2008). In Herzbergs two-factor theory, self-achievement can also satisfy software development employees recognition needs and motivate them to pursue new ways of doing things for their tasks.

Environmental Factors Environmental factors entail the elements that employees appreciate about their outside world. Amar (2004) stresses that a better work environment can be developed to effectively motivate employees to behave consistently and remain focused on their organisational goals. These factors are mainly related to their managers leadership styles, autonomy of their job, and their work and family balance (work-life balance).

Leadership style can increase employees job satisfaction, and thus lower employees turnover intentions (Chen & Silverthorne, 2005). This is because good managers can make their employees feel valued and provide support when their members face work and personal challenges (Messmer, 2006). In addition, good managers usually highly involved
57

with employees as this is a way to foster employees career development and commitment to the organisation (Curtis & Wright, 2001; Hytter, 2007). This is important to employees who are in different career stages and have different career orientation in their life span. Through open and honest communication with the supervisor, employees can better develop their career in which they may pursue technical progression or other alternative options.

Some studies have shown that good employee-supervisor relationships with honest and open communication can strengthen the support from the management that influences software development employees decisions to stay within an organisation (Burnes, 2006; Luftman, 2008). It is noted that 64% of software development employees consider leaving an organisation mainly because of the leadership style, which is top on the list of the ten factors (McGee, 2005). It is deemed that effective communication between employees and management could create an environment of trust and thus promote a positive work environment (Cleveland, 2005). Therefore, leadership style is regarded as an important variable for this study.

Autonomy is believed to be the most important structural factor in motivating behaviour of software development employees because it provides authority for decision-making (Davenport, 2005; Taplin & Winterton, 2007). Hytter (2007) also stresses that software development employees are highly motivated when they are given meaningful assignments that allow them to be involved in the decision-making for their areas of expertise. The demand for autonomy may differ according to different career stages. For instance, people in the Exploration stage may desire more decision making freedom in the technical area rather than the management area. The more opportunities they have to be involved in the process of decision-making, the more the investment in their projects will be linked highly
58

to self-achievement (Messmer, 2006).

Some studies have shown that software development employees are independently motivated by the nature of their work and prefer a high level of discretion and control (Barry, 2007). This is because they are natural problem solvers, and want their work to make a difference by running with their ideas and rules, which may not follow traditional patterns. Feyerherm and Vick (2005) also stress that software development employees want to be valued for their contributions with increased autonomy of operating style, more trust, and they want to be included in the decision making process. This is particularly true of women (Feyerherm & Vick, 2005). It is thus inferred that software development employees could experience greater satisfaction when they are trusted to work on project tasks in which they have more authority and autonomy to make their own decisions (Amundson, 2007). Therefore, autonomy is regarded as an important variable for this study.

The area of work-life balance in human resource management has been an important issue not only for its link to the attraction of employees, but also its link to the retention of employees (Holland, Sheehan, Donohue, & Pyman, 2007). This is because the increase of female workforce and aging population trends require the needs for companies to support valued employees who have family responsibilities (Cappelli, 2005; P. Holland, et al., 2007). As career is dynamic to each person, it is important to retain those who have different motivations at different career stages (Carlson, et al., 2003). For example, employees in the career stage of Exploration may spend much time at work for enhanced competence while employees in career stage of Establishment may want to spend equal time between work and family. In addition, work and family conflict is the most significant factor that causes an employees level of stress and turnover intention, particularly females
59

in the IT industry (Sumner, 2008). This indicates that employees could consider leaving an organisation if their roles are in conflict between work and family.

According to Greenhaus and Beutell (1985), there are three conceptually distinct forms of work and family conflict including time-based, strain-based, and behaviour-based conflicts. As work and family could affect with one another in either direction, this creates six dimensions of work and family conflicts. However, time-based conflict is associated with time pressures for one role of an employee and makes it difficult to participate in another role of family member (Greenhaus & Beutell, 1985). Strain-based conflict is associated with one role that intrudes into and interferes with participation in another role (Greenhaus & Beutell, 1985). Behaviour-based conflict is associated with behaviour of one role that is incompatible with the other behaviour of another role causing inconsistencies in the expectations of dissimilar roles (Greenhaus & Beutell, 1985). It is deemed that work and family conflict is a source of stress which causes emotional exhaustion and lack of accomplishment, and eventually decision to stay or leave within an organisation. Thus, work-life balance is regarded as an important variable for this study.

Consequently, three variables addressed in the environmental factors will be proposed for the study, namely leadership, autonomy, and work-life balance.

Job Satisfaction It is worth noting that job satisfaction is a significant factor that leads employees to either stay longer in the organisation or leave as soon as they are dissatisfied. A physical job withdrawal is a manifestation of career dissatisfaction being taken seriously by the employees (De Cieri & Kramar, 2005). It is believed that although increasing job mobility is unavoidable, a set of retention guidelines is the best and most effective way to retain
60

software development employees. This is all about the psychological success of career actualization, and refers to the feeling of pride and accomplishment that comes from achieving life goals that are not limited to achievements at work (e.g., raising a family and maintaining good physical health) (De Cieri & Kramer, 2005, p.421).

Some studies have shown that generally, software development employees are young, ambitious, and highly paid, so they are fascinated by new jobs and interesting assignments which offer technical challenges or opportunities for self-development (Holland, et al., 2002; Sigler, 1999). This shows that younger skilled employees are no longer loyal to an organisation if their personal needs are not satisfied. In addition, turnover and absenteeism is reduced when software development employees perceive that their jobs meet their important values (De Cieri et al., 2008; McMurtrey, Grover, Teng, & Lightner, 2002). It is deemed that software development employees could be strategically retained through the fulfilment of that persons important job values, as each of the individuals is different in respect to job satisfaction. Therefore, job satisfaction will be proposed as one important variable for the study.

2.6 Research Problem

This section provides two research background problems, including a decreasing core workforce and insufficient software development employees in the IT industry in Taiwan. The decreasing core workforce has been caused by the aging population (CEPD, 2008). The insufficient number of software development employees has been caused by the Taiwans education system which has not supplied enough software development employees for the IT industry in Taiwan (CEPD, 2006; Masuyama & Vandenbrink, 2003). Details of the two research problems are shown as follows.

61

2.6.1 The Problem of a Decreasing Core Workforce

The demographic trend in Taiwan towards an aging population has significantly decreased the labour participation of the core workforce (CEPD, 2008). The birthrate of Taiwan in 2009 is now 1.1% and it is predicted that it will remain flat over next few decades (CEPD, 2008). In addition, the participation of the core workforce (aged between 15 to 44 years) was 64.9% in 2008 and is expected to decrease to 60.9% (about 300,000 labourers) by 2012 (CEPD, 2008). From the quantitative labour perspective, the decreasing core workforce has significantly impacted on the activities of business and the economy of the nation as a whole.

The Taiwanese IT Industry is expected to experience an annual shortage of 45,000 specialized technical and managerial personnel by 2015 (CEPD, 2007). Although the two main academic educators, the National Tsing Hua University and the National Chiao Tung University, support the development of software development employees for the major IT industries, the shortage of software development employees is a significant problem. This is because an insufficient utilization of study for the labour market still remains a serious problem in the education system in Taiwan (CEPD, 2006).

2.6.2 The Problem of Insufficient Software Development Employees

In todays knowledge-based economy, a strong demand for economic transformation, with a shift from quantity to quality has pressured Taiwanese software developers to become more innovative (Chen & Liu, 2003). English language is one of challenges that confines learning progress and innovation in the IT industry in Taiwan as it creates a barrier to access advanced knowledge for a better improvement of product development. That is because, while Mandarin is the official language in Taiwan English is the dominant language in software with about eighty percent of information stored and communicated in
62

the Internet network being in English (Masuyama & Vandenbrink, 2003). The education system is another confining factor that has not encouraged creativity because the emphasis on rote memorization is dominant (Masuyama & Vandenbrink, 2003). These two factors are limiting changes in the Taiwanese people towards innovation-oriented thinking. These factors make the retention of skilled, software development employees more urgent and necessary.

Due to a high voluntary turnover rate, the IT Industry needs to keep the software development employees it already has. It appears that there is an 86% voluntary turnover rate in the IT industry, and those with less than a one year tenure have about a 42.1% turnover rate whereas this decreases with longer tenure of the workers (DGBAS, 2007). To some extent, the age group below 25 and the higher level of education group represent the highest turnover rate compared to older workers and/ or those with a lower education. This indicates that young workers and the more highly educated have different needs for advanced development, such as learning opportunities and promotion. It is this group of employees that the IT industry most needs to retain because of their contemporary, advanced skills and greater openness to innovation.

In addition, geography and family responsibility are significant to influence software development employees decision to stay within an organisation. Wu (2004) stresses that environmental factors are also one of todays HRM challenge as the rise of China and globalization may result in requests for employees to work at different locations in Asian and the world. If reallocation of assignment is in conflict with the values of employees, they may be dissatisfied and then consider leaving as they are capable of moving from one organisation to another where their values and family responsibilities are taken into consideration (Drucker, 2006; Wu, 2004).
63

Furthermore, as the nature of work content in the IT industry involves long working hours and inflexible time, the role of family responsibility could be disrupted and then result in job dissatisfaction. This can be regarded as a source of stress and could lead to the intention of staff towards voluntary turnover (Moore, 2000; Zeffane & McLoughlin, 2006). A balance of work and life is one of many variables that may have impacted on the software development employees decision to stay within an organisation.

In summary, both problems of the quantity of the labour force and the quality of software development employees have negatively impacted on the economic development of the IT industry in Taiwan. Although the shortage of software development employees is a common problem around the world, it is more severe in Taiwan because Taiwans economy relies heavily on the export business. This leads to this study which aims to provide a tentative set of guidelines for retention of the software development employees appropriate to the IT industry in Taiwan for sustainable competitive advantage in a global market. Therefore, the research problem addressed in this study is as follows:

What retention factors are important in retaining software development employees in the IT industry in Taiwan?

2.7 Research Question

Based on the literature review of the parent discipline, the retention of software development employees is an economic way to help IT companies to sustain a competitive advantage in the global market. In addition, based on the literature review of the immediate discipline, five main hypotheses with ten retention factors are important in retaining software development employees, namely financial compensation, career development (training and development and promotion), self-achievement (recognition,
64

challenging work and innovation and creativity), environmental factors (leadership style, autonomy and work-life balance) and job satisfaction.

The group of software development employees was the focus for this study. The reason for choosing software development employees is because they represent a source of profit, skill, and sustainable competitive advantage (Drucker, 1993), and execute daily operations of software programming for the projects of customers. In addition, it is hard to recruit for this group of software development employees in todays labour market (Gordon, 2009b; Hoffman, 2009). Hoch, Roeding, Purkert, and Linder (2000, p.68) also stress that The scarcity of software professionals is the major barrier to growth in the software industry and one of the key challenges of software leaders. Furthermore, HsinChu Science Park (HCSP) is the focus for the survey because it is the oldest and largest industrial science park in Taiwan. Therefore, the research question is as follows:

What retention factors are important in retaining software development employees in the IT industry in Taiwan?

2.8 Theoretical Framework

Based on the previous sections of the immediate disciplines, a research problem and question, a theoretical retention framework has been established for this study. These are five main hypotheses with ten retention factors. These ten retention factors were identified as independent variables and their relationship with the dependent variable, which was the software development employees decision to stay, was examined. Therefore, five main hypotheses with ten retention factors have been developed as follows:

65

Hypothesis 1: Financial compensation is positively related to software development employees decisions to stay in the IT industry. Hypothesis 2: Career development is positively related to software development employees decisions to stay in the IT industry. Hypothesis 2a: Training and development is positively related to software development employees decisions to stay in the IT industry. Hypothesis 2b: Promotion is positively related to software development employees decisions to stay in the IT industry. Hypothesis 3: Self-achievement is positively related to software development employees decisions to stay in the IT industry. Hypothesis 3a: Recognition is positively related to software development employees decisions to stay in the IT industry. Hypothesis 3b: Challenging work is positively related to software development employees decisions to stay in the IT industry. Hypothesis 3c: Innovation and creativity is positively related to software development employees decisions to stay in the IT industry. Hypothesis 4: Environmental factors are positively related to software development employees decisions to stay in the IT industry. Hypothesis 4a: Leadership style is positively related to software development employees decisions to stay in the IT industry. Hypothesis 4b: Autonomy is positively related to software development employees decisions to stay in the IT industry. Hypothesis 4c: Work-life balance is positively related to software development employees decisions to stay in the IT industry. Hypothesis 5: Job satisfaction is positively related to software development employees decisions to stay in the IT industry.
66

2.9 Conclusion
From an overview of the Taiwanese IT industry, the innovation of R&D capability is a core competence that helps Taiwanese IT companies to sustain their competitive advantages in the global competition. It is expected that Taiwans economy is going to rebound more quickly and stronger than many advanced countries after the recent global financial downturn ends. In addition, since Taiwan has adopted an export-led technology strategy for the IT industry, software development employees have become increasingly demanding. These factors indicate that software development employees are valuable to the IT companies as their innovation of human capital can produce an optimal output of quality products and services. Therefore, IT companies must find ways to attract and retain software development employees. The aim of this research is to explore factors that could retain them.

From the literature review of parent disciplines (human resource management), this study begins by analysing the labour market from the broader view of a global war for talent. The global war for talent is a result of higher voluntary staff turnover and rising costs. IT companies could seriously fail because of the loss of competitiveness. In addition, global demographic trends have created a shortage of managerial and engineering professionals, particularly as there is a shift from Western countries to the Asian region. The aging population is also a big issue that creates a multi-generational workforce and diverse needs. Finally, the importance of software development employees is addressed as they own the means of production so they are highly independent and mobile.

From the point of view of retention, a learning process from social network is a better approach to help IT companies to retain their software development employees as it enhances the capability of an employee to fit into the organisation. It is importance to
67

retain software development employees because losing them is expensive, particularly the losses of knowledge, social capital and company memory. It is also suggested that IT companies can effectively retain software development employees through the promotion of job embeddedness and motivation theories. Finally, job satisfaction is a key motivator to intrinsically influence employees appreciation for their work and thus remain in a longer employment relationship within the organisation.

From the point of view of a career satisfaction and actualization, compatibility of an employees career anchor and job setting options within the organisation over time at different career stages can retain a long-term employment relationship. It is also suggested that career development of an employee should be periodically reviewed on a basis of every three years, in order to reduce the effects of psychological contracts, which are highly associated with job-hopping attitudes.

There are two background problems to the research that have been identified as the problem of a decreasing core workforce and insufficient software development employees. The decreasing core workforce has been caused by the aging population, whereas the insufficient number of software development employees has been caused by Taiwans education system. Therefore, the research problem has been identified as what retention factors are important in retaining software development employees in the IT industry in Taiwan?.

Based on the immediate HRM discipline of retention factors that includes the constructs of financial compensation, career development, self-achievement, environmental factors and job satisfaction, a conceptual retention model has been established for this study. This conceptual retention model is the basis for the next chapter.
68

Chapter 3: Methodology
3.1 Introduction
Chapter 2 focused on the reviews of the literature and constructed the conceptual model of the research, which examined the relationship between retention factors and software development employees decisions to stay in the IT industry in Taiwan. Five main hypotheses with eleven variables were identified in Chapter 2 and this chapter discusses the implementation plan of a survey for the quantitative research.

This chapter consists of nine sections (see Figure 3.1). It begins with the Introduction (Section 3.1), and is followed by the justification for the research paradigms (Section 3.2), which includes the justification for the positivist research approach and the process of the quantitative methodology. Research design (Section 3.3) provides methods of data collection and questionnaire design. Methods of data collection include the sample of a target population, sampling method, sampling frame, sampling plan, risk management, sample size, and probability and non-probability sampling. Questionnaire design includes the development of the measurement scale and development of the questionnaire items. Validity and reliability measurements (Section 3.4) and statistical methods for data analysis (Section 3.5) follow. The section outlining steps in the data collection process (Section 3.6) discusses the pilot study and main study. Ethical considerations (Section 3.7) highlight the principles of voluntary participation, anonymity of respondents, confidentiality of data, and informed consent of respondents when conducting the survey. Limitations (Section 3.8) of this study are also included. Finally, the chapter offers a brief conclusion (Section 3.9). Figure 3.1 outlines the structure of Chapter 3.

69

Figure 3.1: Structure of Chapter 3 3.1 Introduction 3.2 Justification for the Research Paradigms 3.3 Research Design 3.4 Validity and Reliability Measurement 3.5 Statistical Methods for Data Analysis 3.6 Steps in the Data Collection Process 3.7 Ethical Considerations 3.8 Limitations 3.9 Conclusion
Source: Developed for this study.

3.2 Justification for the Research Paradigms

A paradigm is a basic set of philosophical beliefs about the nature of the world, which guides the thinking of researchers in the conduct of the research (Burrell & Morgan, 1979; Ticehurst & Veal, 2000). It reflects the fundamental position that researchers have to determine what problems are worthy of research exploration and the availability of the method to solve those problems (Guba & Lincoln, 1994). In general, a research paradigm is a whole system of thinking in the social sciences. It includes basic assumptions, the important questions to be answered or puzzles to be solved, the research techniques to be used, and examples of what good scientific research looks like (Neuman, 2006, p.81).

Although choice of paradigm has been the subject of long-standing debate on the adequacy of research methods in the social sciences, failure to think about the philosophical position
70

can seriously affect the quality of research management. That is, the research techniques for diverse methodologies are totally different from one other among the paradigms. For example, interpretative researchers are likely to collect, analyse, and interpret data simultaneously whereas positivist researchers aim to test hypotheses they previously developed (Neuman, 2006).

In terms of research quality management, there are three useful factors that may help the researchers understand the importance of philosophical beliefs. Firstly, they help to clarify the consequent research designs (Easterby, Thorpe, & Lowe, 1991). This involves an entire system of thinking with every piece of the research. This includes what kind of evidence is gathered, and from where, and how such evidence is interpreted in order to provide good answers to the basic research questions (Easterby et al., 1991, p.21). Secondly, they can help the researchers to recognize which designs will work and which will not work (Easterby et al., 1991). Systematic thinking at the beginning of the research helps to prevent wasting time with incomplete plans. Lastly, they can help the researchers identify designs that may be outside his or her past experience (Easterby et al., 1991). The researchers can therefore adapt research designs in relation to the constraints of different subjects or knowledge structures. Thus, it is believed that the researchers can easily reach their purposes as a result of the study if they establish what they believe about the nature of the world prior to the conduct of the research.

Since paradigms of the social sciences offer a variety of views about the nature of the world, they have been usually divided into a subjective and objective dichotomy. The subjective view is related to the researchers who look at reality as a projection of individuals consciousness whereas the objective view sees reality purely as a concrete structure (Chung & Alagaratnam, 2001). For example, phenomenological insights
71

represent the basic epistemological stance from a position of subjectivity. In contrast, the objective epistemology utilizes lab experiments to study the world. In general, the researchers with a subjective view represent an interpretative paradigm whereas the objective researchers represent a positivist paradigm (Morgan, 1980). However, the following positivist-interpretative paradigms will be discussed using ontology, epistemology, and methodology.

Ontology refers to the nature of reality that researchers fundamentally believe in (Denzin & Lincoln, 2008). The ontology of a positivist paradigm reflects a view that reality exists independently of the observer and is driven by immutable natural laws that take the form of a context-free generalization (Guba & Lincoln, 1994). These laws explain cause and effect relationships to establish predictable patterns of behaviour in the social world. For example, positivist researchers believe that innovation and creative work is a critical part of the decision to stay for the software development employees. In this view of reality, the association between innovation and creative work and software development employees intention to stay is an immutable natural law.

In contrast, the ontology of an interpretive paradigm reflects a view that reality is socially and experientially constructed (Guba & Lincoln, 1994). It indicates that everyone has a different view of reality so they construct their theories based on a form of multiple mental constructions. For example, interpretative researchers could explain that software development employees could stay longer in the company because of the enjoyment of trying new technologies and ideas. This has explained the view of reality with socially constructed meaning rather than as an immutable natural law.

72

Epistemology refers to the nature of knowledge that researchers fully possess and understand (Allison & Pomeroy, 2000). This explains the nature of the relationship between the knower (the researcher) and the knowable (the research). If utilizing the epistemology of a positivist paradigm, the researchers regard themselves as being objective, distant, non-interactive, value and bias free and as having no influence on either the research outcomes or the data they collect and analyse (Guba & Lincoln, 1994; McMurray, 2006). It is important for them to remain detached from the data they collect and analyse from the respondents.

In contrast, the epistemology of an interpretative paradigm freely admits that researchers may have their own set of values, beliefs, and attitudes which may bias the way they collect and analyse their data (Guba & Lincoln, 1994; McMurray, 2006). This is because the researcher and the object of investigation are interactively linked to the findings of an investigation which are the literal creation of the inquiry process (Guba & Lincoln, 1991). This indicates that there is no ontological and epistemological distinction in relation to the constructivist position.

Methodology refers to the ways in which researchers discover or create knowledge about the world (Denzin & Lincoln, 2008). It provides the tool or instrument to researchers that enable them to collect and analyse the data. The methodology of the positivist paradigm is regarded as the quantitative approach as it mainly measures variables and tests its hypotheses using descriptive and inferential data analysis techniques (Neuman, 2006). For example, quantitative researchers construct a theoretical framework and test the hypothetical relationship between independent variables such as innovation and creativity and a dependent variable such as decision to stay.

73

In contrast, the methodology of the interpretative paradigm is regarded as the qualitative approach as it mainly emphasizes detailed explanations of cases with socially constructed meaning (Neuman, 2006). It includes ethnographies, case studies, action research, and grounded theory. The methodological tools used in collecting and analysing data include observation, in-depth interviewing, and focus groups (Neuman, 2006). For example, the retention of software development employees could be explored through the in-depth interviewing of case studies.

In summary, research paradigms are important elements that researchers must justify prior to the conduct of the research. That is because they could seriously affect the quality of research management according to their philosophical position of beliefs in the social world. It is believed that standing on what are the beliefs of the nature of reality can lead to epistemological and methodological variations, which the researchers must be aware of. Table 3.1 summarizes differences of ontology, epistemology, and methodology between the positivist paradigm and the interpretive paradigm.

74

Table 3.1: Summary of research paradigms Positivist Paradigm Ontology (nature of reality) Epistemology (nature of knowledge) Methodology McMurray (2006).

Interpretive Paradigm Reality is multiple and will be constructed differently by people, depending on the meaning they make of their world Findings are subjectively created when individuals accounts of their reality converge Interpretation of the participants feedback information Small samples Qualitative methods

There is a reality that we are able to apprehend It aims to determine what is there Findings are objectively true Verified hypotheses are established as facts or laws Experimental verification of hypotheses Large samples Quantitative methods

Sources: Developed for this study based on Denzin & Lincoln (2008), Guba and Lincoln (1994), and

3.2.1 Justification for the Positivist Research

The positivist approach is traditionally a scientific research that justifies the managerial decision making in the social world (Davis, 2005). It is regarded as an effective problem-solving approach that involves an organized, systematic, critically objective, scientific inquiry into a specific problem that needs a solution. In addition, phenomenal interests of investigation are characterized by the testing of hypothetical relationships (Davis, 2005). The characteristics of positivist research are based on the beliefs that the social world exists externally and its phenomena can be measured through objective methods, rather than being interfered with subjectively through sensation or intuition (Easterby, et al., 1991). These authors have further identified the following elements in order to guide the research of positivism.

1. 2.

Independence: the observer is independent of what of what is being observed; Value-freedom: the choice of what to study, and how to study it, can be determined by objective criteria rather than by human beliefs and interests;
75

3.

Causality: the aim of social sciences should be to identify causal explanations and fundamental laws that explain regularities in human social behaviour;

4.

Hypothetico-deductive: science proceeds through a process of hypothesizing fundamental laws and then deciding what kinds of observations will demonstrate the truth or falsity of these hypotheses;

5.

Operationalisation: concepts need to be operationalised in a way which enables facts to be measured quantitatively;

6.

Reductionism: problems as a whole are better understood if they are reduced into the simplest possible elements;

7.

Generalization: in order to be able to generalize about regularities in human and social behaviour it is necessary to select samples of sufficient size;

8.

Cross-sectional analysis: such regularities can most easily be identified by making comparisons of variations across samples (Easterby et al., 1991, p.23).

These guidelines help to answer the research question that examined the relationship between retention factors and software development employees decisions to stay in the IT companies. The choice of the positivist paradigm for this research was justified for the following reasons. Firstly, the relationship between retention factors and software development employees decisions to stay exists objectively and externally. The ten retention factors naturally exist according to the literature, and none of them are changed or influenced by the researcher. Secondly, the researcher has a lower level of influence on collected data than he would in a qualitative study. While no research is wholly value free or fully objective, using a survey method with questions chosen from established instruments and statistical analysis places the research more towards the objective range of a methodological continuum. Finally, this research tests the ten hypotheses and explains how the ten retention factors affect the software development employees decisions to stay
76

in the IT industry. This study mainly examines the relationships of hypotheses rather than building a new theory. Therefore, positivist research is believed to be more suitable for this research as the research question is involved in causal laws.

3.2.2 The Process of Quantitative Methodology

Neuman (2006) has stressed that quantitative researchers often rely on a positivist approach to social science. The process of this approach is through quantitative techniques using surveys and statistics. This approach precisely measures identified variables and tests hypotheses that are linked to general causal explanations. More importantly, it is an organised method for combining deductive logic with theoretical framework. This consequently discovers and confirms a set of probabilistic causal laws that can be used to predict general patterns of human activity (Neuman, 2006). Therefore, quantitative research is believed to create knowledge through objective, controlled, statistical tests, measurements, stimulated exercises, and models.

Since quantitative research has been built upon the hypothetico-deductive method, it has created a seven-step process of building blocks of social science (Sekaran, 2000). This process begins with the interest of the researcher, and is followed by preliminary information gathering, theory formulation, hypothesizing, further scientific data collection, data analysis, and deduction. Davis (2005, p.59) has also stressed that the direction of the deductive method begins with abstract concepts and theoretical relationships then works towards more concrete empirical evidence.

This research began with the researchers interests in the software development employees in the IT industry as important assets of companies. It is believed that not retaining those important assets would be a risk to the success of companies as voluntary turnover rate is
77

as high as 86% in the IT industry in Taiwan (DGBAS, 2007). This demonstrated the deductive process of the researchers interests at the beginning of the quantitative research.

Skill shortages in the Taiwanese workforce generally, and a shortage of software development employees in the Taiwanese IT industry were two main problems in this study that were identified as the problems or factors for the preliminary information gathering. Both factors were heavily impacting on the economic development of the IT Industry in Taiwan as a consequence of globalization in the knowledge-based economy. These problems indicated that retention strategies were a matter of concern for the Taiwanese IT industry and researching the problem could offer guidelines for the Human Resource Departments regarding staffing of the IT industry in Taiwan.

A conceptual model of this research was then established based on the review of literature, and a theoretical framework with five main hypotheses was also developed. It constructed the decision to stay associated with the variables: financial compensation, career development (training and development and promotion), self-achievement (recognition, challenging work, and innovation and creativity), environmental factors (leadership style, autonomy, and work-life balance), and job satisfaction. These identified variables were tested for the causal explanations of their relationships with software development employees decision to stay.

Finally, a research design was set up to decide on how to collect further data, analyse and interpret those data, and provide an answer to the identified issues (Sekaran, 2000). This is the deductive process (see Figure 3.2) drawing from the logical analysis of scientific research.

78

Figure 3.2: Flow of Deductive process

Observation

Literature Review

Formulating a Theory

Hypothesizing

Data Collection

Data Analysis

Deduction
Source: Developed for this research based on Sekaran (2000).

In summary, quantitative methodology provides a way that creates knowledge through a process of hypothetical deduction methods. It is important to offer such guidelines for the researcher to use when conducting the research. That is, the principles of the positivist approach should follow the hallmarks of scientific research.

3.3 Research Design

Research design is a master plan that specifies methods and procedures for collecting and analysing the needed information (Zikmund, 2003). This is a framework that plans the action for the scope of the study as it provides appropriate techniques to conduct the research. In fact, it sets up the preferences of the researcher with the sources of information, the method of data collection, the sampling methodology, explanation of the research variables, and statistical methods for data analysis.

3.3.1 Methods of Data Collection

There are four typical types of survey methods for quantitative research and they are telephone interview, face-to-face interview, mail questionnaire survey, and web-based
79

survey (online survey). The first two interview-based surveys are administered by the interviewer during the process, whereas the last two surveys are self-completed by respondents on their own. The choice of a survey method depends highly upon the needs of the researcher. There is no best form of survey as each of the surveys has its own advantages and disadvantages (Zikmund, 2003). Table 3.2 provides advantages and disadvantages of the four surveys.

Table 3.2: Advantages and Disadvantages of the four surveys Face-to-face interview Administrative issues Cost Speed Questionnaire length Anonymity of respondent Response rate Research control Open-ended questions Complex questions Sensitive questions Source of bias Interviewer bias Respondent's reading skill Worse No Some No No Yes No Some Expensive Slow to moderate Longest Low Highest Yes Moderate Fast High Short Moderate Moderate Yes Cheap Slowest High Moderate High Lowest No Cheapest Fastest Highest Moderate High (web-based) Low (email) Moderate No Telephone interview Mail survey Online survey

Geographic flexibility Low

Success with different questions Yes Yes Limited Limited Limited Limited Limited Limited Some Yes Yes Yes

Source: Developed for this research based on Neuman (2006) and Zikmund (2003).

Face-to-Face Interviews Face-to-face interviews gather information through a form of direct personal communication. This is a two-way conversation between an interviewer and a respondent
80

and can take place in an office, shopping mall, home or in another setting (Zikmund, 2003).

The advantages of face-to-face interviews are the opportunities for feedback and follow-up questions, a higher response rate, and a longer questionnaire. For instance, sensitive information can be gathered by a feedback process in which interviewees answers can be guaranteed to be strictly confidential. In the follow-up questions, it offers a probing process that allows the researcher to ask for a clear or more comprehensive explanation if a respondents answer is not clear.

The disadvantages of face-to-face interviews are the higher cost of conducting interviews, restriction of geographic flexibility and lack of anonymity of respondents. This type of interview is generally more expensive than mail, internet, and telephone surveys because it is affected by the number of respondents, and geographic proximity of respondents (Zikmund, 2003). Anonymity of respondents is another big challenge because respondents may be reluctant to answer sensitive questions because the process of the interview is not anonymous.

Telephone Interviews Telephone interviews allow respondents to provide detailed and reliable information on a variety of personal topics over the telephone in comparison to face-to-face interviews (Zikmund, 2003).

The advantages of telephone interviews are the speed of data collection, cost of conducting interviews and geographic flexibility. According to Zikmund (2003), the speed of data collection is the major advantage of telephone interviewing. That is, hundreds of telephone
81

interviews can be conducted literally overnight whereas mail surveys or face-to-face interviews can take several weeks.

The disadvantages of telephone interviews are the length of interviews, the limit of open-ended questions, and the limit of complex questions. Zikmund (2003) has stressed that the length of interviews is the major disadvantage of telephone interviewing. That is, respondents could simply hang up the phone if they feel they have spent too much time in the interviews. For example, 10 minutes is acceptable, and 30 minutes is the maximum acceptable time (Zikmund, 2003).

Mail Questionnaire Surveys A mail survey is a self-administrated questionnaire sent by mail to people who qualify for the study. This is a paper-and-pencil method and is commonly used to collect data from a sample of people.

The advantage of a mail survey is that a wide geographical area containing a large number of samples can be covered at a relatively low cost because interviewers are not required. Typically, this survey applies to target respondents who live in rural and urban areas who are difficult to reach but can be contacted by mail (Zikmund, 2003). Mail surveys also offer respondent anonymity, confidentiality, and avoid interviewer bias because respondents can complete the surveys at their own convenience in their home.

The biggest disadvantage of mail surveys is that response rates can run at less than 5 percent in random samples of the general population (Davis, 2005). Another disadvantage of mail surveys is the time taken in data collection. According to Zikmund (2003), a minimum of 2 to 3 weeks is necessary to receive the majority of responses, and usually, 6
82

to 8 weeks is an adequate time to process these, including follow-up mailings. The process of this survey is also beyond researchers control because it is a self-administrated questionnaire.

Online Surveys An online survey is a new survey method being used in the highly penetrated internet markets of developed countries (Davis, 2005). This survey entails two types of techniques; email and web-based. Both surveys are convenient because they enable the respondents to complete the questionnaires by either replying back to the email or visiting the web site. To some extent, a web-based survey preserves the anonymity of respondents in comparison to email techniques (Sue & Ritter, 2007). The response rate for a web-based survey is approximately 30% (Sue & Ritter, 2007).

An online survey has similar advantages to the mail questionnaire survey, including the cheapest cost, highest speed, and geographic flexibility for respondents. This type of survey can reach a large audience (possibly a global one) and secure confidential answers quickly and cost effectively (Zikmund, 2003). Davis (2005) has stressed that an online survey has the many advantages of being easy to administer, interactive, timely, capable of including multimedia, and low in cost. For example, this survey allows utilization of colour, sound, and animation, which may help to increase respondents cooperation in answering the questionnaires.

The disadvantages of the online surveys are coverage, and computer software. The coverage of the sampling may fail to reach respondents who are older, less educated, have a lower-income, and rural people who have less or no internet access (Neuman, 2006). The other concern regards compatibility issues of computer software and hardware
83

combinations for respondents who may be unable to get access because they use different types of computers (Neuman, 2006).

Justification for Selection of Web-Based Online Surveys This research was to examine the relationship between retention factors and software development employees decisions to stay in the IT industry in Taiwan. The participants were full-time employees working in research and development (RD) departments of companies located in HsinChu Science Park. Based on the analysed information of four surveys (Table 3.2), a web-based online survey was adopted for this research for the following reasons.

1.

A web-based online survey provides the cheapest cost and fastest speed from among the four surveys.

2.

A web-based online survey is highly anonymous for respondents so they can answer all questions without any concern.

3.

A web-based online survey has no interviewer bias when compared to face-to-face interviews and telephone interviews.

4.

A web-based online survey is an appropriate method for the software development employees in the IT industry because they are highly computer literate (the disadvantage concern of the coverage was removed) and they get internet access every day.

5.

A web-based online survey is an appropriate method for the computer literate people as the internet access infrastructure in Taiwan is quite well-developed.

6.

This web-based online survey is through SCU Monkey Survey Server, which offers the questionnaire in HTML form. This form of the survey is compatible with Microsoft-related software, which is popular in Taiwan (the disadvantage
84

concern of the computer software issue was removed).

3.3.1.1 Sample of Target Population

The sample of the study was drawn from software development employees in the IT industry in Taiwan. This is because the researcher comes from Taiwan and has a background in the IT industry. Also, the researcher is interested in identifying human resource issues in terms of software development employees retention in the Taiwanese IT industry. Zikmund (2003) has pointed out that the choice of the target population of the sample is important for the study because of constraints of time and budget in the conduct of the research. Sue and Ritter (2007) have also stressed that collecting data from every member of the target population is neither feasible nor practical. Thus, the target population of the sample was Taiwans IT companies with a focus on their existing full-time software development employees (e.g. firmware or software programmers).

The rationale for choosing existing full-time software development employees is that they daily execute the development of important projects with their customers. They are at the operational level of the organizational structure, as they usually innovate or create new products through programming of the software. Thus, it is important to identify whether their decisions to stay or leave the organization are associated with identified financial compensation, career development, self-achievement, environmental factors, and job satisfaction.

3.3.1.2 Sampling Frame

A sampling frame can help to narrow the scale of a study since researchers can make conclusions about population characteristics on the basis of the target population of the sample (Zikmund, 2003). Thus, the researcher chose one representative of the
85

high-technology industrial base for the research. Taiwan has many high-technology industrial bases such as HsinChu Science Park (HCSP), Tainam Science Park, and others, which are all fully sponsored by the central government. HsinChu Science Park in Taiwan is the oldest and biggest industrial base. It consists of six main industries with a total of 384 companies and 115,477 employees, and is the major contributor to industrial economic performance with total revenues of US$32.5 billion in 2004 (HCSP, 2009). This is a world-recognised science park and is representative of Taiwans industrial development, and therefore, the research was geographically focused on HsinChu Science Park.

Six main industries in HsinChu Science Park are: integrated circuits, PC/ peripherals, telecommunication, optoelectronics, precision machinery, and biotechnology. They all strongly value intellectual capital and are fostered by the central government. It is believed that the development of intellectual capital is the only way to transform Taiwan from labour-intensive-industries and technology-industries to knowledge-based industries. In addition, the integrated circuits industry was the most economically valuable contributor to the HsinChu Science Park, which generating US$22,309 million in 2004, with a major total of 164 companies (HCSP, 2009). The following industry was the PC/ peripherals industry, which generated US$4,147 million in 2004, with a total of 58 companies (HCSP, 2009). These two industries were specifically targeted for this study as they were highest economic contributors to the HsinChu Science Park.

According to the statistics of the workforce in HsinChu Science Park (HCSP, 2009), Park companies employed 10,918 R&D-related professional employees, representing 11% of total employees. In fact, the integrated circuits companies employed 6,038 of these professionals, representing 55.3% of them, and the PC/ peripheral companies employed 1,926 of these professionals, representing 17.6% of them (HCSP, 2009). However, there is
86

no evidence about the number of software development employees employed at HsinChu Science Park, so for the purpose of response rates, this research aims to identify them through inquires made to the HR managers of the selected IT companies.

Forty IT companies for this study have been targeted from the integrated circuits industry and the PC/ peripheral industry in the HsinChu Science Park. The size of those IT companies was chosen from a range of between 100 employees and 500 employees. It has been identified that this range of companies has the highest voluntary turnover rate in the IT industry (DGBAS, 2007). The benchmark for those IT companies chosen was measured on the basis of the performance of their stock price in the market. It is believed that the stock price is a good index to reflect the performance of the companies with sustainable competitive advantage in the identical industry.

3.3.1.3 Sampling Plan

As benchmarks for the selection of those top forty IT companies were set up, it was important to approach those companies and invite them to participate. The following stages describe how the participants were approached and invited to participate in the survey.

1.

An invitation letter on SCU letterhead was sent to the managers of human resource departments of those selected IT companies. It was passed on through the contact window found from the companies web sites. It explained the purpose of the research in plain Mandarin and that the research had SCU HREC ethics approval, and invited participation. The research process was clearly explained and the HR managers were kindly asked for the group email address and number of software development employees in their companies (see
87

Appendix A). 2. The survey was not begun until there was a return consent form signed by the director of the human resource department or by higher position managers of the selected companies (see Appendix B). 3. The plain Mandarin questionnaire was stored on the Survey Monkey Server (SCU online survey server) for the participation of the survey (see Appendix C). 4. A notification email was sent to the participants, as a group email address of software development employees was provided in the consent form. 5. The researcher acted as a facilitator to monitor the online activity during the period of the survey.

In order to effectively increase the ease of research implementation, the invitation letter and the follow-up activities were made via email and telephone calls. De Vaus (1995) has stressed that follow-up activities are needed to achieve higher response rates as each follow-up provides opportunities to persuade respondents to complete the survey.

Field work for the online web-based survey was conducted within two months. However, the researcher encountered difficulties with the selected IT companies in the field work so the final data collection was authorised through the Human Resource Bank in Taiwan. Human Resource Bank is a job-seeker company, which holds personal information of the job-seekers in the database system. These difficulties are discussed in the next section of risk management.

88

3.3.1.4 Risk Management

There are two sub-sections in risk management, including pre-risk management and post-risk management. Pre-risk management is a precaution plan for something that could happen in the field work, whereas post-risk management is a reaction to something that happened and how a researcher acted on it.

3.3.1.4.1 Pre-risk management

Pre-risk management provides a researcher an alternative or another way around a solution. It is believed that the more constructive plans the researcher prepares the more successful the research itself will be. The scenarios below describe possible situations that the researcher might encounter during the field work, along with their countermeasures.

Scenario 1 Situation: what would happen if more than half of the selected IT companies were unwilling to participate in the survey? Countermeasure 1: the researcher would keep rolling to find out more IT companies within the integrated circuits industry and PC/peripherals industry. Countermeasure 2: the researcher would contact his professional networks to increase the possibility of more IT companies participating.

Scenario 2 Situation: what would happen if the number of respondents was less than the researchers expectation? Countermeasure 1: the researcher would kindly ask those participating IT companies to introduce other possible IT companies for participation as a method of snowball sampling.
89

Countermeasure 2: the researcher would officially contact his previous IT companies and invite them to participate, as the researcher used to work in the PC/ peripherals industry and telecommunication industry.

Scenario 3 Situation: what would happen if there was no group email address of R&D departments in the participating IT companies? Countermeasure 1: the researcher would ask for an upper level of business unit group email address, in an attempt to eliminate the involvement of people unnecessary to the survey. Countermeasure 2: alternatively, the researcher would kindly ask participating human resource people to relay the survey email (see Appendix D) to their R&D employees.

Scenario 4 Situation: what would happen if participating R&D department people regarded the survey email as junk email? Countermeasure 1: the content of the survey email (see Appendix D) would address that this survey was agreed to by your company, and would provide messages about the purpose of the survey, a short introduction to the researcher, and school information in the content. Countermeasure 1.1: the content of the survey email would also highlight that the software development employees were the focus of the study, and that it might take 10 to 15 minutes to complete. Countermeasure 1.2: the anonymity and confidentiality of the data would be highlighted in the content of the survey email and the hyperlink in a large font would be provided.

90

3.3.1.4.2 Post-risk management

In reality, things will not always follow the plan. So, the stories below were real situations the researcher encountered during the field work arrangement. They also include the analysis of the real situations and some actions the researcher took, as well as the final conclusion and the decision.

Story 1 Real Situation: only 5 out of 40 IT companies were willing to participate in the survey. Analysis: in the phone call conversations with HR contacts, it was sensed that most of them were hesitant to be involved in what they considered as irrelevant job tasks, especially outside research projects. In addition, the contacts might not have had the authority to make a decision and on the other hand, it was a trouble for them to get permission from the upper level of management. So, they simply rejected the participation. Action 1: the researcher kept contacting more companies within the integrated circuits industry and the PC/ peripheral industry. Action 2: the researcher utilized his network to invite more IT companies from the Telecommunication industry. Action 3: the researcher sought advice from his supervisor for an alternative solution, and the possibility of using the Human Resource (HR) Bank to conduct the survey. The HR Bank in Taiwan is a commercial job-seeker company, which involves a variety of businesses. The HR Bank holds personal information on job-seekers in its private database system and offers an online public platform for exchanging information for both individuals and companies. The company was able to distribute the survey information (Information Sheet and online survey address) to each individual who

91

worked as a software development employee in the area of the HsinChu Science Park in Taiwan. The company also required a certain amount of money for the service that distributed the survey information required by this study.

Story 2 Real Situation: all 5 participating IT companies asked the researcher to send them the email with the survey access link. Analysis: this consequence reminded the researcher that a group email address could be confidential property and that the rules of the company could be violated if it was released to the public. Especially as junk email has seriously spread out in the world. It was a good learning experience for the researcher. Action: the researcher sent them the survey access information and gained the confirmation by receiving the emails.

Story 3 Real Situation: two months had elapsed since the day the survey had begun and only 12 respondents had participated in the survey. Analysis: follow-up was rarely enforced due to the survey email being relayed, and there was no obligation on the HR contacts as this was a voluntary survey. In addition, there was a big possibility that the relayed email could be regarded as junk email. Action: the researcher sought advice again from his supervisor and proposed that the Human Resource Bank was a better alternative solution to conduct the survey as they have a wider range of the software development employees emails stored on a private database.

92

Final conclusion The researcher and his supervisor agreed that the survey would be conducted through the Human Resource Bank on the basis of the following requirements. 1. The rights to the collected data belonged to the researcher and the research. 2. The collected data could not be exposed to anyone else. 3. The process of the data collection should not violate the ethics issues as the survey has to be voluntary, anonymous, and the confidentiality of the data must be protected.

Decision making: The conduct of the survey for this research was authorised through the Human Resource Bank in Taiwan.

Management of Human Resource (HR) Bank The researcher managed the process with the HR Bank to ensure that this company followed ethical procedures and surveyed the right people, that is, those who work as software development employees in the area of the HsinChu Science Park in Taiwan. Based on the requirements of the business agreement for conducting the survey, many meetings were set up to communicate these requirements to the representative of the HR Bank in Taiwan. The focus of these meetings was to ensure that the representative of the HR Bank totally understood the ethical issues in relation to the target participants, who must be voluntary, anonymous, and confidentiality of the collected data must be protected. The conduct of the survey was finally established between both parties based on the mutual understanding of the ethical issues and the requirements of the survey.

93

The focus on software development employees who work in the area of the HsinChu Science Park in Taiwan was ensured through a filtered combination of different categories of job occupations and working areas as recorded in the personal information in the HR Banks database system. This mechanism of the survey requirements was set to target participants who are full-time employees in the IT industry and work as firmware or software programmers in the HsinChu Science Park in Taiwan. Question 5 in Section II of the survey (see Appendix C) filtered target participants who came from software research and development departments rather than other departments.

The total target population for this study was 1,150 software development employees. Their membership was empirically confirmed with the HR Banks database system, which aided in the dissemination of the survey. This target population number has been approved by the mechanism of the survey requirements and Question 5 in Section II of the survey (see Appendix C) in which participants are randomly selected from a total number of 6,943 participants (emails). Finally, the raw data from these participants was passed on to the researcher in the Excel file format.

3.3.1.5 Sample Size

Sample size in quantitative research can affect the results of findings if the sample is too small. However, large samples can undermine the efficiency of research and also increase the cost of research (Ruane, 2005). Therefore, sample size presents the efficiency and economy of research in the social sciences. Ruane (2005) also stresses that when it comes to samples, researchers should employ sampling ratio that establish acceptable sample size for various population sizes. In small populations (1,000), a research needs a large sampling ratio of 30% (300) for a high degree of accuracy (Neuman, 2004). Furthermore, Hair, et al. (2010) stress that the researcher generally should have at least five times as
94

many observations as the number of variables to be analysed for factor analysis. Thus, 415 returned samples were available for factor analysis as a minimum of sample size required 250 (5x50 independent variables=250 samples). Manning and Munro (2007) also stress that a sample size range between 300 and 500 is good for data analysis.

Since the data collection was undertaken through the Human Resource Bank in Taiwan, the total target population for this study was 1,150 software development employees. In addition, there were 60 out of 150 respondents who effectively participated in the first run of the pilot survey reaching a response rate of 40%. In order to achieve a similar response rate to that of the pilot survey, there were 415 respondents from a total of 1000 respondents who were invited to participate in the main survey, giving a response rate of 41.5% for the main study.

3.3.1.6 Probability and Non-Probability sampling

Sampling methods have been grouped into two types, namely probability techniques and non-probability techniques. Probability sampling refers to the concepts of a random sample selection in which every member of the population has an equal probability of being selected. It includes simple random sampling, systematic sampling, stratified sampling, and cluster sampling. On the other hand, non-probability sampling is non-random and selects on the basis of personal judgment or convenience. It includes judgment or purposive sampling, quota sampling, convenience sampling, and snowball sampling. Table 3.3 below provides a summary of advantages and disadvantages of probability and non-probability sampling methods.

95

Table 3.3: Summary of advantages and disadvantages of sampling designs Type Simple random sampling Systematic sampling Stratified sampling Cluster sampling Judgment or Purposive sampling Convenience sampling Quota sampling Snowball sampling Brief description
Use by simple random selection

Advantages
- High generalisability of findings - Requires minimum knowledge of population

Disadvantages
- Not as efficient as stratified sampling - High degree of error on sample size - Systematic biases are possible

Probability sampling

Use by sample interval number of population Use by stratified method on sample selection Use by group type on sample selection

- Easy to use if sampling frame is available - Simplicity - Most efficient among all probability designs - Ensures equal representation - Lowest field cost - Use for subsequent sampling

- Stratification must be meaningful

- Requires ability to assign members into different clusters

Non-probability sampling
Use by purposive sample selection - Lower field cost - Faster - Generalisability is questionable - Requires strong assumptions Use by convenient manner Use by equal proportion from each class Use by referral on sample selection - Highly specific application - Not normally generalisable - Faster - Cheaper - Lower field cost - Not easily generalizable - Not generalisable at all

Source: Developed for this research based on Cavana et al. (2001) and Davis (2005).

The probability sampling method was used for this research, as it is believed to provide significant results about the findings in quantitative research (Garson, 2002). This technique provides an ideal method of sampling, as it allows researchers to draw statistical generalizations from the samples to the total population (Lee & Lings, 2008). Kumar (2005) has also stressed that some of these statistical tests are important to establish conclusive correlations.

96

Cluster sampling in the probability technique was used for this research. As the entire sample population of software development employees is located in many cities in Taiwan, area sampling is the most popular type of cluster sampling (Zikmund, 2003). This is because the benefit of area sampling can yield lowest field cost while retaining the characteristics of a probability sample for the study (Zikmund, 2003). This approach saved the researcher a lot of time and money with post and phone calls to invite participation in the survey, as Taiwan has many industrial science parks spread across many cities in the country. The candidate needed to make personal contact with key decision maker in the organisations. In addition, the reason for choosing HsinChu Science Park is because it is the largest and oldest industrial science park in Taiwan. It is deemed that the results from this cluster sampling could closely represent the total population of software development employees in Taiwan, however findings will be interpreted conservatively.

3.3.2 Questionnaire Design

Relevancy and accuracy are two major concerns associated with questionnaire design. Relevancy refers to questions being asked that are relevant to the identified variables, whereas accuracy refers to information that is reliable and valid (Zikmund, 2003). Generally, it relates to principles of wording, which include content and length, wording and language, and form of questions (Sekaran, 2003).

Content and Length Content of questions should properly measure respondents attitudes, beliefs, and opinions based on the variables of a conceptual model (Sekaran, 2003). In this research, there were five main hypotheses with ten independent variables and one dependent variable which were identified from the literature review in Chapter 2. The independent variables were financial compensation, training and development, promotion, recognition, challenging
97

work, innovation and creativity, leadership style, autonomy, work-life balance, and job satisfaction. The dependent variable was the decision to stay.

The length of the web-based survey required respondents to spend about 10 to 15 minutes, which was acceptable for this study. It included 55 interval scale questions in Section One, 8 questions for personal information in Section Two, and one voluntary open-ended question in Section Three (see Appendix C).

Wording and Language Words and wording used for this research were reviewed by the researchers previous colleagues who work as software development employees in IT companies in Taiwan. This ensured that target respondents were properly communicating and using the same language. These were also reviewed by an expert panel of professors and DBA student peers at Southern Cross University. This ensured that the questions were developed based on the art of asking questions, which is using simple and conversational language, and avoiding leading and ambiguous questions (Zikmund, 2003). More importantly, the respondents are Taiwanese so the language of the questionnaire was in plain Mandarin. Thus, a translation of the English version to plain Mandarin was conducted by the researcher and reviewed by a third party who is a teacher at a senior high school in Taiwan. This provided significant help in ensuring that the meaning of the content was precisely expressed in short sentences from the long sentence translation. For instance, the term IT professionals in Mandarin was suggested to replace the term IT professional workers.

Form of Questions Open-ended questions and fixed-alternative questions (closed questions) are two types of phrase questions usually used for surveys. Open-ended questions are where respondents
98

answer the questions in her or his words whereas fixed-alternative questions are where respondents answer the questions from specific limited-alternative responses (Zikmund, 2003). This research was mainly designed for fixed-alternative questions for Section One and Section Two. Section Three was a voluntary open-ended question.

Welcome Message The welcome message was included in the process of data collection which was fully covered in Section, 3.3.1.4 Risk Management. Sue and Ritter (2007) have stressed that the welcome screen can motivate respondents by providing an opportunity to describe the purpose of the survey and discuss the conditions of anonymity and confidentiality. This would be significantly to increase the response rate. It is deemed that welcome message would have improved the participation of the survey as a result of high response rate of 41.5% for the study.

3.3.2.1 Development of the Measurement Scale

A scale is a tool or mechanism that distinguishes the difference of individuals from one another in relation to the variables of interest to the study. It explains abstract concepts by measuring how an individual feels or thinks about variables instruments in the process. A Likert-type scale is widely used for interval data and is usually designed with agree and disagree categories using four to eight points. Thus, the questionnaires for this research were designed using a five-point scale, where 1 indicates that respondents strongly disagree and 5 indicates that respondents strongly agree.

There are four levels of measurement scale, which include nominal scale, ordinal scale, interval scale, and ratio scale. A nominal scale is used to assign subjects to certain categories or groups. Essentially, there is no relationship between them. For instance, the
99

variable of gender can be grouped into two categories: male and female.

An ordinal scale has all the characteristics of the nominal scale plus the rule that it rank-orders (in either ascending or descending order). For instance, an income question can be grouped into categories and rank-orders like above average, average, and below average.

An interval scale has all the characteristics of the ordinal scale plus the rule that the unit is measured with an arbitrary starting and terminating point (Kumar, 2005). For instance, a financial compensation question can be this company pays me fairly for the work I do. The response choice can be 1=strongly disagree, 2=disagree, 3=neutral, 4=agree, 5=strongly agree.

A ratio scale has all the characteristics of nominal, ordinal, and interval scales plus a rule that it has a fixed starting point (Kumar, 2005). For instance, a financial question can be what is your companys annual revenue? The response is fixed with a dollar value on the answer.

In this study, there were five main hypotheses with ten independent variables and one dependent variable. The first section of the questionnaire contained 55 questions using a 5-point interval scale, and the second section contained 8 questions for personal information. Table 3.4 below provides a summary of the questionnaire structure with the measurement scale.

100

Table 3.4: Questionnaire structure with measurement scale

Variables Financial compensation Career Development Training and Development Promotion Self-achievement Recognition Challenging work Innovation and Creativity Environmental factors Leadership style Autonomy Work-life balance Job satisfaction Retention Personal Information Gender Age Working period in the company Working period in the IT industry Department Position Salary Academic Qualification Source: Developed for this study. Q5 Q6 Q7 Q8 Nominal scale Nominal scale Ordinal scale Ordinal scale Q4 Ordinal scale Q1 Q2 Q3 Nominal scale Ordinal scale Ordinal scale H4a H4b H4c H5 Q31 to Q35 Q36 to Q40 Q41 to Q45 Q46 to Q50 Q51 to Q55 5 point interval scale 5 point interval scale 5 point interval scale 5 point interval scale 5 point interval scale H3a H3b H3c Q16 to Q20 Q21 to Q25 Q26 to Q30 5 point interval scale 5 point interval scale 5 point interval scale H2a H2b Q6 to Q10 Q11 to Q15 5 point interval scale 5 point interval scale Hypotheses H1 Questionnaire items Q1 to Q5 Scales 5 point interval scale

3.3.2.2 Development of the Questionnaire Items

The development of questionnaire items refers to the measuring process for each variable in a research model. This presents the association among the measured indicators in observable and empirical terms. Neuman (2006) has stressed that it provides a way that rigorously links abstract ideas to measurement procedures that produce precise quantitative

101

information about empirical reality. In other words, it enables the facts to be measured quantitatively. The measured indicators of questionnaire items were mainly retrieved from the existing sources (Connolly & Connolly, 2005), which ensures the accuracy of information it gathers.

Financial Compensation Financial compensation variable was mainly focused on monetary related matters, which measure the perception of the software development employees for the fairness of the work compensation (Connolly and Connolly, 2005). One example question was this organization pays me fairly for the work I do. The details of the questions are in Appendix C (Section I, Q1-Q5). A 5-point response scale was used in the questionnaire.

Career Development Career development consists of two variables including training and development, and promotion. Connolly and Connolly (2005) have stressed that these variables determine employees overall engagement in their jobs and the organization. They measure their perceived value of the work and future career growth. One example question for the training and development variable was I am given opportunities to improve my skills in this organization. Another example question for the promotion variable was I believe my career aspirations can be achieved at this organization. The details of the questions for the two variables are in Appendix C (Section I, Q6-Q15). A 5-point response scale was used in the questionnaire.

Self-Achievement Self-achievement consists of three variables including recognition, challenging work, and innovation and creativity. Connolly and Connolly (2005) have stressed that a sense of
102

accomplishment and a risk taking policy is measured by the employees of the company. One example question for the recognition variable was I am adequately recognized for my contributions. The other example question for the challenging work variable was My job is challenging. The last example question for the innovation and creativity variable was I am given an opportunity to present and try new ideas. The details of the questions for the three variables are in Appendix C (Section I, Q16-Q30). A 5-point response scale was used in the questionnaire.

Environmental Factor The environmental factor consists of three variables, including leadership style, autonomy and flexible work arrangements (work-life balance). Connolly and Connolly (2005) have stressed that employees always measure their managers, their decision making process, and free leisure time from the work in the organization. One example question for the leadership style variable was My manager involves me in decisions that affect my job. Another example question for the autonomy variable was I have the freedom I need to do my job. The last example question for the work-life balance variable was I am provided with the flexibility needed to balance the demands of my work and personal life. The details of the questions for the three variables are in Appendix C (Section I, Q31-Q45). A 5-point response scale was used in the questionnaire.

Job Satisfaction Connolly and Connolly (2005) have stressed that employees always feel something about the work they do. They measure personal sense of control over the work, sense of satisfaction, sense of accomplishment, and ability to see how the work fits in. One example question was I like my job at this organization. The details of the questions are in Appendix C (Section I, Q46-Q50). A 5-point response scale was used in the questionnaire.
103

Decision to Stay This reveals employees beliefs about the future of the company, work-life balance, and interactions with management (Connolly & Connolly, 2005). This measures employees decisions to stay within the organization and to give fully to the goals of the organization and the tasks of the individual job (Connolly & Connolly, 2005). One example question was I see myself working for this organization three years from now. The details of the questions are in Appendix C (Section I, Q51-Q55). A 5-point response scale was used in the questionnaire.

3.4 Validity and Reliability Measurement

The quality of research depends on prudent measurement techniques, as it requires accuracy and consistency in the quantitative research. The reliability refers to the consistency and stability of a score from a measurement scale while the validity is the degree to which the scale represents and acts like the concept being measured (Davis, 2005). They complement one another to present the values of the quantitative research.

Validity There are three types of validity which include face and content validity, concurrent and predictive validity, and construct validity. Face validity establishes the logical link in an instrument. Each question or item on the scale must have a logical link with an objective (Kumar, 2005). Content validity is the assessment that the items or questions equally cover the full range of the issues or attitudes being measured (Kumar, 2005). It has been argued that the judgment of the instrument may involve subjective logic which results in that no definite conclusions can be drawn.

104

Predictive validity is judged by the degree to which an instrument can forecast an outcome whereas concurrent validity is judged by how well an instrument compares with a second assessment concurrently done (Kumar, 2005). It has been argued that they may highly involve subjective judgment on the selection of an appropriate criterion for validation purposes.

Construct validity is a more sophisticated technique than others using statistical procedures (Kumar, 2005). It involves the contribution of each construct to the total variance observed in a phenomenon. This shows that the greater the variance attributable to the constructs, the higher the validity of the instrument is. In addition, SPSS software was used to check homogeneity of items for validity measurement.

Reliability There are three methods of reliability measurement, which include the test-retest, alternative form, and internal consistency (split-half and Cronbachs Alpha). The test-retest method refers to measuring the same set of individuals at two different times for stability. It is believed to be hard to execute in a questionnaire because of constraints of attitudes which may change over time (Zikmund, 2003). Alternative form refers to measuring the same set of individuals over time with a different questionnaire designed for reliability. It has been argued that reliability may be affected by the alternative form itself.

Split-half method refers to a measurement of the same set of the data that computes the first half of the scale items and checks them against the results from the second half of the items. It also has been argued that the result may depend on the manner in which the items are split. Cronbachs Alpha refers to a measurement of the same set of data that computes the mean reliability coefficient estimates for all possible ways of splitting the set of items
105

in half. This presents a better estimate of reliability and is widely accepted by the quantitative researchers. To achieve a higher coefficient, the pilot study was also used by pre-testing the questionnaire for the main study.

3.5 Statistical Methods for Data Analysis

Data analysis is a process of interpreting the information to discover the reality of the results. The purpose of data analysis is to achieve the research objectives and to answer the research problems the researchers set up. The procedure of conducting the data analysis begins with selection of computer software, preparation for data analysis (measurement), identification of specific statistical techniques, and presentation of the analysis (Davis, 2005).

The Statistical Package for the Social Science (SPSS version 17.0) was chosen as the analytical software for this research. It has been widely accepted by quantitative researchers for data analysis as it provides a Windows-based module and is user-friendly and easy-to-use for the researchers.

Measurements for validity and reliability were used for this research. Validity is ensuring that the instrument measures what it is supposed to measure (Leedy & Ormrod, 2005). In other words, whether the questions asked in the survey are highly correlated with each other as one construct is a general rule of thumb for measurement. This is called homogeneity of items. Gregory (2000) has stated that homogeneity can be evaluated by correlating every component item with the composite variable (called item-total correlations), and by correlating each item with each other item (called inter-item correlations). It is noted that the value of the item-total correlations should exceed .50 and the inter-item correlations should exceed .30 for preliminary data analysis (Hair, Anderson,
106

Tatham, & Black, 1998). It is also noted that the value of the multicollinearity (refers to the correlation among three or more independent variables) should not exceed .90 for multiple regression (Tabachnick & Fidell, 2001).

Reliability is measured with the consistency of the result it intends to measure. In other words, data collected from the same respondents should give the same result over different time periods. It is concluded that reliability and validity are often used in connection with measurement, and reliability is related to consistency whereas validity is related to accuracy (Davis, 2005). Measuring reliability is a good way to show evidence of consistency within the construct (Coakes, Steed, & Dzidic, 2006). Gregory (2000) has stressed that coefficient Alpha is a useful internal consistency technique to estimate reliability, where 1 indicates perfect reliability and 0 indicates no reliability. A widely used measure of reliability of a composite variable is Cronbachs Alpha. Hair et al. (1998) have suggested that exploratory research could reduce the value to .60 as the lower acceptable limit for Cronbachs Alpha, but generally the value should exceed .70, which is considered to represent acceptable reliability, above .80 is good reliability, and above .90 is excellent reliability.

The choice of data analysis techniques is determined based on the research objectives that answer the research question. The research question was to assess the retention factors associated with software development employees decision to stay. Tabachnick and Fidell (2001, p. 17) stress that If the major purpose of analysis is to assess the associations among two or more variables, some form of correlation/ regression or chi square is appropriate. Hair, Black, Babin, and Anderson (2010, p. 23) also stress that With the objective and conceptual model specified, the researcher has only to choose the appropriate multivariate technique based on the measurement characteristics of the
107

dependent and independent variables. Thus, regression will be used to examine the retention factors in association with software development employees decision to stay as both variables were measured with interval scale and satisfied assumption of normality (Manning & Munro, 2007). In addition, the moderate value between those retention factors achieved p < .05 (Manning & Munro, 2007).

Other statistical techniques such as t-test and ANOVA were further used to answer demographic questions. T-tests were used to answer the difference between age groups (referring especially to the two age groups of 20 to 29 years and 30 to 44 years as they are the majority of respondents). The justification for this is the suggested retention factors linked to age group, which is measured with dichotomous nominal scale and interval scale and to satisfy assumption of normality (Manning & Munro, 2007). This is because the research was interested post hoc in identifying the retention factors of software development employees in relation to their career stage of exploration and establishment.

ANOVA was used to explore post hoc the difference between three groups of software development employees. The justification for this is because the research was interested in identifying the differences in retention factors between different types of software development employees.

3.6 Steps in the Data Collection Process

The steps in the data collection process for this study involved the pilot study and main study. Both studies were sequentially important to the consequence of data interpretation with validity and reliability. Gorard (2003) has stressed that comments made in the pilot study could apply equally well to the main study. That is because the pilot study is a pre-tested study which involves a much smaller sample, testing out all aspects of the
108

survey. Therefore, a pilot study was set up as step 1 and the main study was set up as step 2 as follows.

3.6.1 Step 1: Pilot Study

The purpose of a pilot study is to provide information to fine-tune the survey process in the research design. Ticehurst and Veal (2000, p.151) have stressed that this is a small scale trial run of a large survey and it is always advisable to carry out a pilot study for the following purposes (see Table 3.5).

Table 3.5: The purposes of pilot survey Pilot Survey Purposes: 1. Testing questionnaire wording 2. Testing question sequencing 3. Testing question layout 4. Gaining familiarity with respondents 5. Testing fieldwork arrangements (if required) 6. Training and testing fieldworkers (if required) 7. Estimating response rate 8. Estimating interview or questionnaire completion time 9. Testing analysis procedures Applicable: Yes Yes Yes Yes Yes No Yes Yes No

Source: developed for this research based on Ticehurst and Veal (2000, p.151)

In the pilot survey, a two-stage pre-testing process was established into the field work. Firstly, the questionnaire wording, layout, sequencing, and completion time were tested out by the researchers network of 5 ex-colleagues (software development employees) and the responses to the questionnaire survey were positive. Gorard (2003) suggests that the questionnaire should always be tried out on experts, friends, and anyone else who can help with it. It took them about 10 to 15 minutes to complete, which was acceptable. Secondly, the researcher moved on to a large-scale pilot survey to identify any changes needed

109

before proceeding to the main study. This helps the researcher to face up to flaws and react accordingly at an early stage (Gorard, 2003).

In a large-scale pilot survey, upwards of 50 respondents are suggested as a sufficient number of respondents for analysis (Brace, 2004). In this study, 60 software development employees, retrieved from a total of 150 participants in the Human Resource Banks email system, effectively completed the survey. The response rate in the large-scale pilot survey was 40%. These first 150 participants (emails) were not used again in the main study as the Human Resource Bank was able to filter them out from their database. Furthermore, the sampling through the email system was randomly selected and distributed so the respondents completed the web-based survey voluntarily and anonymously.

3.6.2 Step 2: Main Study

The main study is the next step to fulfil the main purpose of the research which is to gather a certain large number of samples and then answer the research questions by using even more statistical techniques. In order to achieve a minimum 40% response rate similar to the pilot survey, 1000 participants were randomly selected for the main study. The web-based main survey was run for one month and a total of 415 respondents or 41.5% effectively completed the surveys.

The main survey was sequential to the pilot survey, and was also conducted through the Human Resource Bank. The Human Resource Bank emailed the survey message to each individual of the selected respondents by mapping out the requirements of the respondents from their database. The respondents then went to the web server to complete the survey by double-clicking the hyperlink provided in the content of the email. As responses were voluntary and anonymous, there was no obligation to participate.
110

Data analysis with validity and reliability were two main scales of measurement in the research design. In addition, principle component analysis (PCA), which is a type of factor analysis (FA), was used to subtract factors that have similar high loadings. PCA was used because the researcher has confidence, driven by theory, in the underlying dimensions of the latent construct. If this confidence does not exist then change to principal axis factoring (PAF) throughout as it is thought to be more effective at finding underlying dimensions to latent constructs (Netemeyer, Bearden, & Sharma, 2003). Also, comment on this as exploratory factor analysis used to assess the dimensionality of the scale and refine items. Figure 3.3 provides standard values for the scale measurement.

Figure 3.3: Criteria of scale measurement for data analysis Scale Measurement Less than 0.60 No significant (P>0.05) Less than 0.50 Less than 0.50 Less than 0.30 Less than 0.70 Delete Items
Source: Developed for this research based on Zhang (2008).

KMO Bartletts test Factor Loadings Correlations (item-total) Correlations (inter-item) Cronbachs Alpha

More than 0.60 Significant (P<0.05) More than 0.50 More than 0.50 More than 0.30 More than 0.70 Retain Items

3.7 Ethical Considerations

Ethics is a code of conduct which guides researchers in behaving properly in the social world. Generally, the principles of ethics basically provide guidelines for social research including integrity, respect for persons, beneficence, and justice. For instance, a researcher has a responsibility to prevent a risk of harm or discomfort to participants in a research project. In addition, participants in the social research must give informed consent to being part of the research (Bouma, 2000). The identity of informants must be protected,

111

researchers must not coerce participants into divulging information, and researchers must keep data privately for five years (Bouma, 2000). These rules have been established in the Southern Cross University (SCU) and all the researchers must be granted permission from the Human Research Ethics Committee (HREC) of the University before conducting the survey.

This research project complied with the rules of the ethical principles required by the HREC of the SCU. The approval number is ECN-08-107 (see Appendix E). In addition, the survey of this study was an online, web-based design in which the participants were voluntary and anonymous in the process. For instance, the participants could stop completing the surveys at any time and were not to be identified. As the conduct of the survey was undertaken through the Human Resource Bank, the message of informed consent was given in the content of the email. Indeed, the participants consented to the participation automatically if they went to the web server and completed the surveys. Finally, the confidentiality of the survey results was ensured as the survey results were only used for the DBA thesis and kept privately by the researcher and his supervisor.

3.8 Limitations
A focus on HsinChu Science Park was the first limitation identified for this research as Taiwan has many industrial bases. This might restrict the results as findings may tend towards more regional aspects. The other limitation was potential retention factors that had been not identified for the examination. It is expected that there might be more retention factors that were significantly associated with software development employees decisions to stay, such as job responsibility. Thirdly, the nature of quantitative research limited exploration of the meanings behind the facts as the findings objectively presented the relationship between retention factors and software development employees decisions to
112

stay.

3.9 Conclusion
Quantitative methodology was chosen for this research based on the justification of the research question which examined the relationship between retention factors and software development employees decisions to stay in the IT industry in Taiwan. The data collection was designed with a web-based online survey, based on the advantages of speed, cost, anonymity, and ease of use and access because the respondents are computer literate. The sample of the target population was full-time software development employees working in the HsinChu Science Park, which is the oldest and biggest industrial base, and is highly representative of the IT industry as a whole in Taiwan. However, the data collection was conducted through the Human Resource Bank in Taiwan as there were difficulties in the field work which were detailed in the section on risk management.

A probability sampling technique with cluster sampling was used for this research. The sample size of around 400 was anticipated as it is in a good sample range of 300 to 500 for the data analysis. The questionnaire design was developed based on the hypotheses and conceptual framework. The questions of the identified variables were designed with a 5-point interval scale, along with nominal and ordinal scales of questions designed for personal information. The instruments were developed to rate respondents attitudes for interval data. Validity and reliability measurements were also addressed to assure the quality of the research and the results of the findings. The t-test, correlation, and multiple linear regressions (MLR) were used as the statistical methods for the data analysis.

In the pilot study, the response rate was 40% and some instruments of the variables were deleted due to the results of lower validity. In order to achieve a response rate similar to
113

the pilot survey, there were 415 respondents from a total of 1000 respondents who were invited to participate, giving a response rate of 41.5% for the main study. The ethical considerations were addressed to comply with the principles of voluntary participation, anonymity of respondents, confidentiality of data, and informed consent of respondents required by the HREC of the SCU.

This chapter is the implemented plan for the conduct of the survey in the field, in accordance with the concept of the model that has been set up in Chapter 2. Indeed, the results of this chapter will be the base for the following Chapter 4, which is data analysis and results.

114

Chapter 4: Data Analysis and Results

4.1 Introduction
This chapter provides the results from the survey detailed in Chapter 3. This survey was based on the literature on retention analysed in Chapter 2. The research question was probed through the survey using quantitative methods.

Chapter 4 consists of five sections (see Figure 4.1), and begins by reviewing the research focus and survey response. Section Two provides a process of data analysis outlining four stages of analysis procedure. Section Three provides data evaluation and testing including demographics, descriptive data analysis, Principal Components Analysis (PCA), normality test, standard Multiple Linear Regression (MLR), t-tests, and ANOVA test. Section Four is a summary of the analysed results as they relate to the research hypotheses and the research question. Finally, a conclusion is provided in Section Five. Figure 4.1 outlines the structure of this chapter.

Figure 4.1: Structure of Chapter 4. 4.1 Introduction 4.2 The Process of Data Analysis 4.3 Data Evaluation and Testing 4.4 Summary of Data Analysis 4.5 Conclusion
Source: Developed for this study.

115

4.1.1 Overview of Research Focus and Survey Response

The focus of this research identified the relationship between retention factors and software development employees in the IT industry in Taiwan. The participants were full-time software development employees working in Research and Development (R&D) departments of companies located in HsinChu Science Park. The questionnaire survey was an online, web-based design, which is regarded as the most appropriate approach for this group as IT workers are highly computer literate. The data was analysed using the SPSS (Statistical Program for Social Sciences) package and the MLR (Multiple Linear Regression) was used to test hypotheses underpinning the model.

In order to achieve a minimum 40% response rate similar to the pilot survey, 1000 participants were randomly selected for the main survey. The online main survey was run for one month and a total of 415 respondents or 41.5% effectively completed their surveys. The results were allocated into three categories of employees, mainly software engineers (313 respondents), project leaders (72 respondents), and assistant managers (30 respondents). As the sample size of the three groups of respondents was unequal and some were relatively small, the results about findings have been interpreted in more tentative terms throughout the entire thesis. This is because small sample size may have an impact on the statistical significance of the correlation coefficient (De Vaus, 2004). Furthermore, no missing data was found in any of those responses so this set of data was suitable for data analysis.

4.2 Data Analysis Procedure

There are several stages of data analysis in quantitative research (Manning & Munro, 2007). Firstly, it begins by identifying the scale of measurement for each item and variable, followed by entering data into SPSS, checking for missing values and data entry mistakes,
116

creating composite variables and evaluating their validity and reliability, conducting descriptive statistical tests for demographic items and variables, testing normality, and finally an appropriate statistical test to evaluate a research question or to test an hypothesis needs to be chosen.

As a result, the nominal scale items were identified with gender and position. The ordinal scale items were identified with age, salary, academic qualification, working period in the current company, and working period in IT the industry. The interval scale items were identified with retention factors across financial compensation, career development, self-achievement, environmental factors, job satisfaction, and decision to stay (see Appendix F). When entering the data, the retention factors were renamed for easy identification and data analysis (Manning & Munro, 2007). For instance, pay1 to pay5 was made up for financial compensation factor. All of these items are detailed in Appendix F. There was no missing value and data entry was all correct. The reversed questions were reversely coded before conducting any inferential statistical test.

The further data analysis was detailed in the next section of data evaluation and testing. The analysis procedure in Figure 4.2 below is provided for this research.

117

Figure 4.2 Data analysis procedure Stage one: Identify the scale of measurement Stage two: Enter data Stage three: Check for missing values Stage four: Create composite variables and evaluate validity and reliability (Data evaluation and testing includes demographics, descriptive data analysis, Principal Components Analysis, normality test, Multiple Linear Regression, t-tests and ANOVA)
Source: Developed for this research.

4.3 Data Evaluation and Testing

This section presents an important part of data analysis for the main study including respondents and their demographics, descriptive data analysis, Principal Components Analysis (PCA), normality test, Multiple Linear Regression (MLR), t-tests and ANOVA tests.

4.3.1 Respondents and Their Demographics

There were seven survey questions which were related to demographics including gender, age, position, salary, academic qualification, working period in the current company, and working period in the IT industry. The responses in Table 4.1 clearly show the detailed statistics of each of the demographics.

118

Table 4.1: Survey respondents by demographics (n=415)

Variable Gender Category Male Female Age 20 to 24 years 25 to 29 years 30 to 34 years 35 to 39 years 40 to 44 years 45 to 49 years 50 to 54 years Position Software engineers Project leaders Assistant managers Salary Under NT$500,000 per annum From NT$500,001 to NT$1,000,000 per annum From NT$1,000,001 to NT$1,500,000 per annum From NT$1,500,001 to NT$2,000,000 per annum Over NT$2,000,000 per annum Academic qualification Below senior high school 2-years or 5-years college Undergraduate Postgraduate (Master) Postgraduate (Doctor) Working period in current company Less than 1 year From 1 to 2 years From 3 to 5 years From 6 to 10 years From 11 to 15 years Working period in IT industry From 1 to 2 years From 3 to 5 years From 6 to 10 years From 11 to 15 years More than 16 years Source: developed for this study. Frequency 343 72 7 122 211 55 15 3 2 313 72 30 107 250 45 67 6 2 28 225 152 8 2 137 175 87 14 35 190 155 28 7 Percent 82.7 17.3 1.7 29.4 50.8 13.3 3.6 0.7 0.5 75.4 17.3 7.2 25.8 60.2 10.8 1.7 1.5 0.5 6.7 54.2 36.6 1.9 0.5 33.0 42.2 21.0 3.3 8.4 45.8 37.3 6.7 1.7

119

The results in Table 4.1 indicated that most respondents were male which had 82.7% against female with 17.3%.

The results in age indicated that the major respondents were located in a range from 25 to 39 years. The age group of 30 to 34 years had the highest response rate (50.8%), followed by the age group of 25 to 29 years (29.4%), and the age group of 35 to 39 years (13.3%).

For the workers job position, most respondents were from the software engineers group (75.4%), followed by the project leaders group (17.3%), and the assistant managers group (7.2%).

In the salary section, the highest yearly earning was located in the group ranging from NT$500,001 to NT$1,000,000 per annum (60.2%). The second largest group was under NT$500,000 (25.8%), and the third largest group was from NT$1,000,001 to NT$1,500,000 per annum (10.8%).

In academic qualification, graduating at an undergraduate level was dominant which had the highest response rate (54.2%), followed by a master level (36.6%), and 2-years or 5-years college level (6.7%).

Working period in the current company was largely dominated by the group of employees who have been working from 3 to 5 years (42.2%), followed by the group of 1 to 2 years (33%), and the group of 6 to 10 years (21.0%).

Working period in the IT industry was largely dominated by the group of employees who have been working from 3 to 5 years (45.8%), followed by the group of 6 to 10 years
120

(37.3%), and the group of 1 to 2 years (8.4%).

4.3.2 Descriptive Data Analysis

This section provides a summary of the descriptive data analysis of the respondents. It includes gender in comparison to age, salary, academic qualification, working period in current company, and working period in the IT industry.

According to Table 4.2, the responses of male in the IT industry mainly came from the age group of 30 to 34 years (53.4%), followed by the age group of 25 to 29 years (26.5%), and the age group of 35 to 39 years (14.3%). The responses of female, on the other hand, mainly came from the age group of 25 to 29 years (43.1%), followed by the age group of 30 to 34 years (38.9%), and the age group of 35 to 39 years (8.3%). This indicates that the workforce in the IT industry in Taiwan is dominated by Generation X (aged between 33 to 44 years old) and Millennial (Y and Net) (aged below 32 years old).

Table 4.2: Summary of Gender in comparison to Age Age Gender 20 to 24 25 to 29 30 to 34 35 to 39 40 to 44 45 to 49 50 to 54 Total years years years years years years years Response Rate Male (n=343) Female (n=72) 0.6% (2) 6.9% (5) 26.5% (91) 43.1% (31) 53.4% (183) 38.9% (28) 14.3% (49) 8.3% (6) 3.8% (13) 2.8% (2) 0.9% (3) 0.0% (0) 0.6% (2) 0.0% (0) 100% 100%

Source: Developed for this study.

According to Table 4.3, the responses of male in the IT industry indicate that they mainly earned a salary of between NT$500,001 to NT$1,000,000 per annum (63%), followed by a range of under NT$500,000 per annum (22.4%). The responses of female, on the other hand, indicate that they mainly earned a salary of between NT$500,001 to NT$1,000,000
121

per annum (47.2%), followed by a range of under NT$500,000 per annum (41.7%) and. In addition, there is a trend in the responses from male that they earned more than female in the upper ranges of salary. This indicates that the male receives a higher salary per annum than does the female in the IT industry in Taiwan.

Table 4.3: Summary of Gender in comparison to Salary Salary/ per annum Gender Male (n=343) Female (n=72) Under > NT$1M NT 500K 22.4% (77) 41.7% (30) 63.0% (216) 47.2% (34) >NT$1.5M >NT$2M 11.1% (38) 9.7% (7) 1.7% (6) 1.4% (1) Over NT$2M 1.8% (6) 0.0% (0) Total Response Rate 100% 100%

Source: Developed for this study.

According to Table 4.4, the responses of male indicate that they mainly graduated as undergraduates (50.7%), followed by postgraduates (Master) (40.5%). The responses of female, on the other hand, indicate that they also mainly graduated as undergraduates (70.8%) and postgraduates (Master) (18.1%). This indicates that the workforce in the IT industry in Taiwan is highly educated.

Table 4.4: Summary of Gender in comparison to Academic Qualification Academic Qualification Gender Male (n=343) Female (n=72) Below 2 or 5-year Undersenior high college graduate 0.3% (1) 1.4% (1) 6.7% (23) 6.9% (5) 50.7% (174) 70.8% (51) Postgraduate Postgraduate Total (Master) (Doctor) Response Rate 40.5% (139) 18.1% (13) 1.7% (6) 2.8% (2) 100% 100%

Source: Developed for this study.

122

According to Table 4.5, the responses of male indicate that they have been mainly working within the current company for a period of 3 to 5 years (42%), followed by a period of 1 to 2 years (34.1%). The responses of female, on the other hand, indicate that they have also been mainly working within the current company for a period of 3 to 5 years (43.1%) and a period of 1 to 2 years (27.8%). In addition, there is a trend that the percentage of working period within the current company is decreasing for the upper ranges of working period. This indicates that majority of working period within the current company in the IT industry in Taiwan is between 3 to 5 years and begins decreasing.

Table 4.5: Summary of Gender in comparison to Working Period in Current Company Working Period in Current Company Gender Male (n=343) Female (n=72) Less 1 year 0.3% (1) 1.4% (1) 1 to 2 years 3 to 5 years 34.1% (117) 27.8% (20) 42.0% (144) 43.1% (31) 6 to 10 years 20.4% (70) 23.6% (17) 11 to 15 years 3.2% (11) 4.2% (3) Total Response Rate 100% 100%

Source: Developed for this study.

According to Table 4.6, the responses of male indicate that they have been mainly working within the IT industry for a period of 3 to 5 years (45.5%), followed by a period of 6 to 10 years (38.2%). The responses of female, on the other hand, indicate that they have also been mainly working within the IT industry for a period of 3 to 5 years (47.2%) and a period of 6 to 10 years (33.3%). In addition, there is a trend that the percentage of working period within the IT industry is decreasing for the upper ranges of working period. This indicates that majority of working period within the IT industry in Taiwan is between 3 to 5 years and begins decreasing.

123

Table 4.6: Summary of Gender in comparison to Working Period in the IT Industry Working Period in IT Industry Gender Male (n=343) Female (n=72) 1 to 2 years 7.6% (26) 12.5% (9) 3 to 5 years 45.5% (156) 47.2% (34) 6 to 10 years 38.2% (131) 33.3% (24) 11 to 15 years 7.0% (24) 5.6% (4) More than 16 Total years Response Rate 1.7% (6) 1.4% (1) 100% 100%

Source: Developed for this study.

4.3.3 Principal Components Analysis (PCA)

Principal components analysis (PCA) is a type of factor analysis which can be used to summarise the relationships between variables and to define the underlying structure among them in the analysis. Hair, et al., (2010) stress that four objectives of factor analysis are to specifying the unit of analysis, achieving data summarisation and/ or data reduction, variable selection, and using factor analysis results with other multivariate techniques. Thus, a PCA was conducted using all original items in the questionnaire survey except for those constituting the dependent variable.

A PCA followed by a varimax rotation, was conducted on the fifty items of our questionnaire. Many bivariate correlations between items were in excess of .30 and so the correlation matrix was considered to be potentially appropriate for PCA. The Kaiser-Meyer-Olkin Measure of Sampling Adequate (KMO = .952) was greater than .6, and Bartletts Test of Sphericity was significant (X = 14187.224, p<.05), so by each of those criteria it was deemed to be appropriate to apply PCA to this data set (see Appendix G). Communalities ranged in magnitude from .447 (TD5) to .776 (PAY4). Manning and Munro (2007, p. 172) stress that The communality for a variable is the variance in that variable as predicted from the components. The larger the communality for a variable, the

124

better that variable is explained by the small number of component extracted. Variables with small communalities (say below .30) are not explained well be the underlying components, and if many variables are like this the solution will be poor. In this case, there is no communality below .40 (see Appendix H), so a good solution is likely.

Six components were extracted according to the results of the scree test (see Appendix I). Hair, et al. (2010, p. 110) state that The scree test is used to identify the optimum number of factors that can be extracted before the amount of unique variance begins to dominate the common variance structure. Six components were extracted accounting for almost 61% of the total variance (see Appendix J). As already noted, the scree test also suggested a six components solution and this would be judged an appropriate data set.

Component 1 accounted for 38.12% of the variance. The item with the largest loading on component 1 was REC3 My boss congratulates me for doing a good job. Other items with large loadings on this component were PRO3 My boss takes a supervisor role in my career development and REC4 I am adequately recognised for my contributions. Other details of loading scores have been included in Appendix K. Manning and Munro (2007) stress that the largest loading on a component commonly indicates what the component might represent. Therefore, given the pattern of the results it was decided to label this component as Job Appreciation for the purpose of this study.

Component 2 accounted for 7.0% of the variance. The item with the largest loading on component 2 was PAY1 This organisation pays me fairly for the work I do. Other items with large loadings on this component were PAY4 People at this organisation are paid fairly according to their job performance and PAY2 My total pay is adequate compared to others in this organisation. Other details of loading scores have been included in
125

Appendix K. The pattern of these results on this component might obviously represent fair remuneration according to employees work and performance. Therefore, it was decided to label this component as Fair Remuneration for the purpose of this study.

Component 3 accounted for 5.59% of the variance. The item with the largest loading on component 3 was AUTO2 I have authority to make the decisions necessary to do my job. Other items with large loadings on this component were IC2 I am given an opportunity to present and try new ideas and AUTO4 I am satisfied with the level of autonomy associated with my positions. Other details of loading scores have been included in Appendix K. The pattern of these results on this component might represent something like freedom in decision making on the job. Therefore, it was decided to label this component as Freedom in Decision Making for the purpose of this study.

Component 4 accounted for 4.61% of the variance. The item with the largest loading on component 4 was CHA1 My job is challenging. Other items with large loadings on this component were CHA4 My boss provides me with challenging work and CHA2 My job offers me the opportunities to gain work experience in challenge new areas. Other details of loading scores have been included in Appendix K. The pattern of these results on this component might obviously represent the work content that should be involved challenging and interesting. Therefore, it was decided to label this component as New Challenging Work for the purpose of this study.

Component 5 accounted for 2.96% of the variance. The item with the largest loading on component 5 was TD1 My confidence in using the knowledge and skills learned has decreased (reverse). Other items with large loadings on this component were IC5 This organisation does not recognise those who come up with new ideas (reverse) and JS5 I
126

have lost interest in my job (reverse). Other details of loading scores have been included in Appendix K. The pattern of these results on this component might represent the importance of personal development in relation to job competence with knowledge, skills and ability. Therefore, it was decided to label this component as Development for the purpose of this study.

Component 6 accounted for 2.32 of the variance. The item with the largest loading on component 6 was WB3 I have sufficient time to spend with my family. Other items with large loadings on this component were WB1 I am provided with the flexibility needed to balance the demands of my work and personal life and WB2 I have the flexibility to arrange my work so that I can meet my business objectives and balance my family and personal needs. Other details of loading scores have been included in Appendix K. The pattern of these results on this component might represent the significance of flexible work. Therefore, it was decided to label this component as Workplace Flexibility for the purpose of this study.

On the basis of the PCA, six new composite variables were generated using the component items. The first composite variable was Job Appreciation, which was calculated by taking the arithmetic mean of the nine items (REC3, PRO3, REC4, TD4, REC2, IC3, LEADER3, PRO4 and AUTO1). This composite variable was renamed RecPro_9 for easy identification. Item-to-total correlations and inter-item correlations were calculated and all were significant, p < .05. Using the criteria presented by Hair, et al. (1998, p.118), nine items were all found to display item-to-total correlations greater than the criteria of .50. Nine items were also found to display inter-item correlations greater than the criteria of .30. The coefficient (Cronbachs) alpha for the nine items was found to have an excellent level of reliability ( = .921) and was therefore used in the subsequent analyses. Details of
127

correlations and Cronbachs alpha have been included in Appendix L.

The second composite variable was Fair Remuneration, which was calculated by taking the arithmetic mean of the six items (PAY1, PAY4, PAY2, PAY3, PRO1 and PRO2). This composite variable was renamed Rem_6 for easy identification. Item-to-total correlations and inter-item correlations were calculated and all were significant, p < .05. Using the criteria presented by Hair, et al. (1998, p. 118), six items were all found to display item-to-total correlations greater than the criteria of .50. Six items were also found to display inter-item correlations greater than the criteria of .30. The coefficient (Cronbachs) alpha for the six items was found to have a good level of reliability ( = .897) and was

therefore used in the subsequent analyses. Details of correlations and Cronbachs alpha have been included in Appendix L.

The third composite variable was Freedom in Decision Making, which was calculated by taking the arithmetic mean of the nine items (AUTO2, IC2, AUTO4, AUTO3, LEADER4, LEADER2, LEADER1, JS4 and IC1). This composite variable was renamed AutoLeader_9 for easy identification. Item-to-total correlations and inter-item correlations were calculated and all were significant, p < .05. Using the criteria presented by Hair, et al. (1998, p. 118), nine items were all found to display item-to-total correlations greater than the criteria of .50. Nine items were also found to display inter-item correlations greater than the criteria of .30. The coefficient (Cronbachs) alpha for the nine items was found to have an excellent level of reliability ( = .901) and was therefore used in the subsequent

analyses. Details of correlations and Cronbachs alpha have been included in Appendix L.

The fourth composite variable was New Challenging Work, which was calculated by taking the arithmetic mean of the nine items (CHA1, CHA4, CHA2, TD3, REC5, JS2, JS3,
128

TD2 and TD5). This composite variable was renamed ChaJs_9 for easy identification. Item-to-total correlations and inter-item correlations were calculated and all were significant, p < .05. Using the criteria presented by Hair, et al. (1998, p. 118), nine items were all found to display item-to-total correlations greater than the criteria of .50. However, two inter-item correlations (r = .25 for CHA1 and r = .28 for CHA4) for TD5 were found to display lower than the criteria of .30. Although these two inter-item correlations were lower than the criteria of .30 they were significant and so they were retained for subsequent analyses. The coefficient (Cronbachs) alpha for the nine items was found to have a good level of reliability ( = .869) and was therefore used in the subsequent

analyses. Details of correlations and Cronbachs alpha have been included in Appendix L.

The fifth composite variable was Development, which was calculated by taking the arithmetic mean of the nine items (TD1, IC5, JS5, PRO5, AUTO5, WB5, LEADER5, REC1 and PAY5). This composite variable was renamed Dev_9 for easy identification. Item-to-total correlations and inter-item correlations were calculated and all were significant, p < .05. Using the criteria presented by Hair, et al. (1998, p. 118), nine items were all found to display item-to-total correlations greater than the criteria of .50. However, three inter-item correlations (r = .24 for IC5, r = .18 for LEADER5 and r = .15 for PAY5) for REC1 were found to display lower than the criteria of .30. One inter-item correlation (r = .28 for JS5) for PAY5 was found to display lower than the criteria of .30. Although these four inter-item correlations were lower than the criteria of .30 they were significant and so they were retained for subsequent analyses. The coefficient (Cronbachs) alpha for the nine items was found to have a good level of reliability ( = .857) and was therefore used in

the subsequent analyses. Details of correlations and Cronbachs alpha have been included in Appendix L.

129

The sixth composite variable was Workplace Flexibility, which was calculated by taking the arithmetic mean of the four items (WB3, WB1, WB2 and WB4). This composite variable was renamed Wlb_4 for easy identification. Item-to-total correlations and inter-item correlations were calculated and all were significant, p < .05. Using the criteria presented by Hair, et al. (1998, p. 118), four items were all found to display item-to-total correlations greater than the criteria of .50. Four items were also found to display inter-item correlations greater than the criteria of .30. The coefficient (Cronbachs) alpha for the four items was found to have a good level of reliability ( = .871) and was

therefore used in the subsequent analyses. Details of correlations and Cronbachs alpha have been included in Appendix L.

The last composite variable was dependent variable, which was decision to stay, was calculated by taking the arithmetic mean of the five items (RET1, RET2, RET3, RET4 and RET5). This composite variable was renamed Ret_5 for easy identification. Item-to-total correlations and inter-item correlations were calculated and all were significant, p < .05. Using the criteria presented by Hair, et al. (1998, p. 118), five items were all found to display item-to-total correlations greater than the criteria of .50. Five items were also found to display inter-item correlations greater than the criteria of .30. The coefficient (Cronbachs) alpha for the five items was found to have a good level of reliability ( = .844) and was therefore used in the subsequent analyses. Details of correlations and Cronbachs alpha have been included in Appendix L.

From the results of the PCA, a six new retention factors solution was empirically identified as independent variables and their relationships with the dependent variable which was the software development employees decision to stay, was examined. Thus, six hypotheses used as a set of empirical retention factors have been developed to answer the research
130

question as follows:

Empirical Hypothesis 1: Job appreciation is positively related to the software development employees decision to stay in the IT industry. Empirical Hypothesis 2: Fair remuneration is positively related to the software development employees decision to stay in the IT industry. Empirical Hypothesis 3: Freedom in decision making is positively related to the software development employees decision to stay in the IT industry. Empirical Hypothesis 4: New challenging work is positively related to the software development employees decision to stay in the IT industry. Empirical Hypothesis 5: Development is positively related to the software development employees decision to stay in the IT industry. Empirical Hypothesis 6: Workplace flexibility is positively related to the software development employees decision to stay in the IT industry.

4.3.4 Multiple Linear Regressions (MLR)

A multiple linear regression (MLR) is chosen to answer the research question as it allows one to assess the relationship between one dependent variable and several independent variables using the equation of a straight line (Tabachnick & Fidell, 2001). In other words, MLR is an extension of bivariate correlation using the equation of that straight line (Manning & Munro, 2007). Furthermore, each independent variable is weighted by the regression analysis and provides relative contribution to the overall prediction and facilitates interpretation (Hair, et al., 2010). Thus, a set of six empirical retention factors was examined to explain how effectively their relationships associated with the software development employees decision to stay within an organisation.
131

Standard MLR was used in this study. The Standard MLR allows testing of all independent variables simultaneously with one dependent variable and provides estimates of the unique contribution for prediction of each of the independent variables. This approach is different from other MLR techniques such as Sequential MLR, which enters independent variables in an order specified by the researcher and reports the change in the strength of their relationships with dependent variables at each step of the analysis (Manning & Munro, 2007).

Standard MLR also requires assumption testing. Tabachnick and Fidell (2001) have stressed that all variables are required to assess normality, linearity and homoscedasticity of residuals prior to conducting MLR. In addition, all variables must be measured at an interval or ratio scale; they must have a significant correlation result between the independent variables and the dependent variable; and the multicollinearity value must be smaller than .90 (Manning & Munro, 2007). Multicollinearity refers to the correlation among independent variables. A value above .90 indicates that they are the same. Tabachnick and Fidell (2001) stress that when multicollinearity is present, regression coefficients may not be significant because of the large size of standard errors.

The multiple correlation coefficient (R) will be examined which represents the combined correlation of all independent variables with the dependent variable. In addition, the variance between them will be explained by an adjusted R Square. Furthermore, ANOVA must reach a level of 95% confidence to allow further analysis of unique contribution to the regression equation to proceed with the set of all independent variables. Consequently, the relationships between the set of retention factors and software development employees decision to stay would be explained by the results of regression coefficients.
132

Normality and Transformation Normality test resulted in five independent variables and a dependent variable identified as negative skew as their z-score was greater than criteria of 3.29 for sample size larger than 300 (Manning & Munro, 2007). They were job appreciation (z-score is 6.725), fair remuneration (z-score is 3.292), freedom in decision making (z-score is 6.05), new challenging work (z-score is 4.383), workplace flexibility (z-score is 5.817), and decision to stay (z-score is 3.992). Development was another independent variable, which satisfied normality assumption with z-score of .075. In addition, kurtosis was satisfied with all variables. To test whether the distributions skew significantly deviated from that of a normal distribution, the value for skew was divided by the standard error of the skew (Manning & Munro, 2007). A similar procedure was conducted for kurtosis where the value for kurtosis was divided by the standard error of the kurtosis (Manning & Munro, 2007). Details of these descriptive statistics have been included in Appendix M.

Transformations for six negative skew variables were justified and satisfied with normal distribution as z-score for job appreciation is 2.84; for fair remuneration is .158; for freedom in decision making is 2.758; for new challenging work is .908; for workplace flexibility is 1.742; and for decision to stay is .042 (see normal distribution in Appendix N). These transformed variables were used in all subsequent analysis and renamed as RecPro_sq was for job appreciation; Rem_sq was for fair remuneration; ChaJs_sq was for new challenging work; AutoLeader_sq was for freedom in decision making; Wlb_sq was for workplace flexibility, and Ret_sq was for decision to stay.

Linearity and Homoscedasticity of Residuals According to Tabachnick and Fidell (2001), residuals scatterplot can examine normality, linearity and homoscedasticity between predicted dependent variable scores and errors of
133

prediction. In addition, moderate to extreme deviations from linearity and variance of the residuals may lead to a serious underestimation of a relationship (Coakes, Steed, & Ong, 2009). Thus, a case with a large residual is an outlier, which needs to be treated carefully with a chi-square value of 22.458 (six independent variables) and p < .001 in this study (Tabachnick & Fidell, 2001). The chi-square value, with degrees of freedom (DF) equal to the number of independent variables is the examination of the Mahalanobis distance for multivariate outliers.

After the examination of residuals scatterplot with the set of six independent variables and one dependent variable, six cases (case 233, 255, 351, 352, 358 and 431) were found to be greater than the critical chi-square value of 22.458. Thus, these six cases were removed. The assumptions of normality, linearity and homoscedasticity were then satisfied for the MLR. The results of residuals statistics, linearity and scatterplot have been included in Appendix O.

Correlation and Multicollinearity Correlation is an indicator of association between two variables measured at interval or ratio scale. Pearson product-comment correlation (simply referred to Pearson r, correlation or represented by r) is commonly used to measure the strength of relationship and determines the direction (positive or negative) of relationship (Davis, 2005). The value of correlation possible ranges from -1 to 1. The value indicates the strength of the relationship. For instance, 0 is no relationship between two variables whereas 1 is perfect relationship between two variables (Coakes, Steed, & Dzidic, 2006). The sign (+ or -) of correlation indicates the direction (+ is positive or is negative) (Coakes, et al., 2006). When two-tail significance with p < .05, it concludes that there is a significant relationship between two variables at a level of 95% confidence.
134

Six independent variables were identified and one variable was a dependent variable. They were all measured at interval scales (see Appendix F). In addition, correlation results between all independent variables and one dependent variable displayed positive and significant (see Table 4.7) as well as correlations among these independent variables did not display multicollinearity (see Appendix P). Thus, MLR were able to proceed with these six retention factors as independent variables and software development employees decision to stay as dependent variable. Furthermore, there were four groups of people for the analysis, including all aggregated employees, software engineers, project leaders, and assistant managers of research and development department. These consequent results were applied for the test of standard Multiple Linear Regression.

Table 4.7: Correlation results for software development employees Hypotheses All aggregated employees (n=409) Sig. (r = .69) Sig. (r = .66) Sig. (r = .64) Sig. (r = .63) Sig. (r = .42) Sig. (r = .52) Software Engineers (n=311) Sig. (r = .69) Sig. (r = .68) Sig.( r = .65) Sig. (r = .64) Sig. (r = .40) Sig. (r = .50) Project Leaders Assistant (n=69) Managers (n=29) Sig. (r = .69) Sig. (r = .61) Sig. (r = .52) Sig. (r = .51) Sig. (r = .42) Sig. (r = .54) Sig. (r = .79) Sig. (r = .59) Sig. (r = .86) Sig. (r = .84) Sig. (r = .78) Sig. (r = .72)

H1: Job appreciation H2: Fair remuneration H3: Freedom in decision making H4: New challenging work H5: Development H6: Workplace flexibility

Source: Developed for this study.

Standard MLR for All Aggregated Employees A standard multiple linear regression was performed between decision to stay (Ret_sq) as the dependent variable and job appreciation (RecPro_sq), fair remuneration (Rem_sq), freedom in decision making (AutoLeader_sq), new challenging work (ChaJs_sq),
135

development (Dev_9), and workplace flexibility (Wlb_sq) as independent variables for all aggregated employees. The results in Table 4.8 indicated that the correlation of the six combined independent variables with one dependent variable was .75. The adjusted R Square was .55, which meant that 55% of variance in decision to stay could be explained by a set of six independent variables for all aggregated employees. The ANOVA test showed the result was significant with F = 85.62, p < .05, as a set of six independent variables was significantly correlated with decision to stay.

Table 4.8: Summary of R and ANOVA for all aggregated employees Model Summary R .75 R Square .56 Adjusted R Square .55 ANOVA F value 85.62 Sig. Sig. < .05

Source: Developed for this study.

Given the result of ANOVA was significant, a further analysis of unique contribution to the regression equation was able to proceed with the set of six independent variables. From the results of Table 4.9, four independent variables: job appreciation (t = 2.09, p < .05), fair remuneration (t = 6.45, p < .05), development (t = 3.02, p < .05), and workplace flexibility (t = 2.05, p < .05), were significant with p < .05. In addition, freedom in decision making (t = 1.92, p = .056) was marginally significant. These indicated that these five independent variables could be uniquely contributing to the regression equation with the decision to stay. However, new challenging work (t = 1.72, p > .05) could provide no unique contribution to the decision to stay. The equation of the relationship between the variables could be predicted as follows.

136

Decision to stay for all aggregated employees = .163 x job appreciation + .310 x fair remuneration + .134 x freedom in decision making + .113 x new challenging work + .047 x development + .081 x workplace flexibility + .115

Table 4.9: Regression coefficients for all aggregated employees

Unstandardized Coefficients Model 1 (Constant) RecPro_sq Rem_sq AutoLeader_sq ChaJs_sq Dev_9 Wlb_sq a. Dependent Variable: Ret_sq B .115 .163 .310 .134 .113 .047 .081 Std. Error .068 .078 .048 .070 .066 .016 .039 .159 .318 .127 .105 .112 .091 Standardized Coefficients Beta t 1.686 2.093 6.447 1.919 1.719 3.019 2.049 Sig. .093 .037 .000 .056 .086 .003 .041 .686 .662 .635 .634 .420 .519 .104 .306 .095 .085 .149 .102 .069 .213 .063 .057 .100 .068 Correlations Zero-order Partial Part

Standard MLR for Software Engineers A standard multiple linear regression was performed between decision to stay (Ret_sq) as the dependent variable and job appreciation (RecPro_sq), fair remuneration (Rem_sq), freedom in decision making (AutoLeader_sq), new challenging work (ChaJs_sq), development (Dev_9), and workplace flexibility (Wlb_sq) as independent variables for the software engineers. The results in Table 4.10 indicated that the correlation of the six combined independent variables with one dependent variable was .76. The adjusted R Square was .57, which meant that 57% of variance in decision to stay could be explained by a set of six independent variables for the software engineers. The ANOVA test showed the result was significant with F = 68.12, p < .05, as a set of six independent variables was significantly correlated with decision to stay.
137

Table 4.10: Summary of R and ANOVA for software engineers Model Summary R .76 R Square .57 Adjusted R Square .57 ANOVA F value 68.12 Sig. Sig. < .05

Source: Developed for this study.

Given the result of ANOVA was significant, a further analysis of unique contribution to the regression equation was able to proceed with the set of six independent variables. From the results of Table 4.11, two independent variables: fair remuneration (t = 6.31, p < .05) and development (t = 3.02, p < .05), were significant with p < .05. These indicated that these two independent variables could be uniquely contributing to the regression equation with the decision to stay. However, job appreciation (t = 1.30, p > .05), freedom in decision making (t = 1.87, p > .05), new challenging work (t = 1.61, p > .05), and workplace flexibility (t = 1.37, p > .05) could provide no unique contribution to the decision to stay. The equation of the relationship between the variables could be predicted as follows.

Decision to stay for the software engineers = .114 x job appreciation + .354 x fair remuneration + .157 x freedom in decision making + .120 x new challenging work + .053 x development + .064 x workplace flexibility + .103

138

Table 4.11: Regression coefficients for software engineers

Unstandardized Coefficients Model 1 (Constant) RecPro_sq Rem_sq AutoLeader_sq ChaJs_sq Dev_9 Wlb_sq a. Dependent Variable: Ret_sq B .103 .114 .354 .157 .120 .053 .064 Std. Error .078 .088 .056 .084 .075 .018 .047 .112 .360 .146 .112 .124 .071 Standardized Coefficients Beta t 1.316 1.299 6.313 1.865 1.605 3.019 1.374 Sig. .189 .195 .000 .063 .110 .003 .170 .686 .681 .650 .644 .397 .497 .074 .340 .106 .092 .171 .079 .049 .236 .070 .060 .113 .051 Correlations Zero-order Partial Part

Standard MLR for Project Leaders A standard multiple linear regression was performed between decision to stay (Ret_sq) as the dependent variable and job appreciation (RecPro_sq), fair remuneration (Rem_sq), freedom in decision making (AutoLeader_sq), new challenging work (ChaJs_sq), development (Dev_9), and workplace flexibility (Wlb_sq) as independent variables for the project leaders. The results in Table 4.12 indicated that the correlation of the six combined independent variables with one dependent variable was .72. The adjusted R Square was .47, which meant that 47% of variance in decision to stay could be explained by a set of six independent variables for the project leaders. The ANOVA test showed the result was significant with F = 11.17, p < .05, as a set of six independent variables was significantly correlated with decision to stay.

139

Table 4.12: Summary of R and ANOVA for project leaders Model Summary R .72 R Square .52 Adjusted R Square .47 ANOVA F value 11.17 Sig. Sig. < .05

Source: Developed for this study.

Given the result of ANOVA was significant, a further analysis of unique contribution to the regression equation was able to proceed with the set of six independent variables. From the results of Table 4.13, only one independent variable, job appreciation (t = 2.52, p < .05) could be uniquely contributing to the regression equation with the decision to stay. However, fair remuneration (t = 1.75, p > .05), freedom in decision making (t = .14, p > .05), new challenging work (t = -.65, p > .05), development (t = .94, p > .05), and workplace flexibility (t = .65, p > .05) could provide no unique contribution to the decision to stay. The equation of the relationship between the variables could be predicted as follows.

Decision to stay for the project leaders = .477 x job appreciation + .220 x fair remuneration + .020 x freedom in decision making - .110 x new challenging work + .037 x development + .062 x workplace flexibility + .313

140

Table 4.13: Regression coefficients for project leaders

Unstandardized Coefficients Model 1 (Constant) RecPro_sq Rem_sq AutoLeader_sq ChaJs_sq Dev_9 Wlb_sq a. Dependent Variable: Ret_sq B .313 .477 .220 .020 -.110 .037 .062 Std. Error .192 .190 .126 .144 .170 .039 .095 .471 .252 .020 -.096 .097 .082 Standardized Coefficients Beta t 1.630 2.516 1.749 .137 -.645 .940 .650 Sig. .108 .014 .085 .891 .521 .351 .518 .678 .613 .522 .506 .418 .539 .304 .217 .017 -.082 .119 .082 .222 .154 .012 -.057 .083 .057 Correlations Zero-order Partial Part

Standard MLR for Assistant Managers A standard multiple linear regression was performed between decision to stay (Ret_sq) as the dependent variable and job appreciation (RecPro_sq), fair remuneration (Rem_sq), freedom in decision making (AutoLeader_sq), new challenging work (ChaJs_sq), development (Dev_9), and workplace flexibility (Wlb_sq) as independent variables for the assistant managers. The results in Table 4.14 indicated that the correlation of the six combined independent variables with one dependent variable was .91. The adjusted R Square was .78, which meant that 78% of variance in decision to stay could be explained by a set of six independent variables for the assistant managers. The ANOVA test showed the result was significant with F = 17.48, p < .05, as a set of six independent variables was significantly correlated with decision to stay.

141

Table 4.14: Summary of R and ANOVA for assistant managers Model Summary R .91 R Square .83 Adjusted R Square .78 ANOVA F value 17.48 Sig. Sig. < .05

Source: Developed for this study.

Given the result of ANOVA was significant, a further analysis of unique contribution to the regression equation was able to proceed with the set of six independent variables. From the results of Table 4.15, two independent variables: fair remuneration (t = 2.17, p < .05) and freedom in decision making (t = 2.78, p < .05) were significant with p < .05. These indicated that these two independent variables could be uniquely contributing to the regression equation with the decision to stay. However, job appreciation (t = -1.68, p > .05), new challenging work (t = .85, p > .05), development (t = -.29, p > .05), and workplace flexibility (t = 1.85, p > .05) could provide no unique contribution to the decision to stay. The equation of the relationship between the variables could be predicted as follows.

Decision to stay for the assistant managers = -.509 x job appreciation + .290 x fair remuneration + .847 x freedom in decision making + .264 x new challenging work - .028 x development + .238 x workplace flexibility - .249

142

Table 4.15: Regression coefficients for assistant managers

Unstandardized Coefficients Model 1 (Constant) RecPro_sq Rem_sq AutoLeader_sq ChaJs_sq Dev_9 Wlb_sq a. Dependent Variable: Ret_sq B -.249 -.509 .290 .847 .264 -.028 .238 Std. Error .231 .303 .134 .305 .311 .095 .129 -.468 .269 .818 .246 -.068 .254 Standardized Coefficients Beta t -1.079 -1.683 2.172 2.782 .852 -.291 1.854 Sig. .292 .106 .041 .011 .404 .774 .077 .788 .592 .856 .836 .775 .722 -.338 .420 .510 .179 -.062 .368 -.149 .193 .247 .076 -.026 .165 Correlations Zero-order Partial Part

4.3.5 Post-Hoc Analysis

It was decided that a post-hoc analysis of age groups and three types of employment would be conducted as a means of addressing the scholarly evidence that characteristics of young workers (e.g. Generation Y) are different in terms of retention factors, in comparison to mid-career workers (e.g. Generation X). Additionally, people in different stages of their career have different career goals and priorities regarding their achievements. For instance, people at the career stage of Exploration may pursue technical competence while people at the career stage of Establishment may pursue managerial competence.

This section provides two subsections for statistical techniques, including t-tests and the ANOVA test. The adoption of t-tests was used to examine two age groups, which include those aged between 20 and 29 and those aged between 30 and 44 across seven composite variables. The ANOVA test was used to examine three different types of software development employees to see whether there was any significance among seven composite variables.
143

4.3.5.1 T- Tests
A t-test is used to determine whether there is a significant difference between two sets of scores for each of composite variables (Coaks, et al., 2006). This is evaluated by comparing mean scores of two groups (Gorard, 2003). There are two types of t-test (independent samples and dependent samples), and an independent samples t-test was chosen because this research aims to analyse two age groups, not one group over different time periods. Therefore, seven retention factors were examined including job appreciation, fair remuneration, freedom in decision making, new challenging work, development, workplace flexibility, and decision to stay.

Sample size is another influencing factor that determines whether a t-test should be applied in this research. Argyrous (1996) has stressed that t-scores are correct in both large and small sample cases whereas z-scores are correct only in a large sample. In addition, the sample size of 30 has much fatter tails in t-distribution than z-distribution, these tails become thinner in a sample size of 90, and it eventually becomes identical to a normal curve when a sample size is greater than 120 (Argyrous, 2000). In this study, two age groups are that those aged between 20 and 29 had a sample size of 127 and those aged between 30 and 44 had a sample size of 277.

A t-test also needs to meet some assumptions before further analysis, because the accuracy of test interpretation depends on whether assumptions have been violated. Coaks, et al. (2006) have stressed that a t-test must satisfy the scale of measurement (data should be at interval or ratio scale), random sampling (score should be randomly sampled), and normality (score should be normally distributed). Furthermore, an independent t-test has two additional assumptions which include independence of groups and homogeneity of variance. The independence of groups is that participants should appear in only one group
144

and two groups should be independent from each other (Coakes, et al., 2006). In homogeneity of variance, the Levene test is equality of variance, where if the result of this test is statistically not significant (p>.05), then equal variance is assumed, or vice versa. Then, if an independent samples t-test is statistically significant with p<.05, it concludes that there is a significant difference between two groups at a level of 95% confidence (Coakes, et al., 2006).

All seven composite variables met with the assumptions of the t-test as they were all measured at an interval scale (see Appendix F) and satisfied with normal curve (see Appendix N). In addition, two age groups were independent from each other and all of their scores were randomly sampled. Therefore, these seven composite variables were able to proceed with an independent samples t-test.

Based on the results in Table 4.16, the only one factor that evidenced significant difference between two age groups was development. For the remaining factors, such as job appreciation, fair remuneration, freedom in decision making, new challenging work, workplace flexibility, and decision to stay, there were no significant differences between the two age groups.

145

Table 4.16: Summary of the t-tests for two age groups comparison Composite variables Job appreciation Fair remuneration Freedom in decision making New challenging work Development Workplace flexibility Decision to stay Age between 20 to 29 Age between 30 to 44 Levenes t-test result result (n=127) (n=277) Mean 1.7 1.57 1.7 1.72 2.95 1.68 1.56 Std. dev. 0.26 0.28 0.25 0.24 0.61 0.29 0.27 Mean 1.68 1.58 1.7 1.72 3.09 1.64 1.59 Std. dev. 0.25 0.26 0.24 0.24 0.62 0.29 0.26 No Sig. No Sig. No Sig. No Sig. No Sig. No Sig. No Sig. No Sig.

No Sig. Sig. < .05 No Sig. No Sig. No Sig. No Sig.

Source: Developed for this study.

For the significant factor of development, when comparing two age groups (those aged between 20 and 29 and those aged between 30 and 44), the Levene test was not significant with p > .05, so the t-value, df, and two-tail significance for the equal variance assumed existed. The two-tail significance indicated that p < .05. Therefore, it was significant. This indicated that the two age groups came from different populations because of their significant difference, t(402) = -2.17, p < .05. Details of t-test results have been included in Appendix Q. In addition, when comparing mean scores of the two age groups (see Table 4.16), those aged group between 20 and 29 (2.95) had lower mean scores than those aged group between 30 and 44 (3.09). The responses from those aged group between 30 and 44 indicated more confidence in career development than did the responses from those aged group between 20 and 29.

When comparing two age groups (those aged between 20 and 29 and those aged between 30 and 44) for the factors of job appreciation, fair remuneration, freedom in decision making, new challenging work, workplace flexibility, and decision to stay, their Levene
146

test were not significant with p > .05, so their t-value, df, and two-tail significance for the equal variance assumed existed. The two-tail significance indicated that these factors were not significant with p > .05. This indicated that the two age groups could come from the same populations for those factors because they had no significant difference, whereby the job appreciation factor had t(402) = .46, p > .05; the fair remuneration factor had t(402) = -.47, p > .05; the freedom in decision making factor had t(402) = .03, p > .05; the new challenging work factor had t(402) = .18, p > .05; the workplace flexibility factor had t(402) = 1.17, p > .05; and the decision to stay factor had t(402) = -1.16, p > .05. Details of t-test results have been included in Appendix Q.

4.3.6 ANOVA Tests

The analysis of variance (ANOVA) test was used to examine the difference between three groups of employees as independent variables and their relationships with each of the dependent variables as seven composite variables. The three groups of employees were software engineers, project leaders, and assistant managers. The seven composite variables were job appreciation, fair remuneration, freedom in decision making, new challenging work, development, workplace flexibility, and decision to stay. Hair, et al., (2010) stress that unlike the t-test, which is designed to compare the mean scores of two groups, ANOVA is an extensive univariate technique for two or more groups. The reason that ANOVA was used is that it could establish whether there were any significant differences between the three groups in terms of the variables which have been demonstrated to contribute to a decision to stay in the workplace.

Assumptions of the ANOVA test must also be met with normality and homogeneity of variance before further analysis (Coakes, et al., 2009). In this study, all seven composite variables met with normal distribution (see Appendix N). In homogeneity of variance, like
147

the t-test, the Levene test determines whether variances are equal or unequal. Note that the composite variable is able to proceed with confidence when the result of the Levene test is statistically not significant with p > .05.

One-way between-groups ANOVA with post-hoc comparison were used in this study. Coakes et al., (2009) stress that the ANOVA test is based on the notion of variance which is derived from two different estimates of population variance from the data and then calculates a statistic from the ratio of these two estimates. In addition, the F statistic represents a ratio of these two estimates, which are between-groups variance and within-groups variance. A significant F-ratio indicates that two groups come from different populations, each with a different mean, but both with the same standard deviation (Manning & Munro, 2007). A post-hoc analysis is thus required to determine the pattern of difference among the three groups of employees.

Based on the results shown in Table 4.17, all composite variables were not significant among three groups of employees. When comparing the three groups (software engineers, project leaders, and assistant managers) for the factors of job appreciation, fair remuneration, freedom in decision making, new challenging work, development, workplace flexibility, and decision to stay, their Levene tests were not significant, p > .05, so their assumptions of homogeneity of variance were able to proceed with confidence. The F statistic indicated that all composite variables were not significant, p > .05. This indicated that the three groups were from the same population for these factors because their mean ratings of employees had no significant difference among them, whereby the job appreciation factor had F(2, 406) = .87, p > .05; the fair remuneration factor had F(2,406) = 1.28, p > .05; the freedom in decision making factor had F(2,406) = 2.00, p > .05; the new challenging work factor had F(2,406) = .36, p > .05; the development factor
148

had F(2,406) = 2.52, p > .05; the workplace flexibility factor had F(2,406) = .25, p > .05; and the decision to stay factor had F(2,406) = 1.77, p > .05. Details of the ANOVA results have been included in Appendix R.

Table 4.17: Summary of the ANOVA test for three groups Composite variables Software Engineers (n=311) Mean Job appreciation Fair remuneration Freedom in decision making New challenging work Development Workplace flexibility Decision to stay
Source: Developed for this study.

Project Leaders (n=69) Mean 1.7 1.56 1.74 1.74 3.13 1.64 1.53

Assistant Managers (n=29) Mean 1.74 1.65 1.74 1.74 3.24 1.69 1.59

Levenes result

ANOVA result

1.68 1.58 1.68 1.71 3.01 1.65 1.59

No Sig. No Sig. No Sig. No Sig. No Sig. No Sig. No Sig.

No Sig. No Sig. No Sig. No Sig. No Sig. No Sig. No Sig.

4.4 Summary of Data Analysis

This section provides a summary of the analysed results relating to the empirical research hypotheses and the research question.

4.4.1 Summary of the Empirical Research Hypotheses

There were six empirical retention factors identified to examine their relationship with software development employees decision to stay in the IT industry in Taiwan. These were job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility. In addition, four groups of people were undertaken for the analysis which included all aggregated employees, software engineers, project leaders, and assistant managers. Table 4.18 below is a summary of the results
149

relating to the six empirical research hypotheses for the four groups of software development employees.

Table 4.18: Conclusions of the results relating to the hypotheses for four groups Empirical Hypotheses
H1: Job appreciation is positively related to retention

All Aggregated Software Project Employees Engineers Leaders Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig. Sig.

Assistant Managers Sig. Sig. Sig. Sig. Sig. Sig.

H2: Fair remuneration is positively related to retention H3: Freedom in decision making is positively related to retention H4: New challenging work is positively related to retention H5: Development is positively related to retention H6: Workplace flexibility is positively related to retention
Source: Developed for this study.

Empirical Hypothesis 1: Job appreciation is positively related to retention According to Tables 4.7 and 4.18, a significant positive correlation could be found between job appreciation and retention of software development employees (all aggregated employees: r = .69, p < .05; software engineers: r = .69, p < .05; project leaders: r = .69, p < .05; and assistant managers: r = .79, p < .05). This indicated that giving a proper job appreciation to software development employees could keep them staying longer in the company.

Empirical Hypothesis 2: Fair remuneration is positively related to retention According to Tables 4.7 and 4.18, a significant positive correlation could be found between fair remuneration and retention of software development employees (all aggregated
150

employees: r = .66, p < .05; software engineers: r = .68, p < .05; project leaders: r = .61, p < .05; and assistant managers: r = .59, p < .05). This indicated that increasing fair payment could stimulate software development employees to stay longer in the company.

Empirical Hypothesis 3: Freedom in decision making is positively related to retention According to Tables 4.7 and 4.18, a significant positive correlation could be found between freedom in decision making and retention of software development employees (all aggregated employees: r = .64, p < .05; software engineers: r = .65, p < .05; project leaders: r = .52, p < .05; and assistant managers: r = .86, p < .05). This indicated that freedom in decision making could affect software development employees decisions to stay longer in the company.

Empirical Hypothesis 4: New challenging work is positively related to retention According to Tables 4.7 and 4.18, a significant positive correlation could be found between new challenging work and retention of software development employees (all aggregated employees: r = .63, p < .05; software engineers: r = .64, p < .05; project leaders: r = .51, p < .05; and assistant managers: r = .84, p < .05). This indicated that providing new challenging work could keep software development employees staying longer in the company.

Empirical Hypothesis 5: Development is positively related to retention According to Tables 4.7 and 4.18, a significant positive correlation could be found between development and retention of software development employees (all aggregated employees: r = .42, p < .05; software engineers: r = .40, p < .05; project leaders: r = .42, p < .05; and assistant managers: r = .78, p < .05). This indicated that providing opportunities for

151

personal development could stimulate software development employees to stay longer in the company.

Empirical Hypothesis 6: Workplace flexibility is positively related to retention According to Tables 4.7 and 4.18, a significant positive correlation could be found between workplace flexibility and retention of software development employees (all aggregated employees: r = .52, p < .05; software engineers: r = .50, p < .05; project leaders: r = .54, p < .05; and assistant managers: r = .72, p < .05). This indicated that workplace flexibility could affect software development employees decisions to stay longer in the company.

4.4.2 Summary of the Research Question

This section provides a summary of the results relating to the research question with a set of empirical retention factors to predict the decision to stay of software development employees. Standard MLR was used to predict what retention factors are important in retaining software development employees in the IT industry in Taiwan.

All Aggregated Employees Six empirical retention factors could evidence a significant positive correlation with the retention of all aggregated employees. They were job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility. In addition, five factors: job appreciation, fair remuneration, freedom in decision making, development, and workplace flexibility, were found to uniquely contribute to the equation with the decision to stay. Furthermore, 55% of variance in decision to stay could be explained by these five unique contribution factors for all aggregated employees. Therefore, the equation of the relationship between the factors could be predicted as follows.
152

Decision to stay for all aggregated employees = .163 x job appreciation + .310 x fair remuneration + .134 x freedom in decision making + .113 x new challenging work + .047 x development + .081 x workplace flexibility + .115

Software Engineers Six empirical retention factors could evidence a significant positive correlation with the retention of software engineers. They were job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility. In addition, two factors: fair remuneration and development, were found to uniquely contribute to the equation with the decision to stay. Furthermore, 57% of variance in decision to stay could be explained by these two unique contribution factors for software engineers. Therefore, the equation of the relationship between the factors could be predicted as follows.

Decision to stay for the software engineers = .114 x job appreciation + .354 x fair remuneration + .157 x freedom in decision making + .120 x new challenging work + .053 x development + .064 x workplace flexibility + .103

Project Leaders Six empirical retention factors could evidence a significant positive correlation with the retention of project leaders. They were job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility. In addition, there is only one job appreciation factor found to uniquely contribute to the equation with the decision to stay. Furthermore, 47% of variance in decision to stay could be explained by this one unique contribution factor for project leaders. Therefore, the equation of the relationship between the factors could be predicted as follows.
153

Decision to stay for the project leaders = .477 x job appreciation + .220 x fair remuneration + .020 x freedom in decision making - .110 x new challenging work + .037 x development + .062 x workplace flexibility + .313

Assistant Managers Six empirical retention factors could evidence a significant positive correlation with the retention of assistant managers. They were job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility. In addition, two factors: fair remuneration and freedom in decision making, were found to uniquely contribute to the equation with the decision to stay. Furthermore, 78% of variance in decision to stay could be explained by these two unique contribution factors for assistant managers. Therefore, the equation of the relationship between the factors could be predicted as follows.

Decision to stay for the assistant managers = -.509 x job appreciation + .290 x fair remuneration + .847 x freedom in decision making + .264 x new challenging work - .028 x development + .238 x workplace flexibility - .249

4.5 Conclusion
This chapter provides data analysis and results using SPSS version 17.0. Data analysis began with demographics and descriptive analysis of the respondents. Principal Component Analysis (PCA) was used to summarise six new components underlying the structure for this study. They were job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility. These six empirical retention factors were therefore used to test and explore the relationship with software development employees decision to stay in the IT industry in Taiwan.
154

The results of standard MLR focused on four groups (all aggregated employees, software engineers, project leaders, assistant managers). The group of all aggregated employees could have unique contribution to the equation associated with the factors of job appreciation, fair remuneration, freedom in decision making, development and workplace flexibility. The group of software engineers could have unique contribution to the equation associated with the factors of fair remuneration and development. The group of project leaders could have unique contribution to the equation associated with the factor of job appreciation. The group of assistant managers could have unique contribution to the equation associated with the factors of fair remuneration and freedom in decision making.

A post-hoc analysis with t-tests and the ANOVA test was used to further assess whether there were any significant differences between the various age groups and types of employment in terms of the six composite variables identified by the Principal Component Analysis (PCA). In t-tests, development was only one factor found to significantly differ between those aged between 20 and 29 and those aged between 30 and 44. However, there were no significant factors to be found in the ANOVA test. As a result of these analyses it could be concluded that there is uniformity amongst different age groups and employment categories with respect to the factors identified as pivotal to the decision to stay in the workplace.

All in all, the findings of the data analysis provided a base to develop knowledge of this thesis which informed discussion in Chapter 5.

155

Chapter 5: Discussion and Conclusion

5.1 Introduction
This chapter begins with an overview of each of the preceding chapters. In doing so, it examines the research problem and its link to prior literature. The methodology is then discussed and the findings reiterated and contextualised within what is known about retention and the IT industry in Taiwan. Finally, limitations of the research, implications / contributions for theory, policy and practice, and possible avenues for further research are discussed.

Chapter 1 introduced the background problems of this research from a broad perspective of global demographic trends and then focused on the importance of software development employees. The background problems of a decreasing core workforce and insufficient software development employees have created an important view of retention strategies for this thesis. The literature has shown that strategic human resource management of retention is a cost-effective way to resolve the shortage of software development employees as well as to sustain a competitive advantage for the IT companies. The objectives of this research are provided below:

3.

To identify what retention factors are important in retaining software development employees in the IT industry in Taiwan.

4.

To develop a tentative set of retention guidelines, informed by the research results, and appropriate to the IT companies in Taiwan, in order for them to effectively retain their software development employees.

156

In Chapter 2, the history of the Taiwanese IT industry was provided as well as the evolution of the Taiwanese economy towards knowledge-based industries. The shortages of the core workforce and IT software development employees were identified as negatively impacting on the competitiveness of the Taiwanese IT companies in todays knowledge-based economy. The research problem was then identified as What retention factors are important in retaining software development employees in the IT industry in Taiwan. In addition, financial compensation, career development (training and development and promotion), self-actualization (recognition, challenging work, and innovation and creativity), environmental factors (leadership style, autonomy, and work-life balance), and job satisfaction were retention factors established for this study. These retention factors constructed a theoretical framework which proposed to examine software development employees decisions to stay and answer the research question.

In Chapter 3, the justification for the use of quantitative methodology was provided as well as for the use of the online web-based survey for this research. This chapter clearly described the process of approaching selected IT companies in Taiwan, including the sample choice and questionnaire design and explained some of the data gathering difficulties encountered and the risk management strategies utilized. The survey was thus conducted through the Human Resource Bank in Taiwan and yielded 415 respondents from a total of 1000, reaching a 41.5% response rate. The quality of the research was assured by the validity and reliability measurement in which the value of the correlations should exceed .30 and the coefficient value of the reliability should exceed .70 of the acceptable level. The statistical methods were also chosen to examine the research hypotheses, which were identified in Chapter 2 and reiterated in Chapter 3 and 4. Finally, the ethical issues relating to the research were discussed.

157

In Chapter 4, data analysis was the main focus, in which the data was analysed on the basis of that collected from the responses of Taiwans software development employees. The results of analysed data were conducted by statistical test methods such as principal component analysis (PCA), multiple linear regression (MLR), t-tests and ANOVA test. These statistical techniques were believed to properly answer the research question. As the sample size of the three groups of valid respondents (software engineers: 313; project leaders: 72; and assistant managers: 30) was unequal and some were relatively small, the results about findings have been interpreted in more tentative terms throughout the entire thesis.

The principal component analysis (PCA) summarised six new retention factors that were constructed as an empirical model for this study. These were renamed as job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility. Therefore, six empirical research hypotheses have been developed as follows:

Empirical Hypothesis 1: Job appreciation is positively related to the software development employees decision to stay in the IT industry. Empirical Hypothesis 2: Fair remuneration is positively related to the software development employees decision to stay in the IT industry. Empirical Hypothesis 3: Freedom in decision making is positively related to the software development employees decision to stay in the IT industry. Empirical Hypothesis 4: New challenging work is positively related to the software development employees decision to stay in the IT industry.

158

Empirical Hypothesis 5: Development is positively related to the software development employees decision to stay in the IT industry. Empirical Hypothesis 6: Workplace flexibility is positively related to the software development employees decision to stay in the IT industry.

The results of correlation tests indicated that all six retention factors were significantly associated with the decision to stay for the four groups of all aggregated employees, software engineers, project leaders, and assistant managers. The results of MLR indicated that five retention factors could significantly and uniquely contribute to the regression equation with the decision to stay for all aggregated employees. Two retention factors could significantly and uniquely contribute to the regression equation with the decision to stay for software engineers and assistant managers. There was only one significant retention factor which could uniquely contribute to the regression equation with the decision to stay for project leaders. The results of t-tests indicated that there could be one significant retention factor in the comparison between two age groups (e.g. those aged between 20 and 29 and those aged between 30 and 44). Finally, the results of the ANOVA test indicated that there could be no significant retention factor among three groups of software development employees.

This Chapter (Chapter 5), which presents a discussion of the findings and implications / contributions of this research, consists of nine sections. It begins with the Introduction (Section 5.1), followed by the conclusions about the research hypotheses (Section 5.2), which discuss the outcome of the six empirical research hypotheses. Conclusions about the research question (Section 5.3) discuss the unique retention factors on the different types of software development employees. Conclusions about the research problem (Section 5.4) discuss a set of tentative guidelines for retention of each of the different types of software
159

development employees. Implications / contributions for theory, policy, and practice (Section 5.5 and 5.6) explain the contributions of the study to the knowledge as well as to the Taiwanese IT companies. Limitations of this research (Section 5.7) and implications for further research (Section 5.8) are also addressed. Finally, the chapter offers a brief conclusion (Section 5.9). Figure 5.1 outlines the structure of this chapter.

Figure 5.1: Structure of Chapter 5 5.1 Introduction 5.2 Conclusions about the Empirical Research Hypotheses 5.3 Conclusions about the Research Question 5.4 Conclusions about the Research Problem 5.5 Contributions for Theory 5.6 Contributions for Policy and Practice 5.7 Limitations of this Research 5.8 Implications for Further Research 5.9 Conclusion
Source: Developed for this study.

5.2 Conclusions about the Empirical Research Hypotheses

There were six retention factors summarised in Chapter 4 which have been used to form the empirical research hypotheses as the research model for this study. These empirical hypotheses aimed to examine the relationship between retention factors and software development employees decisions to stay in the IT industry in Taiwan. These empirical hypotheses were job appreciation, fair remuneration, freedom in decision making, new
160

challenging work, development, and workplace flexibility. This section discusses these empirical hypotheses based on the findings analysed in Chapter 4.

The empirical research hypotheses were examined across four groups of participants, namely all aggregated employees, software engineers, project leaders, and assistant managers. In the results of analysed data, the responses from four groups of software development employees could be all positively and significantly associated with the six retention factors (see Table 4.18).

Empirical Hypothesis 1: Job appreciation could be positively related to software development employees decisions to stay in the IT industry in Taiwan. This hypothesis was supported (see Figure 5.2, 5.3, 5.4, and 5.5), as the results for job appreciation variable were supported for all aggregated employees (r = .69, p < .05), software engineers (r = .69, p < .05), project leaders (r = .69, p < .05), and assistant managers (r = .79, p < .05). In the principal component analysis (PCA), the question, My boss congratulates me for doing a good job received the highest score from the total items of this component. This indicates that job appreciation from boss could be positively related to these software development employees decisions to stay in the IT industry in Taiwan.

These findings are supported by previous studies which show that software development employees would stay longer if they are appreciated as valued members of the organization (Luftman, 2008; Young, 2008). Gostick and Elton (2001) also stress that 79% of employees resign their positions because their professional contributions are under-appreciated. This is because recognition pays out psychic income which craves the form of tangible and intangible expressions of acknowledgment and appreciation

161

(Foote, 2006). It is inferred that a recognition program could be a cheap and effective way to forge a long term relationships with employees, resulting in steadily rising employee satisfaction and retention (Jakobson, 2008). In addition, a recognition program is also inferred to be more effectively practiced through the use of bonuses and innovative awards. For instance, a $100 bonus for an employee-of-the-month award may not be a significant amount of money, but employees could appreciate receiving this type of award and feel that they are being valued (DeMers, 2002).

Empirical Hypothesis 2: Fair remuneration could be positively related to software development employees decisions to stay in the IT industry in Taiwan. This hypothesis was supported (see Figure 5.2, 5.3, 5.4, and 5.5), as the results for fair remuneration variable were supported for all aggregated employees (r = .66, p < .05), software engineers (r = .68, p < .05), project leaders (r = .61, p < .05), and assistant managers (r = .59, p < .05). In the principal component analysis (PCA), the question, This organisation pays me fairly for the work I do received the highest score from the total items of this component. This indicates that a fair remuneration system could be positively related to these software development employees decisions to stay in the IT industry in Taiwan.

These findings are supported by previous studies which show that although money is successfully a temporary motivator, a fair and comparable pay can significantly attract and retain top talents (Gee & Burke, 2001; Huang, Lin, & Chuang, 2006; Mosley & Hurley, 1999). It is noted that top-tier IT companies pay 32% more than average for their software development employees (Tynan, 2006), and especially use stock as a form of compensation (Dobson, 2007; Hansen, 2001). Chiu, Luk, and Tang (2001) also argue that software development employees always rationally evaluate various work behaviours and

162

choose those that lead to the work-related rewards that they value most. Therefore, it is inferred that a fair remuneration could play an important role in influencing software development employees decisions to stay in the IT companies.

Empirical Hypothesis 3: Freedom in decision making could be positively related to software development employees decisions to stay in the IT industry in Taiwan. This hypothesis was supported (see Figure 5.2, 5.3, 5.4, and 5.5), as the results for freedom in the decision making variable were supported for all aggregated employees (r = .64, p < .05), software engineers (r = .65, p < .05), project leaders (r = .52, p < .05), and assistant managers (r = .86, p < .05). In the principal component analysis (PCA), the question, I have authority to make the decisions necessary to do my job received the highest score from the total items of this component. This indicates that personal control could be positively related to these software development employees decisions to stay in the IT industry in Taiwan.

These findings are supported by previous studies which show that software development employees naturally resist work content that is command-and-control style, and always seek autonomy in their work (Davenport, Thomas, & Cantrell, 2002; Horwitz, Heng, & Quazi, 2003; Nair & Vohra, 2010). It is essential that their work is meaningful and allows them to get involved and participate in the decision-making process as they may feel rewarded in the knowledge and skills that are their expertises (Haar & Spell, 2009; Nair & Vohra, 2010). More importantly, a considerable level of freedom in decision making has been viewed as a non-monetary reward, which may help satisfy employee needs in the same way that pay does (Campion & Berger, 1990; Nair & Vohra, 2010; Williams, McDaniel, & Nguyen, 2006). Therefore, it is inferred that a considerable level of freedom in decision making in the workplace could be a critical factor that influences these

163

software development employees decision to stay in the IT industry in Taiwan.

Empirical Hypothesis 4: New challenging work could be positively related to software development employees decisions to stay in the IT industry in Taiwan. This hypothesis was supported (see Figure 5.2, 5.3, 5.4, and 5.5), as the results for new challenging work variable were supported for all aggregated employees (r = .63, p < .05), software engineers (r = .64, p < .05), project leaders (r = .51, p < .05), and assistant managers (r = .84, p < .05). In the principal component analysis (PCA), the question, My job is challenging received the highest score from the total items of this component. This indicates that new challenging work could be positively related to these software development employees decisions to stay in the IT industry in Taiwan.

These findings are supported by previous studies which show that the nature of work content for software development employees is governed primarily by their own expertise rather than by a routine or system (Kochanski & Ledford, 2001; Lock, 2003). Lacity, Iyer, and Rudramuniyaiah (2008) also stress that software development employees do not like programming from predefined specifications, because task variety or skill set utilization are the major reasons for their satisfaction. It is believed that software development employees value and thrive on interesting work that challenges them and uses their skills and talents (Kochanski & Ledford, 2001). Furthermore, it is also noted that almost 95% of software development employees rate challenging work as being very important in determining turnover intention (Liz, Blackman, & Hurst, 2007). Therefore, it is inferred that new challenging work could be a critical influence on these software development employees decisions to stay in the IT industry in Taiwan.

164

Empirical Hypothesis 5: Development could be positively related to software development employees decisions to stay in the IT industry in Taiwan. This hypothesis was supported (see Figure 5.2, 5.3, 5.4, and 5.5), as the results for development variable were supported for all aggregated employees (r = .42, p < .05), software engineers (r = .40, p < .05), project leaders (r = .42, p < .05), and assistant managers (r = .78, p < .05). In the principal component analysis (PCA), the question, My confidence in using the knowledge and skills learned has decreased (reverse) received the highest score from the total items of this component. This indicates that offering an opportunity for career development could be positively related to these software development employees decisions to stay in the IT industry in Taiwan.

These findings are supported by previous studies which show that software development employees tend to stay longer if organizations provide long-term career developments (Acton & Golden, 2003; Kochanski, Mastropolo, & Ledford, 2003). Whitaker (1999) also supports the view that career development is an important factor to retain IT workers, especially those who are under 30 years old. This is because career development is involved with specific learning opportunities, which provide future prospects for younger professionals as well as to preempt offers from competitors (Hansen, 2007; Sweeney, 2008). In addition, 86% of software development employees see self-improvement as important to success while 97% see up-to-date skills as crucial (Shah, Sterrett, Chesser, & Wilmore, 2001, p.22). Therefore, it is interred that career development could make a difference in retaining the group of Millennial (Generation Y or Net Generation: born between the years of 1977 and 1997; aged between 12-32 years old) in the IT industry in Taiwan.

165

Empirical Hypothesis 6: Workplace flexibility could be positively related to software development employees decisions to stay in the IT industry in Taiwan. This hypothesis was supported (see Figure 5.2, 5.3, 5.4, and 5.5), as the results for workplace flexibility variable were supported for all aggregated employees (r = .52, p < .05), software engineers (r = .50, p < .05), project leaders (r = .54, p < .05), and assistant managers (r = .72, p < .05). In the principal component analysis (PCA), the question, I have sufficient time to spend with my family received the highest score from the total items of this component. This indicates that flexible work could be positively related to these software development employees decisions to stay in the IT industry in Taiwan.

These findings are supported by previous studies which show that workplace flexibility is a trend among professionals in Western countries such as Australia to remain in a long-term employment relationship (Cooper, 2006; De Cieri et al., 2008). More particularly, balance life between work and family activity is more important than money, as values in society may be shifting towards a greater emphasis on quality of life and less on material gain (Blyton & Jenkins, 2007; Tynan, 2006). It is noted that 60% of software development employees rate time and flexibility as key factors in deciding whether or not to take or keep a job (Tynan, 2006). However, only 35% of employers rate it as important (Tynan, 2006), so employers attitudes play a key role in determining flexible work schedules within an organization (Trinczek, 2006). This could particularly be the major reason that more than half the women leave the IT industry in their mid-to-late 30s, as the IT industry is regarded as having a family-unfriendly atmosphere (Anonymous, 2008). Therefore, it is inferred that flexible work schedules could make a difference in retaining the group of generation Xers (born between the years of 1965 and 1976; aged between 33-44 years old), particular those who have small children or elder family members (Anonymous, 2006; Colman, 2004; Zetlin, 2002).

166

The diagrams below illustrates the correlation of the six retention factors for the four groups of software development employees, namely all aggregated employees, software engineers, project leaders, and assistant managers.

Figure 5.2: Correlation results for the retention of all aggregated employees

0.69 (+)

H1: Job Appreciation

0.66 (+)

H2: Fair Remuneration

0.64 (+)

H3: Freedom in Decision Making

0.63 (+)

Retention of All Aggregated Employees

H4: New Challenging Work

0.42 (+)

H5: Development

0.52 (+)

H6: Workplace Flexibility

Source: Developed for this research.

167

Figure 5.3: Correlation results for the retention of software engineers

0.69 (+)

H1: Job Appreciation

0.68 (+)

H2: Fair Remuneration

0.65 (+)

H3: Freedom in Decision Making

Retention of Software Engineers

0.64 (+)

H4: New Challenging Work

0.40 (+)

H5: Development

0.50 (+)

H6: Workplace Flexibility

Source: Developed for this research.

168

Figure 5.4: Correlation results for the retention of project leaders

0.69 (+)

H1: Job Appreciation

0.61 (+)

H2: Fair Remuneration

0.52 (+)

H3: Freedom in Decision Making

Retention of Project Leaders

0.51 (+)

H4: New Challenging Work

0.42 (+)

H5: Development

0.54 (+)

H6: Workplace Flexibility

Source: Developed for this research.

169

Figure 5.5: Correlation results for the retention of assistant managers

0.79 (+)

H1: Job Appreciation

0.59 (+)

H2: Fair Remuneration

0.86 (+)

H3: Freedom in Decision Making

Retention of Assistant Managers

0.84 (+)

H4: New Challenging Work

0.78 (+)

H5: Development

0.72 (+)

H6: Workplace Flexibility

Source: Developed for this research.

170

5.3 Conclusions about the Research Question

This section is divided into four subsections in order to answer the research question, which is What retention factors are important in retaining software development employees decisions to stay in the IT industry in Taiwan. The principal component analysis (PCA) was used to summarise the importance of retention factors that have been developed into a retention model for this study. Six retention factors: job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility, have been identified as factors that influence the software development employees decision to stay in the IT industry in Taiwan. As a result, four subsections: conclusions for correlation tests, conclusions for standard multiple linear regression (MLR) tests, conclusions for t-tests, and conclusions for ANOVA tests, are presented in this section.

The correlation results concluded how each of the four groups of the software development employees could be positively and strongly correlated with each of the retention factors. The four groups of software development employees included all aggregated employees, software engineers, project leaders, and assistant managers. In addition, standard MLR results concluded how the software development employees decisions to stay in the IT industry in Taiwan could be significantly predicted by a set of retention factors. Furthermore, t-tests results concluded how two age groups of the software development employees could be significantly different for each of the retention factors. The two age groups are those aged between 20 and 29 and those aged between 30 and 44. Finally, the ANOVA results concluded how three different types of software development employees could be significantly different with each of the retention factors. These conclusions were based on the findings analysed in Chapter 4.

171

5.3.1 Conclusions for Correlation Tests

The correlation results indicated that six retention factors could all be positively correlated with decisions to stay for all aggregated employees, software engineers, project leaders, and assistant managers in the IT industry in Taiwan. Table 5.1 provides the correlation results for the four groups of software development employees.

Table 5.1: Correlation results for four groups of software development employees All Aggregated Software Employees Engineers (n=409) (n=311) Job Appreciation Fair Remuneration Freedom in Decision Making New Challenging work Workplace Flexibility Development
Source: Developed for this study.

Project Leaders (n=69) 1 (r = .69) 2 (r = .61) 4 (r = .52) 5 (r = .51) 3 (r = .54) 6 (r = .42)

Assistant Managers (n=29) 3 (r = .79) 6 (r = .59) 1 (r = .86) 2 (r = .84) 5 (r = .72) 4 (r = .78)

1 (r = .69) 2 (r = .66) 3 (r = .64) 4 (r = .63) 5 (r = .52) 6 (r = .42)

1 (r = .69) 2 (r = .68) 3 (r = .65) 4 (r = .64) 5 (r = .50) 6 (r = .40)

The results of the correlation test indicated that job appreciation could be the most important factor correlated with decisions to stay for the all aggregated employees, followed by fair remuneration, freedom in decision making, new challenging work, workplace flexibility, and development. In addition, when looking at a pattern for retention factors across three different types of employees, the fair remuneration factor could be strongly and significantly correlated with bottom-line software engineer employees. The result of software engineers for the fair remuneration factor (r = .68) was almost the same as the highest correlation for the job appreciation factor (r = .69).

In contrast, the pattern for assistant managers as upper level employees was that freedom in decision making (which had the highest correlation: r = .86) could be strongly and
172

significantly correlated with their decision to stay. The results of these patterns indicated that the empirical study of this retention model could be explained by the Herzbergs two-factor theory that hygiene factors such as fair payment not only to prevents employees from job dissatisfaction but could also strongly retain software engineers in the IT industry in Taiwan. The motivator of freedom in decision-making not only satisfies employees, but could also strongly retain assistant managers in the IT industry in Taiwan.

5.3.2 Conclusions for Standard MLR Tests

The standard MLR results indicated that five retention factors significantly and uniquely contributed to the regression equation with the decision to stay for all aggregated employees These included job appreciation (p < .05), fair remuneration (p < .05), freedom in decision making (p < .05), development (p < .05), and workplace flexibility (p < .05). Therefore, it is deemed that these five retention factors could be a set of unique contributors which predict all aggregated employees decisions to stay, and fair remuneration (t = 6.45) is valued most highly, followed by development (t = 3.02), job appreciation (t = 2.09), workplace flexibility (t = 2.05), and freedom in decision making (t = 1.92).

The standard MLR results also indicated that two retention factors could significantly and uniquely contribute to the regression equation with the decision to stay for software engineers. These factors were fair remuneration (p < .05) and development (p < .05). Therefore, it is deemed that these two retention factors could be a set of unique contributors which predict software engineers decisions to stay, and fair remuneration (t = 6.31) is valued most highly, followed by development (t = 3.02).

173

The standard MLR results also indicated that job appreciation (p < .05) could be one factor that significantly and uniquely contributed to the regression equation with the decision to stay for project leaders. Therefore, it is deemed that job appreciation (t = 2.52) could be a critical factor to uniquely predict the decision to stay for project leaders.

The standard MLR results also indicated that two retention factors significantly and uniquely contributed to the regression equation with the decision to stay for assistant managers. These two factors were fair remuneration (p < .05) and freedom in decision making (p < .05). Therefore, it is deemed that these two retention factors could be a set of unique contributors which predict assistant managers decisions to stay, and freedom in decision making (t = 2.78) is valued most highly, followed by fair remuneration (t = 2.17).

Table 5.2 provides MLR results for the four groups of software development employees in descending order of importance from left to right.

Table 5.2: Standard MLR results for four groups of software development employees Retention factors that uniquely contribute to the decision to stay All Aggregated Employees Software Engineers Project Leaders Assistant Managers Fair Remuneration Fair Remuneration Job Appreciation Freedom in Fair Decision Making Remuneration Development Freedom in Job Workplace Decision Appreciation Flexibility Making

Development

Source: Developed for this study.

174

All Aggregated Employees Fair remuneration, development, job appreciation, workplace flexibility, and freedom in decision making were the five unique contributors to the regression equation that predicted all aggregated employees decisions to stay. In addition, fair remuneration could be ranked as the first unique retention factor for the all aggregated employees. It is deemed that base pay is an exchange for employees effort and ideas for an organization, and incentive pay (e.g. stock bonus) is a motivator to keep them staying longer within an organization (Chiu, Luk, & Tang, 2002; Rynes, Colbert, & Brown, 2002; Wang, 2007). In addition, the three highest scored items for this factor: This organisation pays me fairly for the work I do., People at this organisation are paid fairly according to their job performance, and My total pay is adequate compared to others in this organisation., all indicated that a fair treatment of pay structure could be critical to retain software development employees in the IT industry in Taiwan. This finding is supported by some studies that individuals assess the fairness of their pay by comparing their own ratio of inputs (e.g., skills and efforts) and outcomes (e.g., pay and recognition) to the inputs and outcomes ratio of others, as the equity theory suggests (Messmer, 2006; Pare, Tremblay, & Latrick, 2001; Ramlall, 2004; Rynes, Gerhart, & Minette, 2004).

Development could be ranked as the second unique retention factor to the regression equation that predicted all aggregated employees decisions to stay. It is deemed that an individual career development supported by their supervisor could retain software development employees, as their social psychological needs and preferences have been satisfied (Bhal & Gulati, 2006; Mendonsa, 1998). In addition, the three highest scored items for this factor: My confidence in using the knowledge and skills learned has decreased (reverse)., This organisation does not recognise those who come up with new ideas (reverse), and I have lost interest in my job (reverse)., indicated that an emphasis
175

on the individuals employability and growth development could be critical to retain software development employees in the IT industry in Taiwan. This finding is supported by some studies that self-confidence and up-to-date skills are required to develop an individuals competence in responding to the impacts of globalisation and demographic changes (Baruch, 2006; D'Art & Turner, 2006; Lamb & Sutherland, 2010).

Job appreciation could be ranked as the third unique retention factor to the regression equation that predicted all aggregated employees decisions to stay. It is deemed that recognition with considerable rewards can retain and motivate software development employees by letting them know that their efforts have been appreciated (Alexander, 2002; Jakobson, 2008). In addition, the three highest scored items for this factor: My boss congratulates me for doing a good job., My boss takes a supervisor role in my career development., and I am adequately recognised for my contributions., indicated that an opportunity for a next step in career growth could influence software development employees decisions to stay. This is because recognition awards may be effective for the first three years of employees tenure (Bhatnagar, 2007), and as well as this, it is the time that base people regularly review their psychological contracts with their organisations (Agarwal & Ferratt, 2000; Thomson, 2007). However, some studies have supported the view that recognition awards are truly symbols of achievement as Maslows higher order needs suggest (Gee & Burke, 2001; Hansen, Smith, & Hansen, 2002; Jakobson, 2008; Ramlall, 2004).

Workplace flexibility could be ranked as the fourth unique retention factor to the regression equation that predicted all aggregated employees decisions to stay. It is deemed that a considerate policy of flexible work arrangement can retain software development employees, especially for those who have suffered from significant burnout and stress in
176

the workplace (Dychtwald, et al., 2006; Hayes, 2003; Holland, Sheehan, Donohue, & Pyman, 2007). In addition, the three highest scored items for this factor: I have sufficient time to spend with my family., I am provided with the flexibility needed to balance the demands of my work and personal life, and I have the flexibility to arrange my work so that I can meet my business objectives and balance my family and personal needs., indicated that a flexible work arrangement could be a critical factor that influences software development employees decisions to stay. This finding is supported by some studies that workplace flexibility can create a better working relationship that simultaneously recognises and realises the needs of organisations and workers, especially for those who have small children (Budhwar, Saini, & Bhatnagar, 2005; Sumner, 2008; Wickramasinghe & Jayabandu, 2007).

Freedom in decision making could be ranked as the last unique retention factor to the regression equation that predicted all aggregated employees decisions to stay. It is deemed that giving freedom in decision making for the area of employees expertise is a way to show respect to them, as autonomy and independence are their identities (Davenport, 2005; Drucker, 2007). In addition, the three highest scored items for this factor: I have authority to make the decisions necessary to do my job., I am given an opportunity to present and try new ideas, and I am satisfied with the level of autonomy associated with my positions., indicated that allowing freedom in the way of doing things could influence software development employees decisions to stay. This finding is supported by many studies that satisfactory autonomy can be achieved through an increase in authority and responsibility to try new ideas or to explore a new ways of achieving things (Feyerherm & Vick, 2005; Kuo & Chen, 2004; Yigitcanlar, Baum, & Horton, 2007).

177

Software Engineers Fair remuneration and development were the two unique retention factors to the regression equation that predicted software engineers decisions to stay. In addition, fair remuneration could be the first unique retention factor, followed by development, for software engineers. This indicated that this group of software engineers could be motivated by a fair and comparable salary and individuals employability, as stated in the previous subsection of all aggregated employees. It is inferred that most software engineers could be juniors at the career stage of Exploration, so they could have needs in ongoing development of new skills which can stimulate personal growth and stay in tune with industry (Amundson, 2007; Chen, Chang, & Yeh, 2006; Moderi & Foote, 2005; Pare, et al., 2001).

Project Leaders Job appreciation was the only one unique retention factor to the regression equation that predicted project leaders decisions to stay. This indicated that this group of project leaders could only be motivated by a next step of career appreciation, as stated in the previous subsection of all aggregated employees. It is inferred that project leaders could be experienced employees and tied in with the second career stage of Establishment, which seeks achievements such as career progression. The career progression could be learning broad up-to-date skills, managerial skills, or even promotion. In addition, some project leaders might not have any interest or demonstrated ability to manage people, but could have an interest in the development of technical skills for their career progression (Nelson & Todd, 2004). As a result, dual tracks (e.g., managerial and technical) for career progression could be inferred as the unique retention factor for project leaders (Crepeau, Crook, Goslar, & McMurtrey, 1992; Davenport, 2005; Hsu, Chen, Jiang, & Klein, 2003; Petroni, 2000; Yarnall, 2008).

178

However, a continuously open and honest communication between subordinate and superior must be fostered to prevent a misfit situation of the new arrangement (Hsu, et al., 2003). This is because project leaders could be smart enough to recognise temporary retention measures for what they are, and their market value (Moderi & Foote, 2005; Thomson, 2007). Consequently, career appreciation for project leaders could be achieved, as well as sustainable retention measures for an organisation.

Assistant Managers Freedom in decision making and fair remuneration were the two unique retention factors to the regression equation that predicted assistant managers decisions to stay. In addition, freedom in decision making could be the first unique retention factor, followed by fair remuneration, for assistant managers. This indicated that this group of assistant managers could be motivated by having the power to exercise new ideas and by receiving a different level of remuneration from their pay structure, as stated in the previous subsection of all aggregated employees. It is inferred that assistant managers could be the successful members of an organisation, so they could be naturally given more responsibilities and trusts that allow them to allocate resources according to their preferences (Kuo & Chen, 2004; Yigitcanlar, et al., 2007). It is also inferred that assistant managers could be the successor of career achievement at the career stage of Establishment, so they could be motivated by receiving different pay levels and pay content as fair treatment (Huang, Lin, & Chuang, 2006; Pare, et al., 2001; Wang, 2007).

5.3.3 Conclusions for T-Tests

This section identifies whether or not there is a difference between two age groups of employees, which includes those aged between 20 and 29 and those aged between 30 and 44 across seven composite variables. The conclusion of this section is based on the
179

post-hoc analysis of the t-tests results in Chapter 4. The results of t-tests indicated that only one factor development could be significantly related to those aged between 20 and 29 in comparison to those aged between 30 and 44 in the IT industry in Taiwan. Table 5.3 provides a summary of the t-tests results for the two age groups.

Table 5.3: Summary of t-tests results for age groups Composite variables H1: Job Appreciation H2: Fair Remuneration H3: Freedom in Decision Making H4: New Challenging work H5: Development H6: Workplace Flexibility H7: Decision to stay
Source: Developed for this study.

Age group between 20 and 29 years old vs. Age group between 30 and 44 years old No Sig. No Sig. No Sig. No Sig. Sig. No Sig. No Sig.

The results of development indicated that when comparing mean scores of these two age groups, those aged between 20 and 29 (2.95) had lower mean scores than those aged between 30 and 44 (3.09). The responses from those aged between 30 and 44 indicated that they could be more confident in using the knowledge and skills learned than did the responses from those aged between 20 and 29. It is inferred that those aged between 30 and 44 could be senior members, project leaders or assistant managers who have been familiar with the project tasks and work environment, leading them to feel more confident in using their knowledge and skills. On the other hand, those aged between 20 and 29 could be juniors or newcomers, who are just beginning their career and trying to establish their rapport relationships with their project tasks and work environment as a whole. Rollag, Parise, and Cross (2005) stress that although newcomers (mostly software

180

engineers) are a source of fresh ideas and expertise, they tend to be more risk-averse in their decision making and activities because they are fitting in with the organization. Therefore, the responses from those aged between 30 and 44 inferred that they could have more self-confidence in using knowledge and skills at work than did the responses from those aged between 20 and 29.

5.3.4 Conclusions for ANOVA Test

This section discusses whether or not there is a difference among three different types of software development employees, including software engineers, project leaders, and assistant managers across seven composite variables. The conclusion of this section is based on the post-hoc analysis of the ANOVA test results in Chapter 4. The results of the ANOVA test indicated that there could be no significant factor related to three different types of software development employees in the IT industry in Taiwan. Table 5.4 provides a summary of the ANOVA results for the comparisons between three different types of software development employees.

Table 5.4: Summary of ANOVA results Composite variables H1: Job Appreciation H2: Fair Remuneration H3: Freedom in Decision Making H4: New Challenging work H5: Development H6: Workplace Flexibility H7: Decision to stay
Source: Developed for this study.

Software Engineers vs. Project Leaders vs. Assistant Managers No Sig. No Sig. No Sig. No Sig. No Sig. No Sig. No Sig.

181

5.4 Conclusions about the Research Problem

The background problems of this study have been identified as the IT industry in Taiwan suffering from the impact of a national shortage in its labour force, as well as a shortage of software development employees (CEPD, 2006). More importantly, there is a strong demand for economic transformation in Taiwan with a shift from quantity to quality under todays knowledge-based economy (Chen & Liu, 2003). These severe situations are related to human resource issues which attracted the researcher to find a way for an appropriate solution. Although training is an important approach to solving the shortage within the labour force, retention of existing software development employees is important and is the focus of this research. Therefore, the research problem has been identified as follows:

What retention factors are important in retaining software development employees in the IT industry in Taiwan?

In addition, a diverse set of useful tentative guidelines for retention, appropriate to IT companies, is provided for effectiveness in retaining the three different types of software development employees.

The concept of retention is an established theory in human resource management. However, in recent years it has regained prominence due to highly skilled workforce shortages across all industries and most nations (Dychtwald, et al., 2006). This is because demographics in the labour market has been dramatically changed with the disproportionate size of the retiring group of the baby boomer generation, a decreasing core workforce, and decreasing birthrates of the young cohort (Dychtwald, et al., 2006). The IT industry in Taiwan is also inevitably affected by the impact of the demographic changes, as the majority of respondents (98.7%) for this study is from the core workforce
182

(aged between 15 to 44 years old) which is expected to decrease by about 4% (about 300,000 labourers) by 2012 (CEPD, 2008).

From the aspects of labour availability, the activities of business are severely undermined if existing software development employees keep leaving. This is because projects cannot be conducted, and the know-how of IT companies cannot be passed on to junior employees from senior employees. In addition, from the scale of age allocation, software development employees mainly span across two generations as Generation X and Millennials (Generation Y and Net Generation). This is important for a set of retention guidelines as each generation has different needs according to their characteristics of career stages (Chen, et al., 2006; Nicholas & Guzman, 2009). For example, Generation Xers may have reached middle management status whereas Millennials are just beginning their careers (Sattar, 2007).

From the productivity aspect of IT knowledge work, the innovation of products and services is severely undermined if there are not sufficient qualified workers in todays knowledge-based economy. Because the programming work of software development employees requires a high level of technical knowledge, Taiwan has been undergoing huge pressures for an improvement of the intellectual capital supplied from the education system. Creativity also relies on intellectual capital, and this can be undermined by the educational emphasis on rote memorization. In addition, as Mandarin is the official language of Taiwan, and English is the dominant language on the Internet, learning opportunities through this network are limited, and access to advanced knowledge or scholarly publishing opportunities is constrained. Therefore, it is deemed that a high transaction cost is involved with the development of human intellectual capital (either in poaching or promoting a career development within an organization), as it is an exchange
183

for knowledge which is highly related to a source of profit, skill, and sustainable competitive advantage (Drucker, 1993).

Prior studies have shown that turnover hurts business because of the increased administration expenses associated with recruiting, hiring, and training replacements, which could cost up to 2.5 times the departing employees salary (Cascio, 2006; Luftman & Kempaiah, 2007). More importantly, it is highly associated with loss of skills and company knowledge, low morale, poor customer satisfaction, and eventually financial losses (Frank, Finnegan, & Taylor, 2004; Somaya & Williamson, 2008). These studies show that turnover is not only involved with the expense of replacing departing employees, but also highly associated with the loss of sustainable competitive advantages of IT companies. IT companies could fail if nothing is done in terms of retention strategies for existing software development employees. Although the nature of the IT work environment is involved with work exhaustion (e.g. work overload), a useful set of tentative retention guidelines is expected to mitigate the impact of the shortage of software development employees. This is because this set of tentative retention guidelines is tailored to meet unique retention factors as well as common retention factors of each type of different software development employees based on the literature evidence and empirical research hypotheses.

Table 5.5 provides a summary of retention factors of each type of different software development employees.

184

Table 5.5: Summary of retention factors of software development employees

All Aggregated Software Employees Engineers

Project Leaders Common Common Unique Common Common Common

Assistant Managers Unique Common Common Common Unique Common

Fair Remuneration Development Job Appreciation Workplace Flexibility Freedom in Decision Making New Challenging work
Source: Developed for this study.

Unique Unique Unique Unique Unique Common

Unique Unique Common Common Common Common

A Set of Retention Guidelines for All Aggregated Employees According to Table 5.5, the unique retention factors of all aggregated employees could be related to fair remuneration, development, job appreciation, workplace flexibility, and freedom in decision making. The managers of human resource departments could regard this group as representative of software departments for effective human resource management. The common retention factor of all aggregated employees could be related to new challenging work, which the managers of human resource departments could keep in mind as a reference. Therefore, a useful set of retention guidelines all aggregated employees based on their unique retention factors as detailed in the findings could be concluded as follows.

1. As all aggregated employees span across two generations including Generation X and Millennials (Generation Y and Net Generation), the emphasis of their motivations is suggested to link to each individuals values, where software engineers may seek technical competence; project leaders may seek dual career development; and assistant managers may seek further freedom in decision making and status. More importantly, any decision-making regarding their career

185

development may need well communicated, as a long-term employment relationship is developed based on mutual understanding of career growth and progression (Coleman, 2001). 2. Fair remuneration is a straightforward value of employees contributions to the organization. From this, they are able to satisfy physiological and security needs, or even higher-order needs of recognition in society, as they have more money for play, consumption, and leisure activities (Chiu, et al., 2002). It is always suggested that a fair and comparable compensation of pay level and pay content may be important to sustain a long-term employment relationship. This is because software development employees may value a high level of internal equity (e.g. compared to counterparts within an organization) and external equity (e.g. compared to counterparts outside an organization) (Pare, et al., 2001). 3. A career development program is suggested to make software development employees feel that they may have gained opportunities for self-development and career advancement. This is important because they may be interested in trying new technologies and learning up-to-date skills. In addition, training in interpersonal / communication skills is also fundamental, because they may be often promoted to managerial positions based on technical skills and little is done to assess their leadership skills. 4. Recognition with considerable rewards is suggested to make these types of software development employees feel that they have been appreciated for the efforts they have put in at work. An ongoing consultation with their career development may further recognise their personal needs and preferences as well as their contributions to the organisation. 5. A flexible work schedule may be much more important than making good money because these types of software development employees may prefer a balanced
186

life between work and family activities (Nicholas & Guzman, 2009). 6. A considerate level of freedom in decision making over project tasks is suggested because these types of software development employees may want to be led rather than managed. It is deemed that their self-achievement may be satisfied when they are highly involved in the process of decision-making over their project tasks (Feyerherm & Vick, 2005).

A Set of Retention Guidelines for Software Engineers According to Table 5.5, the unique retention factors of software engineers could be related to fair remuneration and development, which the managers of human resource departments could pay more attention to for effective human resource management. The common retention factors of software engineers could be related to job appreciation, workplace flexibility, freedom in decision making, and new challenging work, which the managers of human resource departments could keep in mind as a reference. Therefore, a useful set of guidelines for software engineers based on their unique retention factors could be concluded in the findings as follows.

1. Fair remuneration is an exchange for employees efforts and ideas for an organization, so it is suggested that offering a comparable salary may attract and retain software engineers. This is because they may rationally evaluate a fair and comparable pay with others. 2. The emphasis of an individuals value is a key to retaining software development employees, because of the individuals fulfilment relying heavily on the measurement of their job satisfaction. As most software engineers are newcomers who are in the first career stage of Exploration, opportunities for self-development with technical competence could be emphasized. It is also
187

suggested that a self-development program may tie in with the individuals needs.

A Set of Retention Guidelines for Project Leaders According to Table 5.5, job appreciation could be the only unique retention factor that applied for project leaders. Thus, the managers of human resource departments could be aware of this unique retention factor of the project leaders. The common retention factors of project leaders could be related to fair remuneration, development, workplace flexibility, freedom in decision making, and new challenging work, which the managers of human resource departments could keep in mind as a reference. Therefore, a useful set of tentative retention guidelines for project leaders based on their unique retention factors could be concluded in the findings as follows.

1. As project leaders are deemed in the second career stage of Establishment, they usually attempt to progress their own career development to a particular occupational field which could be either management responsibility or technology management. Thus, their career appreciation could be met within the field they value the most. It is suggested that section managers could properly communicate with them before any career development decision has been made. 2. It is also suggested that small amounts of incentive rewards (e.g. money for a team dinner) could be an encouragement for them as well as for the team members each time they complete the project.

A Set of Retention Guidelines for Assistant Managers According to Table 5.5, the unique retention factors of assistant managers could be related to freedom in decision making and fair remuneration, which the managers of human
188

resource departments could pay more attention to for effective human resource management. The common retention factors of assistant managers could be related to development, job appreciation, workplace flexibility, and new challenging work, which the managers of human resource departments could keep in mind as a reference. Therefore, a useful set of tentative retention guidelines for assistant managers based on their unique retention factors could be concluded from the findings as follows.

1. Assistant managers are also deemed to be in the second career stage of Establishment, which is regarded as a competitive stage with high payoffs for superior performance, so they may seek overall achievement by getting ahead of competition, and with personal growth, self-esteem and competence (Chen, et al., 2006). Thus, a considerate level of freedom in decision making could be a symbol of their achievement of professional success. 2. As different levels of IT employees value fair treatment differently, remuneration (including an annual bonus and managerial perks) for assistant managers could reflect a different pay level and pay content.

5.5 Contributions for Theory

Although retention theory has recently attracted attention for many human resource scholars, there is less literature in the Taiwanese IT context. This makes the contribution of knowledge more extraordinary within the Taiwanese context as this research has been developed based on the literature evidence and empirical research hypotheses. Much knowledge of retention studies has come from Western countries such as the United States and shows that qualified workers are getting more difficult to find, specifically those who possess skills in the IT industry (Agarwal & Ferratt, 2002; Williams, 2007). The impact of global demographic changes and rapid changes in technology are the main reasons that
189

affect the need for software development employees so dramatically. It is also noted that a long-term employment relationship is a way to retain software development employees by developing their career and providing them security, better work content and a friendly work environment (Agarwal & Ferratt, 2001; Amaram, 2005; Horwitz, Heng, & Quazi, 2003; Perry, 2001; Rumpel & Medcof, 2006). Therefore, the knowledge contribution of this retention model has provided a better concept of managing voluntary turnover of software development employees within the Taiwanese IT industry context.

The second theoretical contribution of this study is that the cluster of IT workers focuses on software development employees, namely software engineers, project leaders and assistant managers. The findings indicate that there are similarities and differences among these three different types of software development employees, which provide a better understanding of their differing unique and common retention factors in relation to the retention factors. This is a significant knowledge contribution as it provides a reference for those who need this kind of knowledge and information for effective management of software development employees in the IT industry in Taiwan.

The third theoretical contribution of this study is that the research has further compared the differences in retention factors between two age groups of the software development employees. This is a unique contribution to the knowledge. The results of the two age groups are deemed to provide new aspects of retention theory within the Taiwanese context. This is because different age groups of the software development employees have different needs as they mainly span across two generations including Generation X and Millennials (Generation Y and Net Generation).

The final theoretical contribution of this study is that the knowledge gap about retention in
190

the IT industry in Taiwan can be filled by the recommendations of this research. This is concluded based on the diverse retention factors of different software development employees, as career growth and progression can mainly satisfy them under a severe work environment which is highly involved with long hours and high pressure. The knowledge of this study is deemed to provide a better approach for human resource managers to deal with the shortage of software development employees.

5.6 Contributions for Policy and Practice

A comprehensive human resource management policy on retention is necessary for the different types of software development employees as they are diverse in needs according to their career stages and generations. Therefore, a diverse set of tentative retention guidelines for the human resource managers of IT companies is developed based on their unique retention factors. This is deemed to provide a significant difference for the IT companies that have not applied any retention policy to their software development employees. This could also be a good reference for the IT companies that have applied some retention policies to their software development employees as well. Table 5.6 provides a summary of career stages and generations for the software development employees.

191

Table 5.6: Summary of career stages and generations for software development employees All Aggregated Employees Career Stage Generation
Unique retention factors

Software Engineers Exploration M/X FR DEV

Project Leaders Establishment M/X JA

Assistant Managers Establishment X FDM/ FR DEV/ JA/ WF/ NCHA

Exploration/ Establishment M/X FR/ DEV/ JA/ FDM/ WF NCHA

Common retention factors

JA/ FDM/ WF/ FR/ DEV/ FDM/ NCHA WF/ NCHA

Source: Developed for this study.

Guidelines for retention policy and practice for software development employees have been developed as follows.

Career stage of Exploration, which is mainly applied to most software engineers, and may also be applied to some junior project leaders. 1. Emphasis of a fair and comparable payment is a good start to attract and retain them within the organization as this is hygiene factor which can prevent individuals from dissatisfaction. In addition, this also plays an important role in satisfying physiological and security needs at the base level of Maslows theory. 2. Offering opportunities for self-development with new technology courses are important as they desire technical competence. A step-by-step approach with small challenging work assignments can make them happy and satisfy their fundamental need to prove themselves. 3. It is also important to help in establishing their internal social network, or providing a mentor who can assist them with an easily accessible resource and confidant.
192

Career stage of Establishment, which may be mainly applied to most project leaders, and may also be applied to some senior software engineers. 1. Emphasis of dual career development may be important in retaining them as some are interested in further managerial positions while some are interested in expanding their technology progression. So, managers may keep communicating with them to see what they value most for their career growth and progression. 2. Preventing project leaders from job hopping may be important especially for senior project leaders who stay with an organization for more than three years. It is deemed that they are capable to work independently and may seek other opportunities such as managerial promotion or better compensation and benefits.

Career stage of Establishment, which may be applied to assistant managers. 1. A considerable level of freedom in their decision-making may be important, as they are capable of working independently and want to be led rather than be managed. They may desire implementing own way of doing things, as they may be pursuing the higher-order needs of motivation factors and self-actualization. 2. Offering flexible work schedules may be important in retaining them as they are pursuing their own autonomy rather than focusing on the base level of physiological and security needs. 3. Emphasis on a fair and comparable remuneration of pay level and pay content may be important in retaining them as they value equal treatment in comparison to their counterparts internally and externally.

193

5.7 Limitations of this Research

There are seven limitations identified by this research. The first limitation is that because of difficulties with the field work arrangement the online survey was conducted through the Human Resource Bank in Taiwan, which was not the original sampling plan that focused on top performing IT companies. This might result in the findings not strongly reflecting the unique retention factor of challenging work. Secondly, the focus on HsinChu Science Park may have restricted the results of findings to those that tend towards more regional aspects as Taiwan has many industrial bases. Thirdly, there could be more retention factors which are highly associated with software development employees decisions to stay, which are not included in this research.

Fourthly, the nature of a quantitative approach is unable to explain the meanings behind the findings that new challenging work was not unique and significant for software development employees. This is a limitation of this research. Fifthly, the survey period is also regarded as a limitation of this research. That is, the survey was running from October to December, 2008, when the impact of the economic downturn started hitting the economy of Taiwan, especially as Taiwans IT industry is heavily reliant on export business. This could distort the intention of the data presented from the software development employees, because layoff activities had been active during that period of time.

Another limitation was that the two double-barrelled questions in the work-life balance factor of the survey were found after the survey was done. This is therefore a limitation of this research and interpretations in the findings chapter have been conservatively interpreted. Finally, the process of translating the survey back into Mandarin was not as precise as it could have been done. This is also regarded as a limitation of this research and
194

interpretations in the findings chapter have been conservatively interpreted as well.

5.8 Implications for Further Research

There are six implications that have been identified for consideration of further research. Firstly, the research has utilised a quantitative, exploratory approach to retention factors for the Taiwanese software development employees, so it does not offer a finely grained analysis of motives or aspirations. Therefore, qualitative techniques such as interviews could be undertaken to understand more deeply the reasons behind these workers intentions to stay within an organisation. For instance, interviews could probe the top values of the software development employees, as well as to understand why new challenging work was not unique retention factor. The geographic location of other industrial parks could be also undertaken to broadly explain the importance of retention guidelines for the entire IT industry in Taiwan. This is because the findings could be different by the fact that the living expenses in northern Taiwan are higher than the living expenses in southern Taiwan, so fair remuneration may be rated higher than expected.

The level of employees focus could also be undertaken to include other departments inside IT companies including hardware employees and marketing and sales employees. This could extend the scope of knowledge and understand how they are different, as some IT companies are successfully led by such departments. The survey could also be repeated within the next few months, as the global financial crisis could have impacted on retention factors. Furthermore, culture as a moderating factor could have included future research, as some employees may still have strong effects in Confucian values. For example, senior members of an organisation may desire to have more respect from others than the junior members within an organisation. Finally, as the results of comparisons between three different types of software development employees were not significant using 5-point
195

scales, 7-point scales could be undertaken when conducting a similar survey next time.

5.9 Conclusion
This research aims to examine retention factors with software development employees decisions to stay in the IT industry in Taiwan. In addition, there are two objectives that have been identified for this research. Firstly, the research identifies what retention factors are important in retaining software development employees, including job appreciation, fair remuneration, freedom in decision making, new challenging work, development, and workplace flexibility. Finally, the research develops a diverse set of tentative retention guidelines, informed by the research results, and appropriate to the IT companies in Taiwan for effective human resource management.

The results of the statistical tests show that the retention of software engineers could be significantly related to two unique retention factors (fair remuneration and development) and four common retention factors (job appreciation, freedom in decision making, workplace flexibility, and new challenging work). The retention of project leaders could be significantly related to one unique retention factor (job appreciation) and five common retention factors (fair remuneration, development, freedom in decision making, workplace flexibility, and new challenging work). The retention of assistant managers could be significantly related to two unique retention factors (freedom in decision making and fair remuneration) and four common retention factors (development, job appreciation, workplace flexibility, and new challenging work). A set of retention guidelines is therefore provided for the different types of software development employees for effective human resource management.

196

The results of two age groups tests show that development could be only one significant factor when comparing those aged between 20 and 29 and those aged between 30 and 44. However, the results of three comparison tests show that there could not be any significant factor when comparing three different types of software development employees.

Implications for theory, policy, and practice are also discussed in detail. Firstly, this research has contributed to the knowledge of retention studies within the Taiwanese context. Secondly, this research has found that each of the different types of software development employees have different unique retention factors and common retention factors. Thirdly, this research has also found that two age groups of software development employees have similarities and differences. Fourthly, this research has closed the knowledge gap by providing a set of useful retention policies for the Taiwanese IT companies. Finally, the guidelines for retention policies are recommended as the appropriate practices to help Taiwanese IT companies to cope with the shortage of software development employees.

197

References
Abell, A., & Oxbrow, N. (2001). Competing with Knowledge: the Information Professional in the Knowledge Management Age. London: Library Association Publishing. ACS. (2008). Dire ICT skills forecast - now's the time for 2020 vision. Retrieved 5th October, 2009, from http://www.acs.org.au/news/110808.htm Acton, T., & Golden, W. (2003). Training the knowledge worker: a descriptive study of training practices in Irish software companies. Journal of European Industrial Training, 27(2), 137-146. Adams, J. S. (1965). Inequity in social exchange. Advances in Experimential Social Psychology, 2, 267-299. Agarwal, R., De, P., & Ferratt, T. W. (2002). Explaining an IT professional's preferred employment duration: emperical tests of a causal model of antecedents. Communications of the ACM, 14-24. Agarwal, R., & Ferratt, T. W. (1999). Coping with Labor Scarcity in IT: Strategies and Practices for Effective Recruitment and Retention. OH: Pinnaflex. Agarwal, R., & Ferratt, T. W. (2000). Retention and the career motives of IT professionals. Communications of the ACM, 158-166. Agarwal, R., & Ferratt, T. W. (2001). Rafting and HR strategy to meet the need for IT workers. Communications of the ACM, 44(7), 59-64. Agarwal, R., & Ferratt, T. W. (2002). Enduring practices for managing IT professionals. Communications of the ACM, 45(9), 73-79. Alexander, H. (2002). Motivational Management: Inspiring Your People for Maximum Performance. New York: AMACOM. Allen, D. G., Shore, L. M., & Griffeth, R. W. (2003). The role of perceived organizational support and supporting human resource practices in the turnover process. Journal of Management, 29(1), 99-118. Allen, N. J., & Meyer, J. P. (1993). Organizational commitment: evidence of career stage effects? Journal of Business Research, 26(1), 49-61. Allison, P., & Pomeroy, E. (2000). How show we "know?" Epistemological concerns in research in experiential education. The Journal of Experiential Education, 23(2), 91-98. Amabile, T. M. (1997). Motivating creativity in organizations: on doing what you love and loving what you do. California Management Review, 40(1), 39-58. Amabile, T. M., & Conti, R. (1999). Changes in the work environment for creativity during downsizing. Academy of Management Journal, 42(6), 630-640. Amar, A. D. (2004). Motivating knowledge workers to innovate: a model integrating motivation dynamics and antecedents. European Journal of Innovation Management, 7(2), 89-101.
198

Amaram, D. (2005). Issues in recruitment and retention for the IT workforce. Journal of American Academy of Business, 6(2), 49-54. Amundson, N. E. (2007). The influence of workplace attraction on recruitment and retention. Journal of Employment Counseling, 44(4), 154-162. Anonymous. (2001). One-third of employees plan to resign within two years. European Business Review, 13(6), 373-376. Anonymous. (2003). Survey highlights skill shortages in ICT. Journal of European Industrial Training, 27(2/3/4), 2001-2002. Anonymous. (2006). How to get & keep the best employees. HR Focus, 83(12), 1-15. Anonymous. (2007). Talent crunch threatens tech industry. CMA Management, 81(5), 12. Anonymous. (2008). Why women quit technology. Computerworld, 42(25), 34-35. Antonopoulos, A. M. (2009). Security job security. Retrieved 5th October, 2009, from http://www.cio.com/article/print/499057 Arellano, N. E. (2007). North American companies "can't find" qualified IT talent. Retrieved 5th October, 2009, from http://www.cio.com/article/108054/North_American_Companies_Can_t_Find_Qua lified_IT_Talent?page=2 Argyrous, G. (1996). Statistics for Social Research. South Melbourne: Macmillan Education Australia. Argyrous, G. (2000). Statistics for Social and Health Research with a Guide to SPSS. London: SAGE Publications. Armstrong-Stassen, M. (2006). Encouraging retirees to return to the workforce. Human Resource Planning, 29(4), 38-44. Bae, J., & Lawler, J. J. (2000). Organizational and HRM strategies in Korea: Impact on firm performance in an emerging economy. Academy of Management Journal, 43, 502-517. Baker, T., & Aldrich, H. E. (1996). 'Prometheus stretches: building identity and cumulative knoweldge in muliemployer careers'. In M. B. Arthur & D. M. Rousseau (Eds.), The Boundaryless Career: A New Employment Principle for a New Organizational Era (pp. 132-149). New York: Oxford University Press. Baldoni, J. (2005). Great Motivation secrets of Great Leaders. New York: McGraw-Hill. Barry, T. (2007). Europe's business brains: poorly led and leaving. Industrial and Commercial Training, 39(2), 113-115. Baruch, Y. (2006). Career development in organisations and beyond: balancing traditional and contemporary viewpoints. Human Resource Management Review, 16(2), 125-138. Bergiel, E. B., Nguyen, V. Q., Clenney, B. F., & Taylor, G. T. (2009). Human resource practices, job embeddedness and intention to quit. Management Research News,
199

32(3), 205-219. Bernhardt, D., & Scoones, D. (1993). Promotion, turnover and preemptive wage offers. The American Economic Review, 83(4), 771-791. Bhal, K. T., & Gulati, N. (2006). Predicting turnover intentions: incorporating the role of organization and work-group level variables. Global Journal of Flexible Systems Management, 7(3/4), 41-50. Bhatnagar, J. (2007). Talent management strategy of employee engagement in Indian ITES employees: key to retain. Employee Relations, 29(6), 640-663. Bird, A. (1996). 'Careers as repositories of knowledge: considerations for boundaryless careers'. In M. B. Arthur & D. M. Rousseau (Eds.), The Boundaryless Career: A New Employment Principle for a New Organizational Era (pp. 150-168). New York: Oxford University Press. Blyton, P., & Jenkins, J. (2007). Key Concepts in Work. Los Angeles: SAGE Publications. Booth, S., & Hammer, K. (2007). Labour turnover in the retail industry. International Journal of Retail & Distribution Management, 35(4), 289-307. Bostock, R. (1996). Rethinking tomorrow's company for the twenty-first century. Career Development International, 1(1), 19-22. Bouma, G. D. (2000). The Research Process (4th Edn). Melbourne: Oxford University Press. Brace, I. (2004). Questionnaire Design: how to plan, structure and write survey material for effective market research. London: Kogan Page. Brandel, M. (2007). Irresistible IT skills. Computerworld, 41(32), 36-38. Branham, F. (2000). Keeping The People Who Keep You In Business: 24 Ways to Hang on to your most Valuable Talent. New York: AMACOM. Budhwar, P. S., Saini, D. S., & Bhatnagar, J. (2005). Women in management in the new economic environment: the case of India. Asia Pacific Business Review, 11(2), 179-193. Burnes, P. T. (2006). Voluntary employee turnover: why IT professionals leave. IT Pro, 46-48. Burrell, G., & Morgan, G. (1979). Sociological Paradigms and Organisational Analysis. Portsmouth. Buyens, D., Van Dijk, H., Dewilde, T., & De Vos, A. (2009). The aging workforce: perceptions of career ending. Journal of Managerial Psychology, 24(2), 102-117. Campion, M. A., & Berger, C. J. (1990). Conceptual integration and empirical test of job design and compensation relationships. Personnel Psychology, 43(3), 525-553. Cappelli, P. (2005). 'Will there really be a labor shortage'. In M. Losey, S. Meisinger & D. Ulrich (Eds.), The Future of Human Resource Management (pp. 5-14). New Jersey: John Wiley & Sons.
200

Carlson, D. S., Derr, C. B., & Wadsworth, L. L. (2003). The effects of internal career orientation on multiple dimensions of work-family conflict. Journal of Family and Economic Issues, 24(1), 99-116. Carlson, D. S., & Rotondo, D. M. (2001). Differences in promotion stress across career stage and orientation. Human Resource Management, 40(2), 99-110. Carrell, M. R., & Dittrich, J. E. (1978). Equity theory: the recent literature, methodological considerations and new directions. Academy of Management Review, 3(2), 202-210. Cascio, W. F. (2002). Responsible Restructuring. San Francisco: Berrett Koehler. Cascio, W. F. (2006). Managing Human Resources: Productivity, Quality of Work Life, Profits (7th Edn). New York: McGraw-Hill Irwin. Castilla, J. I., & Ruiz, O. R. (2008). EFQM model: knowledge governance and competitive advantage. Journal of Intellectual Capital, 9(1), 1469-1930. Cavana, R. Y., Delahaye, B. L., & Sekaran, U. (2001). Applied Business Research: Qualitative and Quantitative Methods. Milton: John Wiley & Sons Australia. CEDEFOP. (2008). Future Skill Needs in Europe. Luxembourg: Office for Official Puiblications of the European Communities. CEPD. (2006). The first stage three-year Sprint Program of the Executive Yuan's Economic Development Vision for 2015. Taipei: Executive Yuan. CEPD. (2007). What future changes will we see in Taiwan's labor market demand and supply trends? What policy responses will be adopted? Retrieved 2nd June, 2009, from http://www.cepd.gov.tw/m1.aspx?sNo=0009110 CEPD. (2008). Taiwan' 2008~2056 Population Projections. Taipei: Executive Yuan. Chatterjee, S. R., & Nankervis, A. R. (2007). Asian Management in Transition: Emerging Themes. New York: Palgrave Macmillan. Chen, H. L., & Huang, Y. S. (2006). Employee stock ownership and corporate R&D expenditures: evidence from Taiwan's information-technology industry. Asia Pacific Journal of Manage, 23, 369-384. Chen, J. C., & Silverthorne, C. (2005). Leadership effectiveness, leadership style and employee readiness. The Leadership and Organization Development Journal, 26(4), 280-288. Chen, S. H., & Liu, M. C. (2003). 'Taiwan's transition from an industrialising economy to a knowledge-based economy'. In S. Masuyama & D. Vandenbrink (Eds.), Towards a Knowledge-based Economy: East Asia's Changing Industrial Geography (pp. 83-111). Singapore: Institute of Southeast Asian Studies. Chen, S. J. (1997). The development of HRM practices in Taiwan. Asian Pacific Business Review, 3(4), 152-169. Chen, T. Y., Chang, P. L., & Yeh, C. W. (2004). A study of career needs, career
201

development programs, job satisfaction and the turnover intentions of R&D personnel Career Development International, 9(4), 424-437. Chen, T. Y., Chang, P. L., & Yeh, C. W. (2006). The effects of career development programs on R&D personnel in Taiwan. Asia Pacific Journal of Human Resources, 44(3), 318-341. Chiu, R. K., Luk, W.-M., & Tang, L. P. (2001). Hong Kong and China: the cash mentality revisited. Compensation and Benefits Review, 33(3), 66-72. Chiu, R. K., Luk, W. M., & Tang, L. P. (2002). Retaining and motivating employees: compensation preferences in Hong Kong and China. Personnel Review, 31(4), 402-431. Chung, E., & Alagaratnam, S. (2001). "Teach ten thousand stars how not to dance": a survey of alternative ontologies in marketing research. Qualitative Marketing Research, 4(4), 224-234. Claes, R., & Heymans, M. (2008). Long-term career development and employability. HR professionals' views on work motivation and retention of older workers: a focus group study. Career Development International, 13(2), 95-111. Clarke, M. (2009). Knowledge is power. Utility Week, 31(3), 25. Cleveland, B. (2005). Seven essential principles of effective communication. Call Center Magazine, 18(8), 30-33. Coakes, S. J., Steed, L., & Dzidic, P. (2006). SPSS version 13.0 for Windows: Analysis without Anguish. Milton: John Wiley & Sons Australia. Coakes, S. J., Steed, L., & Ong, C. (2009). SPSS Version 16.0 for Windows: Analysis without Anguish. Milton: John Wiley & Sons Australia. Coleman, B. (2001). ITS professional development: redefine, reward, retain. Communications of the ACM, 29, 23-26. Colin, C. J., Hanges, P. J., & Locke, E. A. (2004). The relationship of achievement motivation to entrepreneurial behavior: a meta-analysis. Human Performance, 17(1), 95-117. Colman, R. (2004). Telecommuting transitions. CMA Management, 78(5), 23-27. Connolly, P. M., & Connolly, K. G. (2005). Employee Opinion Questionnaire. San Francisco: John Wiley & Sons. Conway, E. (2004). Relating career stage to attitudes towards HR practices and commitment: evidence of interaction effects? European Journal of Work and Organizational Psychology, 13(4), 417-446. Cooper, C. (2006). The challenges of managing the changing nature of workplace stress. Journal of Public Mental Health, 5(4), 6-9. Craig, E., & Hall, D. T. (2005). "The new organizational career: too important to be left to HR". In R. J. Burke & C. L. Cooper (Eds.), Reinventing Human Resource
202

Management: Challenges and New Directions (pp. 115-132). London: Routledge. Crepeau, R., Crook, C., Goslar, M., & McMurtrey, M. (1992). Career anchors of information systems personnel. Journal of Management Information System, 9(2), 145-165. Crook, C., Crepeau, R., & McMurtrey, M. (1991). Utilization of the career anchor/ career orientation construct for management of I/S professionals. Communications of the ACM, 13(2), 12-23. Curtis, S., & Wright, D. (2001). Retaining employees - the fast track to commitment. Management Research News, 24(8), 59-64. D'Art, D., & Turner, T. (2006). New working arrangements: changing the nature of the employment relationship? International Journal of Human Resource Management, 17(3), 523-538. Davenport, T. H. (2005). Thinking for a Living. Boston: Harvard Business School Press. Davenport, T. H., Thomas, R. J., & Cantrell, S. (2002). The mysterious art and science of knowledge-worker performance. MIT Sloan Management Review, 44(1), 23-30. Davila, T., Epstein, M. J., & Shelton, R. (2006). Making Innovation Work: How to Manage It, Measure It, and Profit from It. New Jersey: Pearson Education. Davis, D. (2005). Business Research for Decision Making (6th End). Australia: Thomson South-Western. De Cieri, H., & Kramar, R. (2005). Human Resource Management in Australia 2E Strategy People Performance. NSW: McGraw-Hill. De Cieri, H., Kramar, R., Noe, R. A., Hollenbeck, J. R., Gerhart, B., & Wright, P. M. (2008). Human Resource Management in Australia: Strategy/ People/ Performance (3rd End). Sydney: McGraw-Hill Irwin. De Vaus, D. (2004). Analysing Social Science Data. London: SAGE Publications. De Vaus, D. A. (1995). Surveys in Social Research (4th Edn). St Leonards: Allen & Unwin. DeLong, T. J. (1982). Reexamining the career anchor model. Personnel, 59(3), 50-61. DeMers, A. (2002). Solutions and strategies for IT recruitment and retention: a manager's guide. Public Personnel Management, 31(1), 27-40. Denzin, N. K., & Lincoln, Y. S. (2008). "Paradigms and perspectives in contention". In N. K. Denzin & Y. S. Lincoln (Eds.), The Landscape of Qualitative Research (3rd End) (pp. 245-253). London: SAGE Publications. DGBAS. (2007). 2007 Employment Report with Statistic Analysis. Retrieved 5th June, 2009, from http://www.dgbas.gov.tw/lp.asp?ctNode=3321&CtUnit=951&BaseDSD=7. DGBAS. (2009). 2009 Taiwan's GDP Growth Projection. Retrieved 5th June, 2009, from http://www.dgbas.gov.tw/ct.asp?xItem=24964&ctNode=497. Dobson, S. (2007). Salary surveys challenging but crucial tool in tight market. Canadian
203

HR Reporter, 20(11), 1-2. Doh, J. P., Stumpf, S. A., Tymon, W., & Haid, M. (2008). How to retain talent in India. MIT Sloan Management Review, 50(1), 6-7. Drucker, P. F. (1992). Focusing on the new world economy. Modern Office Technology, 37(11), 8-10. Drucker, P. F. (1993). Post-Capitalist Society. Oxford: Butterworth-Heinemann. Drucker, P. F. (1998). The future that has already happened. The Futurist, 32(8), 16-18. Drucker, P. F. (1999). Knowledge-worker productivity: the biggest challenge. California Management Review, 2, 79-94. Drucker, P. F. (2000). Knowledge work. Executive Excellence, 17(4), 11-12. Drucker, P. F. (2006). Classic: the Man Who Invented Management. Boston: Harvard Business School Publishing Corporation. Drucker, P. F. (2007). Management Challenges for the 21st Century. Oxford: Butterworth-Heinemann. Dychtwald, K., Erickson, T., & Morison, B. (2004). It's time to retire retirement. Harward Business Review, 82(3), 48-58. Dychtwald, K., Erickson, T. J., & Morison, R. (2006). Workforce Crisis. Boston: Harvard Business School Press. Easterby, M., Thorpe, R., & Lowe, A. (1991). Management Research: an introduction. London: Sage. Eyring, A. R. (2008). Executing growth strategies in China: perspective on people and organization. People & Strategy, 16-17. Farh, J. L. (1995). "Human resource management in Taiwan, Republic of China". In L. F. Moore & P. D. Jennings (Eds.), Human Resource Management on the Pacific Rim: Institutions, Practices and Attitudes (pp. 256-294). Berlin: Walter de Gruyter. Feyerherm, A., & Vick, Y. H. (2005). Generation X women in high technology: overcoming gender and generational challenges to succeed in the corporate environment. Career Development International, 10(3), 216-227. Fitzgerald, W. D. (2006). Organisational Behaviour: Core Concepts and Application (1st Edn). Milton: John Wiley & Sons. Fong, S., & Shaffer, M. (2003). The dimensionality and determinants of pay satisfaction: a cross-cultural investigation of a group incentive plan. International Journal of Human Resource Management, 14(4), 559-580. Foote, D. (1998). Non-cash rewards and incentives: secret weapons in the high-tech worker staffing wars. ACA Journal, 7(4), 24-27. Foote, D. (2006). Making your organization the 'it' place to work for IT. Canadian HR Reporter, 19(7), 14-17. Frank, F. D., Finnegan, R. P., & Taylor, C. R. (2004). The race for talent: retaining and
204

engaging workers in the 21st century. Human Resource Planning, 27(3), 12-25. Gagg, P. (2005). Developing skills for the future. Works Management, 58(6), 28-30. Garmise, S. (2006). People and the Competitive Advantage of Place: Building a Workforce for the 21st Century. New York: M. E. Sharpe. Garson, G. D. (2002). Guide to Writing Empirical Papers, Theses, and Dissertations. New York: Marcel Dekker, Inc. Gee, C., & Burke, M. E. (2001). Realising potential: the new motivation game. Managemnt Decision, 39(2), 131. Glebbeck, A. C., & Bax, E. H. (2004). Is high employee turnover really harmful? An empirical test using company records. Academy of Management Journal, 47, 277-286. Gopinath, D. (2007). China and India are failing to produce top quality graduates. The Business, 1-3. Gorard, S. (2003). Quantitative Methods in Social Science. New York: Continuum. Gordon, E. E. (2009a). The 2010 surprise; global talent shortages, wage inflation loom as employers will soon be competing for skilled workers in a shrinking global market. Here's how to prepare. Employee Benefit News, 23(9), 14-16. Gordon, E. E. (2009b). The global talent crisis: contrary to popular opinion, there are plenty of open jobs. What's missing are cadidates with skills. The Futurist, 43(5), 34-39. Gostick, A., & Elton, C. (2001). Show me the rewards. Canadian HR Reporter, 14(5), 7-10. Greenhaus, J. H., & Beutell, N. J. (1985). Sources of conflict between work and family roles. Academy of Management Review, 10(1), 76-88. Gregory, R. J. (2000). Psychological Testing: History, Principles, and Applications (3rd Edn). Boston: Allyn and Bacon. Griffeth, R. W., Hom, P. W., & Gaertner, S. (2000). A meta-analysis of antecedents and correlates of employee turnover: update, moderator tests, and research implications for the millennium. Journal of Management, 26(3), 463-488. Guba, E. G., & Lincoln, Y. S. (1991). "What is the constructivist paradigm?". In D. S. Anderson & B. Liddle (Eds.), Knowledge for Policy: Improving Education through Research. London: Falmer Press. Guba, E. G., & Lincoln, Y. S. (1994). "Competing paradigm in qualitative research". In N. K. Denzin & Y. S. Lincoln (Eds.), Handbook of Qualitative Research (pp. 105-117). London: Sage. Guidice, R. M., Heames, J. T., & Wang, S. (2009). The indirect relationship between organizational-level knowledge worker turnover and innovation: an integrated application of related literature. The Learning Organization, 16(2), 143-167.
205

Haar, J. M., & Spell, C. S. (2009). How does distributive justice affect work attitudes? The moderating effects of autonomy. The International Journal of Human Resource Management, 20(8), 1827-1842. Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate Data Analysis: A Global Perspective (7th End). New Jersey: Pearson Education. Hair, J. R., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivariate Data Analysis (5th Edn). New Jersey: Prentice-Hall. Hammer, M., Leonard, D., & Davenport, T. (2004). Why don't we know more about knowledge? MIT Sloan Management Review, 45(4), 14-18. Hansen, F. (2001). Currents in compensation and benefits. Compensation and Benefits Review, 33(3), 6-12. Hansen, F. (2007). Pushing money toward the top. Workforce Management, 86(19), 44. Hansen, F., Smith, M., & Hansen, R. B. (2002). Rewards and recognition in employee motivation. Compensation and Benefits Review, 34(5), 64-72. Harman, W. S., Lee, T. W., Mitchell, T. R., Felps, W., & Owens, B. P. (2007). The psychology of voluntary employee turnover. Current Directions in Psychological Science, 16(1), 51-54. Hatala, J. P. (2006). Social network analysis in human resource development: a new methodology. Human Resource Development Review, 5(1), 45-71. Hawser, A. (2008). China faces growing skills shortage. Global Finance, 22(1), 4. Hayes, F. (2003). Avoid burnout. Computerworld, 37(12), 58. HCSP. (2009). Advantages of Taiwan. Retrieved 5th March, 2009, from http://investintaiwan.nat.gov.tw/en/env/taiwan.html Hess, N., & Jepsen, D. M. (2009). Career stage and generational differences in psychological contracts. Career Development International, 14(3), 261-283. Hobday, M. (1994). Export-led technology development in the four dragons: the case of electronics. Development and Change, 25(2), 333-361. Hoch, D. J., Roeding, C. R., Purkert, G., & Lindner, S. K. (2000). Secrets of Software Success: Management Insights from 100 Software Firms around the World. USA: Harvard Business School Press. Hoffman, T. (2009). The 9 hottest IT skills for 2009. Retrieved 5th October, 2009, from http://www.cio.com/article/print/475372 Holland, P., Sheehan, C., Donohue, R., & Pyman, A. (2007). Contemporary Issues and Challenges in HRM. Prahran: Tilde University Press. Holland, P. J., Hecker, R., & Steen, J. (2002). Human resource strategies and organisational structures for managing gold-collar workers. Journal of European Industrial Training, 26(2-4), 72-80. Holt, G. D., Love, P. E., & Nesan, L. J. (2000). Employee empowerment in construction:
206

an implementation model for process improvement. Team Performance Management: An International Journal, 6(3/4), 47-51. Holtom, B. C., & Inderrieden, E. J. (2006). Integrating the unfolding model and job embeddedness model to better understand voluntary turnover. Journal of Managerial Issues, XVIII(4), 435-452. Holtom, B. C., Mitchell, T. R., & Lee, T. W. (2006). Increasing human and social capital by applying job embeddedness theory. Organizational Dynamics, 35(4), 316-331. Holtom, B. C., Mitchell, T. R., Lee, T. W., & Inderrieden, E. J. (2005). Shocks as causes of turnover: what they are and how organizations can manage them. Human Resource Management, 44(3), 337-352. Holtom, B. C., & O'Neill, B. (2004). Job embeddedness: a theoretical foundation for developing a comprehensive plan for retaining health care employees. Journal of Nursing Administration, 34, 216-217. Hom, P. W., & Griffeth, R. W. (1995). Employee Turnover. Cincinnati: South/Western. Horwitz, F. M., Heng, C. T., & Quazi, H. A. (2003). Finder, keepers? Attracting, motivating and retaining knowledge workers. Human Resource Management Journal, 13(4), 23-44. Hsu, M. K., Chen, H. G., Jiang, J. J., & Klein, G. (2003). Career satisfaction for managerial and technical anchored IS personnel in later career stages. Database for Advances in Information Systems, 34(4), 64-71. Hu, M. C., & Mathews, J. (2005). National innovative capacity in East Asia. Research Policy, 34(9), 1322-1349. Hu, M. C., Zheng, C., & Lamond, D. (2007). Recruitment and retention of ICT skills among MNCs in Taiwan. Chinese Management Studies, 1(2), 78-92. Huang, E. Y., & Lin, S. C. (2006). How R&D management practice affects innovation performance: an investigation of the high-tech industry in Taiwan. Industrial Management & Data Systems, 106(7), 966-996. Huang, I. C., Lin, H. C., & Chuang, C. H. (2006). Constructing factors related to worker retention. International Journal of Manpower, 27(5), 491-508. Huang, T. C. (2001). "Human resource management in Taiwan". In P. S. Budhwar & Y. A. Debrah (Eds.), Human Resource Management in Developing Countries (pp. 56-74). New York: Routledge. Hytter, A. (2007). Retention strategies in France and Sweden. Irish Journal of Management, 28(1), 59-79. Igbaria, M., Greenhaus, J. H., & Parasuraman, S. (1991). Career orientations of MIS employees: an emperical analysis. MIS Quarterly, 15(2), 151-169. Igbaria, M., & McCloskey, D. W. (1996). Career orientation of MIS employees in Taiwan. Computer Personnel, 17(2), 3-24.
207

Jakobson, L. (2008). Met life rethinks recognition. Incentive, 182(9), 20-30. Jaskyte, K., & Kisieliene, A. (2006). Determinants of employee creativity: a survey of lithuanian nonprofit organizations. International Journal of Voluntary and Nonprofit Organizations, 17, 133-141. Jones, B., & Miller, B. (2007). Innovation Diffusion in the New Economy. New York: Routledge. Jouhette, S., & Romans, F. (2006). Eu labor force survey principal results 2005. Retrieved 5th June, 2009, from http://eeurostat.ec.europa.eu/cache/ITY_OFFPUB/KS-NK-06-013/EN/KS-NK-06013-EN.PDF. Jowitt, T. (2009). Enterprises still face 'significant' IT skills shortage. Retrieved 5th October, 2009, from http://www.cio.com/article/print/503158 Kaliprasad, M. (2006). The human factor I: attracting retaining, and motivation capable people. Cost Engineering, 48(6), 20-27. Kavanagh, M. (2008). Skills shortage still applies in technical subjects. Financial Times, 5-7. Kelley, L. L., Blackman, D. A., & Hurst, J. P. (2007). An exploration of the relationship between learning organisations and the retention of knowledge workers. The Learning Organisation, 14(3), 204-221. Khatri, N., Chong, T. F., & Budhwar, P. (2001). Explaining employee turnover in an Asian context. Human Resource Management Journal, 11(1), 54-74. Kochanski, J., & Ledford, G. (2001). "How to keep me" - Retaining technical professionals. Research and Technology Management, 31-38. Kochanski, J., Mastropolo, P., & Ledford, G. (2003). People solutions for R&D. Research and Technology Management, 59-61. Koh, S., Lee, S., Yen, D. C., & Havelka, D. (2004). The relationship between information technology professionals' skill requirements and career stage in the e-commerce era: an empirical study. Journal of Global Information Management, 12(1), 68-82. Korane, K. (2009). How to cope with technical staff shortages. Machine Design, 81(7), 52. Kreiner, K. (2002). Tacit knowledge management: the role of artifacts. Journal of Knowledge Management, 6(2), 112-123. Kuijpers, M. A. C. T., Schyns, B., & Scheerens, J. (2006). Career competencies for career success. The Career Development Quarterly, 55(2), 168-178. Kuo, Y. F., & Chen, L. S. (2004). Individual demographics differences and job satisfaction among information technology personnel: an empirical study in Taiwan. International Journal of Management, 21(2), 221-231. Lacity, M. C., Iyer, V. V., & Rudramuniyaiah, P. S. (2008). Turnover intentions of Indian IS professionals. Springer, 10, 225-241.
208

Lamb, M., & Sutherland, M. (2010). The components of career capital for knowledge workers in the global economy. The International Journal of Human Resource Management, 21(3), 295-312. Lazear, E. P. (1986). Raids and offer matching. Research in Labor Economics, 8, 141-165. Lee, C. B. (2002). Career goals and career management strategy among information technology professionals. Career Development International, 7(1), 6-13. Lee, N., & Lings, I. (2008). Doing Business Research: A Guide to Theory and Practice. London: SAGE Publications. Lee, T. H., Gerhart, B., Weller, I., & Trevor, C. O. (2008). Understanding voluntary turnover: path-specific job satisfaction effects and the importance of unsolicited job offers. Academy of Management Journal, 51, 651-671. Lee, T. W., & Maurer, S. D. (1997). The retention of knowledge workers with the unfolding model of voluntary turnover. Human Resource Management Review, 7(3), 247-275. Lee, T. W., & Mitchell, T. R. (1994). An alternative approach: the unfolding model of voluntary employee turnover. Academy of Management Journal, 19, 51-58. Lee, T. W., Mitchell, T. R., Holtom, B. C., McDaniel, L. S., & Hill, J. W. (1999). The unfolding model of voluntary turnover: a replication and extension. Academy of Management Journal, 42(4), 450-461. Lee, T. W., Mitchell, T. R., Sablynski, C. J., Burton, J. P., & Holtom, B. C. (2004). The effects of job embeddedness on organizational citizenship, job performance, volitional absences, and voluntary turnover. Academy of Management Journal, 47(5), 711-722. Leedy, P. D., & Ormrod, J. E. (2005). Practical Research: Planning and Design (8th Edn). New Jersey: Pearson Education International. Leung, F. L. (1995). Overseas Chinese Management: Myths and Realities. East Asian Executive Reports, 17(2), 6-13. Lewin, K. (1951). Field Theory in Social Science. New York: Harper. Liz, L. K., Blackman, D. A., & Hurst, J. P. (2007). An exploration of the the relationship between learning organizations and the retention of knowledge workers. The Learning Organization, 14(3), 204-221. Lock, G. E. (2003). Living valuing and sharing- a case study of retaining IT professionals in the British Columbia public service. Career Development International 8(3), 152-158. Loomis, L. (2006). The aging workforce is your company ready ? Accounting & Tax Periodicals, 46-48. Luftman, J. (2008). Companies can't afford to lose their best IT pros. Computerworld, 42(37), 37.
209

Luftman, J., & Kempaiah, R. M. (2007). The IS organization of the future: the IT talent challenging. Information Systems Management, 24(2), 129-138. Lui, S. S., Lau, C. M., & Ngo, H. Y. (2004). Global convergence, human resources best practices, and firm performance: a paradox. Management International Review, 44(2), 67-86. Maertz, C. P., & Campion, M. A. (1998). 25 years of voluntary turnover research: a review and critique. International Review of Industrial and Organizational Psychology, 13(49-81). Mah, M. (2009). When talk fails, so does software. Software Quality Professional, 11(4), 47-48. Mahatanankoon, P. (2007). The effects of post-educational professional development activities on promotion and career satisfaction of IT professionals. Communications of the ACM, 9-14. Mallol, C. M., Holtom, B. C., & Lee, T. W. (2007). Job embeddedness in a culturally diverse environment. Journal of Business and Psychology, 22(1), 35-44. Manning, M., & Munro, D. (2007). The Survey Researcher's SPSS Cookbook (2nd Edn). Sydney: Pearson Education Australia. Martin, G. (2008). Managing talent across generations. China Staff, 14(2), 28-34. Masuyama, S., & Vandenbrink, D. (2003). 'The new industrial geography of east Asia under the knowledge-based economy'. In S. Masuyama & D. Vandenbrink (Eds.), Towards a Knowledge-based Economy: East Asia's Changing Industrial Geography (pp. 3-39). Singapore: Institute of Southeast Asian Studies. Mata, F., Fuerst, W., & Barney, J. (1995). Information technology and sustained competitive advantage: a resource based analysis. MIS Quarterly, 19(4), 487-505. Mayfield, J., & Mayfield, M. (2008). The creative environment's influence on intent to turnover: a structural equation model and analysis. Management Research News, 31(1), 41-56. McGee, M. K. (2005). Retention Tension. InformationWeek, 49-56. McMurray, D. (2006). Qualitative Research Methods (3rd Edn). New South Wales: Southern Cross University. McMurtrey, M. E., Grover, V., Teng, J. T. C., & Lightner, N. J. (2002). Job satisfaction of information technology workers: the impact of career orientation and task automation in a CASE environment. Journal of Management Information System, 19(2), 273-302. McQuade, E., Sjoer, E., Fabian, P., Nascimento, J. C., & Schroeder, S. (2007). Will you miss me when I'm gone? A study of the potential loss of company knowledge and expertise as employees retire. Journal of European Industrial Training, 31(9), 758-768.
210

McVey, C. C., & McVey, R. T. (2005). Responding to today's work force: attracting, retaining and developing the new generation of workers. Correlations Today, 67(7), 80-109. Mendonsa, R. (1998). Keeping who you want to keep: retaining the best people. SuperVision, 59(1), 10-12. Messmer, M. (2006). Four keys to improved staff retention. Strategic Finance, 88(4), 13-14. Mgaya, K. V., Uzoka, F. M. E., Kitindi, E. G., & Shemi, A. P. (2009). Examining career orietations of information systems personnel in an emgerging economy context. Paper presented at the SIGMIS-CPR'09. Milkovich, G. T., & Newman, J. M. (2005). Compensation (8th Edn). Boston: McGraw-Hill Irwin. Milliman, J., Czaplewski, A. J., & Ferguson, J. (2003). Workplace spirituality and employee work attitudes: an exploratory empirical assessment. Journal of Organistional Change Management, 16(4), 426-447. Mitchell, T. R., Holtom, B. C., & Lee, T. W. (2001). How to keep your best employees: developing an effective retention policy. Academy of Management Executive, 15(4), 96-109. Mitchell, T. R., Holtom, B. C., Lee, T. W., & Erez, M. (2001). Why people stay: using job embeddedness to predict voluntary turnover. Academy of Management Journal, 44(6), 1102-1121. Moderi, M., & Foote, D. (2005). Are IT retention efforts working? Optimize, 4(12), 23-24. MOEA. (2008). Taiwan Investment Guide. Retrieved 25th September, 2009, from http://www.dois.moea.gov.tw/tig/english/page.asp?class_id=1&page=1 Moore, J. E. (2000). One road to turnover: an examination of work exhaustion in technology professionals. MIS Quarterly, 24(1), 141-168. Morgan, G. (1980). Paradigm, metaphors, and puzzle solving in organization theory. Administrative Science Quarterly, p.605-627. Morrell, K., Load-Clarke, J., & Wilkinson, A. (2004). The role of shocks in employee turnover. British Journal of Management, 15, 335-349. Morrell, K., Loan-Clarke, J., Arnold, J., & Wilkinson, A. (2008). Mapping the decision to quit: a refinement and test of the unfolding model of voluntary turnover. Applied Psychology: An International Review, 57(1), 128-150. Mosley, V., & Hurley, M. (1999). IT skill retention. Information Management & Computer Security, 7(3), 129-132. Nair, N., & Vohra, N. (2010). An exploration of factors predicting work alienation of knowledge workers. Management Decision, 48(4), 600-615. Nankervis, A., Compton, B., & Baird, M. (2005). Human Resource Management:
211

Strategies and Processes (5th Edn). Southbank Victoria: Nelson Australia. Nelson, R. R., & Todd, P. A. (2004). Peopleware: the hiring and retention of IT personnel. 1-17. Nelson, R. R., & Winter, S. J. (1982). An Evolutionary Theory of Economic Change. Cambridge: Harvard University Press. Netemeyer, R. G., Bearden, W. O., & Sharma, S. (2003). Scaling Procedures: Issues and Applications. Thousand Oaks: Sage Publications. Neuman, W. L. (2004). Basics of Social Research: Qualitative and Quantitative Approaches. Sydney: Pearson Education. Neuman, W. L. (2006). Social Research Methods: qualitative and quantitative approaches (6th Edn). Sydney: Pearson International Edition. Ng, J. M., & Li, K. X. (2003). Implications of ICT for knowledge management in globalization. Information Management & Computer Security, 11(4), 167-174. Nicholas, A. J., & Guzman, I. R. (2009). IS teleworking for the millennials? Communications of the ACM, 197-208. Niles, S. P., & Harris-Bowlsbey, J. (2005). Career Development Interventions in the 21st Century (2nd End). Ohio: Pearson Prentice Hall. OECD. (2008). OECD Information Technology Outlook 2008 Highlights. Paris: Organization for Economic Co-operation and Development. Pare, G., Tremblay, M., & Latrick, L. (2001). Workforce retention: what do IT employees really want? Communications of the ACM, 1-7. Perry, P. M. (2001). Holding your top talent. Research Technology Management, 44(3), 26-30. Petroni, A. (2000). Strategic career development for R&D staff: a field research. Team Performance Management: An International Journal, 6(3/4), 52-61. Policastro, E. F. (2007). Untapped knowledge, hidden sweet spot. InTech, 54(11), 18-22. Punia, B. K., & Sharma, P. (2008). Employees' perspective on human resource procurement practices as a retention tool in Indian IT sector. The Journal of Business Perspective, 12(4), 57-69. Quesenberry, J. L. (2006). Career anchors and organizational culture: a study of women in the IT workforce. Communications of the ACM, 342-344. Quesenberry, J. L., & Trauth, E. M. (2007). What do women want?: an investigation of career anchors among women in the IT workforce. Communications of the ACM, 122-127. Ramakrishna, H., & Potosky, D. (2001/2002). Structural shifts in career anchors of information systems personnel: a preliminary empirical analysis. The Journal of Computer Information Systems, 42(2), 83-89. Ramlall, S. (2004). A review of employee motivation theories and their implications for
212

employee retention within organizations. Journal of American Academy of Business, 5(1/2), 52-63. Reiche, B. S. (2009). To quit or not to quit: organizational determinants of voluntary turnover in MNC subsidiaries in Singapore. The International Journal of Human Resource Management, 20(6), 1362-1380. Rhee, M. (2004). Network updating and exploratory learning environment. Journal of Management Studies, 41(6), 932-949. Rollag, K., Parise, S., & Cross, R. (2005). Getting new hires up to speed quickly. MIT Sloan Management Review, 46(2), 35-41. Ross, J., Beath, C., & Goodhue, D. (1996). Develop long-term competitiveness through IT assets. Sloan Management Review, 38(1), 31-42. Rowley, C., Benson, J., & Warner, M. (2004). Towards an Asian model of human resource management? A comparative analysis of China, Japan and South Korea. The International Journal of Human Resource Management, 15(4), 917-933. Ruane, J. M. (2005). Essentials of Research Methods: A Guide to Socail Research. Malden: Blackwell Publishing. Rubens, K. (2008). The future of work-cubed. People & Strategy, 31(4), 13-14. Rumpel, S., & Medcof, J. W. (2006). Total rewards: good fit for Tech Workers. Research Technology Management, 49(5), 27-35. Rynes, S., Colbert, A., & Brown, K. G. (2002). HR professionals' beliefs about effective human resource practices: correspondence between research and practice. Human Resource Management, 41(2), 149-174. Rynes, S., Gerhart, B., & Minette, K. A. (2004). The importance of pay in employee motivation: discrepancies between what people say and what they do. Human Resource Management, 43(4), 381-394. Salopek, J. (2008). Retention buzz. T + D, 62(1), 23-25. Sattar, S. (2007). Young professionals. Civil Engineering, 15(1), 16-17. Schein, E. H. (1978). Career dynamics: matching individual needs and organizational needs. MA: Addison-Wesley. Schein, E. H. (1987). "Individuals and careers". In J. W. Lorsch (Ed.), Handbook of Organizational Behavior. New Jersey: Prentice-Hall. Schein, E. H. (1996). Career anchors revised: implications for career development in the 21st century. The Academy of Management Executive, 10(4), 80-88. Schramm, J., & Burke, M. E. (2004). SHRM 2004-2005 Workplace Forecast: A Strategic Outlook. Alexandria: Society for Human Resource Managemnet. Schulz, E., Camp, R. R., & Waltman, J. L. (2008). Incremental effectiveness of two key IT recruitment methods. Journal of Managerial Issues, 20(2), 195-213. Sekaran, U. (2000). Research Methods for Business: A Skill Building Approach (3rd Edn).
213

New York: John Wiley & Sons. Sekaran, U. (2003). Research Methods for Business: A Skill Building Approach (4th Edn). New York: John Wiley & Sons. Shah, A., Sterrett, C., Chesser, J., & Wilmore, J. (2001). Meeting the need for employee development in the 21st century. S.A.M. Advanced Management Journal, 66(2), 22-28. Shaw, J. D., Delery, J. E., Jenkins, G. D., & Gutpa, N. (1998). An organisation-level analysis of voluntary and involuntary turnover. Academy of Management Journal, 41(5), 511-525. Sigler, K. (1999). Challenges of employee retention. Management Research News, 22(10), 1-5. Slagter, F. (2007). Knowledge management among the older workforce. Journal of Knowledge Management, 11(4), 82-96. Smith, A. D., & Rupp, W. T. (2002). Communication and loyalty among knowledge workers: a resource of the firm theory view. Journal of Knowledge Management, 6(3), 250-261. Somaya, D., & Williamson, I. O. (2008). Rethinking the "War fo Talent". MIT Sloan Management Review, 49(4), 29-34. Stinglhamber, F., & Vandenberghe, C. (2003). Organizations and supervisors as sources of support and targets of commitment: a longitudinal study. Journal of Organizational Behavior, 24(3), 251-270. Stone, R. J. (2005). Human Resource Management. Sydney: Wiley. Sue, V. M., & Ritter, L. A. (2007). Conducting Online SURVEY. Los Angeles: Sage Publications. Sullivan, S. E., Carden, W. A., & Martin, D. F. (1998). Careers in the next millennium: directions for future research. Human Resource Management Review, 8(2), 165-185. Sumner, M. (2008). An investigation of work family conflict among IT professionals. Communications of the ACM, 127-132. Sumner, M., & Yager, S. (2004). Career orientation of IT personnel. Communications of the ACM, 92-96. Sumner, M., Yager, S., & Franke, D. (2005). Career orientation and organizational commitment of IT personnel. Communications of the ACM, 75-80. Super, D. E. (1990). "A life-span, life-space approach to career development". In D. Brown & L. Brook (Eds.), Career Choice and Development: Applying Contemporary Theories to Practice (2nd End) (pp. 197-261). San Francisco: Jossey-Bass. Sverko, B., & Vizek-Vidovic, V. (1995). 'Studies on the meaning of work: approaches, models, and some findings'. In D. E. Super & B. Sverko (Eds.), Life Roles, Values,
214

and Careers: International Findings of the Work Importance Study (pp. 3-21). San Francisco: Jossey-Bass. Sweeney, J. (2008). Career paths for Gen X, Gen Y. Canadian HR Reporter, 21(12), 16. Tabachnick, B. G., & Fidell, L. S. (2001). Using Multivariate Statistics (4th Edn). Sydney: Allyn and Bacon. Tanova, C., & Holtom, B. C. (2008). Using job embeddedness factors to explain voluntary turnover in four European countries. The International Journal of Human Resource Management, 19(9), 1553-1568. Taplin, I. M., & Winterton, J. (2007). The importance of management style in labour retention. International Journal of Sociology and Social Policy, 27(1), 5-18. Tapscott, D. (2009). Grown Up Digital: How the Net Generation is Changing your World. New York: McGraw-Hill. Taylor, L. (2008). Occupational Skill Shortage Analysis - Tasmania Phase One Report. Thapanachai, S. (2006). Research exposes skills gap: cape group: manage and retain your talent. Knight Ridder Tribune Business News, 1-3. The Economist. (2007). Capturing talent - Asia's skills shortage. The Economist, 58-64. Thomson, R. (2007). Training is key to soothing IT workers' three-year itch. Computer Weekly, 31. Ticehurst, G. W., & Veal, A. J. (2000). Business Research Methods: A Managerial Approach. French Forest: Pearson Education Pty Limited. Ticehurt, G. W., & Veal, A. J. (2000). Business Research Methods: A Managerial Approach. French Forest: Pearson Education Pty Limited. Trevor, C. O., Gerhart, B., & Boudreau, J. W. (1997). Voluntary turnover and job performance: curvilinearity and the moderating influences of salary growth and promotions. Journal of Applied Psychology, 82, 44-61. Trinczek, R. (2006). 'Work-life balance and flexible work hours - the German experience'. In P. Blyton, B. Blunsdon, K. Reed & A. Dastmalchian (Eds.), Work-Life Integration: International Perspectives on the Balancing of Multiple Roles (pp. 113-134). Basingstroke: Palgrave Macmillan. Trower, C. A. (2008). "Young faculty and their impact on academe". In D. E. Heller & M. B. d' Ambrosio (Eds.), Generational Shockwaves and the Implications for Higher Education (pp. 41-58). Cheltenham: Edward Elgar. Tynan, D. (2006). Executive order: attractive and retain top talent. InfoWorld, 28(38), 40-42. Vavra, B. (2009). Skills gap. Plant Engineering, 63(2), 21-25. Veal, A. J. (2005). Business Research Methods: A Managerial Approach (2nd Edn). South Melbourne: Pearson Addison Wesley. Walker, B. C. (2007). The race for talent. Leadership Excellence, 24(3), 8.
215

Walker, J. W., & LaRocco, J. M. (2002). Perspectives: talent pools: the best and the rest. Human Resource Planning, 25(3), 12-14. Wang, C. (2007). Building a theory of IT compensation. Communications of the ACM, 221-223. Waterman, R. H., Waterman, J. A., & Collard, B. A. (1994). Toward a career resilient workforce. Harvard Business Review, 72(4), 87-95. Wenger, E. C., & Snyder, W. M. (2000). Communities of practice: the organizational frontier. Harvard Business Review, 139-145. Whitaker, A. (1999). What causes IT workers to leave. Management Review, 88(9), 8. Wickramasinghe, V., & Jayabandu, S. (2007). Towards workplace flexibility: flexitime arrangements in Sri Lanka. Employee Relations, 29(6), 554-575. Williams, D. (2007). How to find and keep skilled workers. Area Development Site and Facility Planning, 42(1), 22-26. Williams, M. L., McDaniel, M. A., & Nguyen, N. T. (2006). A meta-analysis of the antecedents and consequences of pay level satisfaction. Journal of Applied Psychology, 91(2), 392-413. Wirtz, J., Heracleous, L., & Pangarkar, N. (2008). Managing human resources for service excellence and cost effectiveness at Singapore Airlines. Managing Service Quality, 18(1), 4-19. Woodruffe, C. (1999). Winning the Talent War: a strategic to attracting, developing and retaining the best people. Winning John Wiley & Sons. Wu, P. C. (2004). "HRM in Taiwan". In P. S. Budhwar (Ed.), Managing Human Resources in Asia-Pacific (pp. 93-112). New York: Routledge. Wynekoop, J. L., & Walz, D., B. (2000). Investigating traits of top performing software developers. Information Technology & People, 13(3), 186-195. Yarnall, J. (2008). Strategic Career Management Development Your Talent. Oxford: Butterworth-Heinemann. Yeh, R. S. (1991). Management practices of Taiwanese firms: as compared to those of American and Japanese subsidiaries in Taiwan. Asia Pacific Journal of Management, 8(1), 1-14. Yigitcanlar, T., Baum, S., & Horton, S. (2007). Attracting and retaining knowledge workers in knowledge cities. Journal of Knowledge Management, 11(5), 6-17. Young, L. (2008). Disconnect between what IT wants, HR offers. Canadian HR Reporter, 21(16), 7-8. Young, M. B. (2006). We've studied older workers to death. How come we know so little about them? The Conference Board Review, 48-54. Zeffane, R., & McLoughlin, D. (2006). Cooperation and stress: exploring the differential impact of job satisfaction, communication and culture. Management Research
216

News, 29(10), 618-631. Zetlin, M. (2002). Put balance in work and home life. Computerworld, 36(19), 40-41. Zhang, Y. (2008). The Relatonship between motivation factors and the retention of key management and professional technical staff in government-run construction enterprises in China. Unpublished DBA Thesis, Southern Cross University, Tweed Heads. Zheng, C., Soosay, C., & Hyland, P. (2008). Manufacturing to Asia: who will win the emerging battle for talent between Dragons and Tigers? Journal of Manufacturing Technology Management, 19(1), 52-72. Zikmund, W. G. (2003). Business Research Methods (7th Edn). Ohio: Thomson South-Western.

217

Appendix A

) Information Sheet for HR Managers ( )

SOUTHERN CROSS UNIVERSITY ( Name of Project: The Retention of Software Development Employees in the IT Industry in Taiwan

:
My name is Chin-Yao Tseng and I am conducting research on the retention of software development employees in the IT Industry in Taiwan. I am a Doctor of Business Administration candidate at Southern Cross University in Australia. As part of the course, all DBA candidates are required to take a course that includes a research project that examines an issue in the workplace.

,. . ,,.
I used to work in the HsinChu Science Park and I saw many IT professionals had left their companies due to a variety of reasons. Thus, this study is to focus on examining factors that determine employees decision to stay in the organizations. Five major factors have been identified that include financial compensation, career development, self-achievement, environmental factors, and job satisfaction. It is hoped that the research can help IT companies address issues regarding the retention of their software development employees in the Taiwanese IT Industry.

, IT . , IT . ,, , , , . , .
I would appreciate your cooperation in this research to permit me to conduct the survey in your company with your staff. The benchmark of choosing your company as part of my survey sample is that your company has performed with outstanding results in the stock market. This shows that your company really values knowledge workers, as this study focuses on the full-time employed software development employees. It is voluntary and they can withdraw from this study or discontinue participation at any time.

,. . , . .
The questionnaire survey is an online, web-based design so it is anonymous and the confidentiality of the data will be highly protected by the researcher. The questionnaire will be in Mandarin. It will take appropriately 10 -15 minutes for your staff to complete the questionnaire through the Internet.

,. . 10 15 .
The combined results from software development employees in all of the organizations will be summarized to you and in appreciation of allowing me to conduct research in your organization. No individual or particular organization will be identifiable in these combined

218

results. The research findings may also be submitted for publication.

, . . .
It is necessary that you sign a consent form to enable your organization to participate. All signed consent form will be held in safe storage at the University for a period of five years before being destroyed.

, . , .
In addition, could you please provide a group email address of your software development employees in the returned consent form so that I can notify them regarding participation and how to access the survey. I would also appreciate you if you could introduce some possible IT companies that might be interested in participating in this study. However, if you have any questions with regard to this study, please communicate to us at any time:

,, . . , ,.
Researcher

Chin-Yao Tseng

Supervisor

Associate Professor Michelle Wallace Graduate College of Management Southern Cross University Tweed Gold Coast Campus PO Box 42 Tweed Heads 2485 Email: michelle.wallace@scu.edu.au

Graduate College of Management Southern Cross University Tweed Gold Coast Campus PO Box 42 Tweed Heads 2485 Email: c.tseng.10@scu.edu.au

The ethical aspects of this study have been approved by the Southern Cross University Human Research Ethics Committee. The Approval Number is ECN-08-107. Or if you have any problems about any ethical aspect of your participation in this research, you may contact the Committee through the Ethics Complaints Officer:

. ECN-08-107. ,.
Ms Sue Kelly Ethics Complaints Officer and Secretary HREC Southern Cross University, PO Box 157, Lismore, NSW, 2480 Telephone: 0011-61-2-6626-9139 or fax 0011-61-2-6626-9145 Email: sue.kelly@scu.edu.au

219

Appendix B

) Informed Consent Form for HR Managers ( )

SOUTHERN CROSS UNIVERSITY ( Name of Project: The Retention of Software Development Employees in the IT Industry in Taiwan

: Researcher
Mr. Chin-Yao Tseng Email: c.tseng.10@scu.edu.au Or you may contact the Ethics Complaints Officer:

Supervisor

Associate Professor Michelle Wallace Email: michelle.wallace@scu.edu.au

Ms Sue Kelly, Ethics Complaints Officer and Secretary, Email: sue.kelly@scu.edu.au, Phone: 0011-2-6626-9139

I have been provided with information at my level of comprehension about the purpose, demands, and possible outcomes of this research (including any form of publication of results).

,,

I agree that my organization can participate in the above research project. I have read and understand the details contained in the

. . .
Information Sheet. I have had the opportunity to ask questions about the study and I am satisfied with the answers received. I understand that participation in this research by my staff will be anonymous and confidential and that participants can withdraw or discontinue participation at any time.

I understand that all information gathered in this research is confidential. It is kept securely and confidentially for 5 years, at the University.

I have read the information above and agree to participate in this study. I am over the age of 18 years.

. . RD group email address (RD ): __________________________________________________ Total amount of your software development employees (): ________________________________________ Other possible IT companies (): ___________________________________ Your Name ():__________________________________________________________________ Your Position ():_______________________________________________________________ Your Signature ():____________________________Date ():_________________________
220

Appendix C
Research Questionnaire (

Section I: Please select the number that best represents your opinion about your work experiences at your organization according to the following statements. Please circle how strongly you agree or disagree using below scale with each of the following statements:

: . , 1 , 2 , 3 , 4 , 5 .
1= Strongly Disagree, 2= Disagree, 3= Neutral, 4= Agree, 5= Strongly Agree SD 1. This organization pays me fairly for the work I do. 1 D 2 U 3 A 4 SA 5

.
2. My total pay is adequate compared to others in this organization. 1 2 3 4 5

,,.
3. My total pay is adequate compared to similar work elsewhere. 1 2 3 4 5

, .
4. People at this organization are paid fairly according to their job performance. 1 2 3 4 5

,.
5. This organization does not provide appropriate salaries considering their contributions. (reverse) 1 2 3 4 5

,. ()
6. My confidence in using the knowledge and skills learned has decreased (reverse). 1 2 3 4 5

. ()
7. I am given opportunities to improve my skills in this organization. 1 2 3 4 5

.
8. I have successfully generated new ideas as a result of attending training courses. 1 2 3 4 5

,.
9. My boss has coached me on how to apply the knowledge and skills learned on the job. 1 2 3 4 5

.
10. I receive the training I need to do my job. 1 2 3 4 5

.
11. I am satisfied with my opportunities for promotion. 1 2 3 4 5

.
12. I believe my career aspirations can be achieved at this organization. 1 2 3 4 5

,.
13. My boss takes a supportive role in my career development. 1 2 3 4 5

221

.
14. I have opportunities for career advancement at this organization. 1 2 3 4 5

,.
15. My job does not offer me the opportunities to further my career development. (reverse) 1 2 3 4 5

. ()
16. I really do not feel I accomplish anything worthwhile. (reverse) 1 2 3 4 5

.()
17. My boss ensures that people who do a good job are recognized. 1 2 3 4 5

.
18. My boss congratulates me for doing a good job. 1 2 3 4 5

.
19. I am adequately recognized for my contributions. 1 2 3 4 5

.
20. I am doing something that is considered really worthwhile. 1 2 3 4 5

.
21. My job is challenging. 1 2 3 4 5

.
22. My job offers me the opportunities to gain work experience in challenging new areas. 1 2 3 4 5

.
23. I have ability to challenge the way things done. 1 2 3 4 5

.
24. My boss provides me with challenging work. 1 2 3 4 5

.
25. I prefer to work with routine task. (reverse) 1 2 3 4 5

. ()
26. People are encouraged to try new ways of doing things. 1 2 3 4 5

.
27. I am given an opportunity to present and try new ideas. 1 2 3 4 5

.
28. I feel encouraged to come up with new and better ways of doing things. 1 2 3 4 5

.
29. Management is genuinely interested in employee ideas on how to improve our products and services. 1 2 3 4 5

.
30. This organization does not recognize those who come up with new ideas. (reverse) 1 2 3 4 5

. ()
31. My boss involves me in decisions that affect my job. 1 2 3 4 5

.
222

32.

My boss does a good job of keeping me informed about matters affecting me.

.
33. I am satisfied with the fair treatment I receive from my boss. 1 2 3 4 5

.
34. I am satisfied with the amount of support I receive from my boss. 1 2 3 4 5

.
35. I have no confidence in the leadership of this organization. (reverse) 1 2 3 4 5

()
36. I have the freedom I need to do my job. 1 2 3 4 5

.
37. I have authority to make the decisions necessary to do my job. 1 2 3 4 5

.
38. I have personal control over the way my work is done. 1 2 3 4 5

.
39. I am satisfied with the level of autonomy associated with my position. 1 2 3 4 5

.
40. In general, I am not satisfied with the amount of autonomy I have (reverse) 1 2 3 4 5

, . ()
41. I am provided with the flexibility needed to balance the demands of my work and personal life. 1 2 3 4 5

,.
42. I have the flexibility to arrange my work so that I can meet my business objectives and balance my family and personal needs. 1 2 3 4 5

, .
43. I have sufficient time to spend with my family. 1 2 3 4 5

.
44. The company supports an appropriate work-life balance. 1 2 3 4 5

.
45. I am not satisfied with the level of balance between my work and personal life. (reverse) 1 2 3 4 5

.()
46. I like my job at this organization. 1 2 3 4 5

.
47. My work gives me a feeling of personal accomplishment. 1 2 3 4 5

.
48. I enjoy my job. 1 2 3 4 5

.
49. I am doing something that I consider satisfying in my job. 1 2 3 4 5

223

.
50. I have lost interest in my job. (reverse) 1 2 3 4 5

. ()
51. Considering everything, I will stay working for this organization. 1 2 3 4 5

,.
52. I am proud to be part of this organization. 1 2 3 4 5

.
53 This company is able to retain quality employees. 1 2 3 4 5

.
54. I see myself working for this organization three years from now. 1 2 3 4 5

.
55. I am not committed to staying with this organization (reverse) 1 2 3 4 5

. ()
Section II: Please answer General Information This section aims to identify your personal information in general. This includes your gender, age, annual salary, position, the period of your staying in the company and your suggestions. Therefore, please response the most appropriate answer in the box below:

: . , , , , , , . , . 1. What is your gender? ? Male Female 2. What is your age? ? .Below 20 years 20 .Between 20 to 24 years 20 24 .Between 25 to 29 years 25 29 .Between 30 to 34 years 30 34 35 39 .Between 35 to 39 years .Between 40 to 44 years 40 44 .Between 45 to 49 years 45 49 .Between 50 to 54 years 50 54 .Over 55 years 55 3. How long have you worked within the company? ? .Less than 1 year 1 1 2 .From 1 to 2 years .From 3 to 5 years 3 5 .From 6 to 10 years 6 10
224

.From 11 to 15 years .More than 16 years 4.

11 15 16

How long have you worked within the IT Industry? .Less than 1 year 1

5.

.From 1 to 2 years 1 2 .From 3 to 5 years 3 5 .From 6 to 10 years 6 10 .From 11 to 15 years 11 15 .More than 16 years 16 Which department are you working in? ?
...Sales and Marketing Department ...Human Resource Department Others

...Software Research and Development Department

RD

6.

What position are you working with?

? Software Engineer Project Leader Above Assistant Manager Others

Under NT$500,000 per annum Between NT$500,001 to NT$1,000,000 per annum Between NT$1,000,001 to NT$1,500,000 per annum Between NT$1,500,001 to NT$2,000,000 per annum Above NT$2,000,0001 per annum

7.

What is your current annual salary (salary and stock bonus, etc)?

8.

What is your highest education?

? Below Senior High School 2-year or 5-year College Undergraduate Postgraduate (Master) Postgraduate (Doctor)

225

Section III: Voluntary open-ended question

If you have other comments or suggestions, please describes below. Or what are the most important factors determine your decision to stay in your organization?

,. ?

_____________________________________________________________________________________________________________ _____________________________________________________________________________________________________________ _____________________________________________________________________________________________________________ _____________________________________________________________________________________________________________ _____________________________________________________________________________________________________________ _____________________________________________________________________________________________________________.

If you have any questions, or concerns with regard to this study, please contact project researcher Mr. Chin-Yao Tseng at c.tseng.10@scu.edu.au or research supervisor Associate Professor Michelle Wallace at michelle.wallace@scu.edu.au, Graduate College of Management, Southern Cross University, Australia.

, , c.tseng.10@scu.edu.au Michelle Wallace, michelle.wallace@scu.edu.au, .

THANK YOU I really appreciate your time and effort in filling out this questionnaire. Thank you very much and best wishes to you.

. , , .

226

Appendix D Survey Email

Dear Participates, Your company [ Name ] has agreed to act as a sample for my research project. I am a Doctor of Business Administration candidate at Southern Cross University in Australia. This is my doctoral dissertation about human resources management in retention of Software Development Employees. The combined results from software development employees in all of the organizations will be summarized and available to your HR managers. No individual or particular organization will be identifiable in these combined results.

, ,. . , . . .
The purpose of this survey is designed to identify the factors that determine your decision to stay in the organization. As you may agree, financial compensation is the main purpose for the work specifically yearly stock bonus. However, career development and self-achievement could also impact the employees decision of staying in the organization. So, your opinions are valuable and will help us understand how the company can retain Software Development Employees.

. , , . , . , , .
This survey is online, web-based survey. I hope you could spare 10 to 15 minutes to assist completing this survey. Your participation in this survey is anonymous and confidential. You will NOT be identified outside the research project team if you participate. So, please complete the survey as honestly as you can.

. 10 15 . . . .
If you wish to complete this online survey, please go to

, ,
http://www.surveymonkey.com/s.aspx?sm=_2b5jcQqJQxdqIOjVw YCbkBg_3d_3d
The survey runs from now to 30 November 2008.

2008 11 30 . Thank you for your assistance, , ChinYao Tseng

227

Appendix E
HUMAN RESEARCH ETHICS COMMITTEE (HREC) NOTIFICATION

To:

A/Prof Michelle Wallace/Chin Yao Tseng Graduate College of Management michelle.wallace@scu.edu.au,c.tseng.10@scu.edu.au

From:

Secretary, Human Research Ethics Committee Graduate Research College, R. Block

Date:

2 September 2008

Project:

The Retention of Software Development Employees in the IT Industry in Taiwan. Approval Number ECN-08-107

The Southern Cross University Human Research Ethics Committee has established, in accordance with the National Statement on Ethical Conduct in Human Research Section 5/Processes of Research Governance and Ethical Review, a procedure for expedited review by a delegated authority.

This application was considered by the Human Research Ethics Sub-Committee at the Tweed/Gold Coast Campus.

This application is approved subject to the usual standard conditions of approval. --------------------------------------------------------------------------------------------------------------------------------------------------------------------

Ethics Complaints Officer Sue Kelly Secretary HREC Ph: +61 +2 6626 9139 sue.kelly@scu.edu.au

Professor William Boyd Chair, HREC Ph: (02) 6620 3569 william.boyd@scu.edu.au

228

Appendix F
Question Hypothesis 1 H1 2 H1 3 H1 4 H1 5 H1 6 H2a 7 H2a 8 H2a 9 H2a 10 H2a 11 H2b 12 H2b 13 H2b 14 H2b 15 H2b 16 H3a 17 H3a 18 H3a 19 H3a 20 H3a 21 H3b 22 H3b 23 H3b 24 H3b 25 H3b 26 H3c 27 H3c 28 H3c 29 H3c 30 H3c 31 H4a 32 H4a 33 H4a 34 H4a 35 H4a 36 H4b 37 H4b 38 H4b 39 H4b 40 H4b 41 H4c 42 H4c 43 H4c 44 H4c 45 H4c 46 H5 47 H5 48 H5 49 H5 50 H5 51 Retention (DV) 52 Retention (DV) 53 Retention (DV) 54 Retention (DV) 55 Retention (DV) Demographics Variable name pay1 pay2 pay3 pay4 pay5 td1 td2 td3 td4 td5 pro1 pro2 pro3 pro4 pro5 rec1 rec2 rec3 rec4 rec5 rec1 rec2 rec3 rec4 rec5 ic1 ic2 ic3 ic4 ic5 leader1 leader2 leader3 leader4 leader5 auto1 auto2 auto3 auto4 auto5 wb1 wb2 wb3 wb4 wb5 js1 js2 js3 js4 js5 ret1 ret2 ret3 ret4 ret5 Variable value 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree 1 Strongly Disagree 2 Disagree 3 Neutral 4 Agree 5 Strongly agree Range values 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 Scale Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval Interval

229

1 2

Demographics Demographics

Gender Age

Demographics

Position

Demographics

Salary

Demographics

Academic Qualification

Demographics

Working period in current company

Demographics

Working period in IT industry

1 male 2 female 1 below 20 2 20 to 24 years 3 25 to 29 years 4 30 to 34 years 5 35 to 39 years 6 40 to 44 years 7 45 to 49 years 8 50 to 54 years 9 over 55 years 1 software engineers 2 project leaders 3 above assistant managers 1 Under NT$500,000 per annum 2 From NT$500,001 to NT$1,000,000 per annum 3 From NT$1,000,001 to NT$1,500,000 per annum 4 From NT$1,500,001 to NT$2,000,000 per annum 5 Over NT$2,000,000 per annum 1 Below senior high school 2 2-years or 5-years college 3 Undergraduate 4 Postgraduate (Master) 5 Postgraduate (Doctor) 1 Less than 1 year 2 From 1 to 2 years 3 From 3 to 5 years 4 From 6 to 10 years 5 From 11 to 15 years 6 More than 16 years 1 Less than 1 year 2 From 1 to 2 years 3 From 3 to 5 years 4 From 6 to 10 years 5 From 11 to 15 years 6 More than 16 years (rec3+pro3+rec4+td4+rec2+ic3+leader3+ pro4+auto1)/9 (pay1+pay4+pay2+pay3+pro1+pro2)/6 (auto2+ic2+auto4+auto3+leader4+leader2+ leader1+js4+ic1)/9 (cha1+cha4+cha2+td3+rec5+js2+js3+td2+td5)/9 (td1+ic5+js5+pro5+auto5+wb5+leader5+rec1+ pay5)/9 (wb3+wb1+wb2+wb4)/4 (ret1+ret2+ret3+ret4+ret5)/5

1-2 1-9

Nominal Ordinal

1-3

nominal

1-5

ordinal

1-5

ordinal

1-6

ordinal

1-6

ordinal

New composite variables 1 H1 2 3 4 5 6 7 H2 H3 H4 H5 H6 Retention (DV)

RecPro_9 Rem_6 AutoLeader_9 ChaJs_9 Dev_9 Wlb_4 Ret_5

Interval Interval Interval Interval Interval Interval Interval

230

Appendix G
KMO and Bartlett's Test Kaiser-Meyer-Olkin Measure of Sampling Adequacy. Bartlett's Test of Sphericity Approx. Chi-Square df Sig. .952 14187.224 1225 .000

231

Appendix H
Communalities Initial PAY1 TD2 PRO1 REC1 CHA1 PAY2 TD5 PRO2 REC2 CHA2 PAY3 TD3 PRO3 REC3 CHA3 PAY4 TD4 PRO4 REC4 CHA4 PAY5 TD1 PRO5 REC5 CHA5 IC1 LEADER1 AUTO1 WB1 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Extraction .725 .527 .642 .505 .622 .697 .447 .682 .600 .552 .689 .453 .699 .687 .637 .776 .556 .648 .688 .564 .623 .519 .600 .491 .487 .535 .645 .621 .727

232

JS1 IC2 LEADER2 AUTO2 WB2 JS2 IC3 LEADER3 AUTO3 WB3 JS3 IC4 LEADER4 AUTO4 WB4 JS4 IC5 LEADER5 AUTO5 WB5 JS5

1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

.680 .565 .596 .625 .730 .674 .692 .709 .698 .698 .648 .571 .677 .669 .697 .653 .603 .601 .647 .762 .606

Extraction Method: Principal Component Analysis.

233

Appendix I

234

Appendix J
Total Variance Explained Extraction Sums of Squared Initial Eigenvalues % of Component Total 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 19.057 3.500 2.796 2.306 1.478 1.160 1.149 .999 .938 .865 .831 .808 .759 .711 .692 .657 .638 .592 .556 .548 .541 .528 .456 .451 .435 .409 Variance 38.115 6.999 5.593 4.613 2.956 2.321 2.297 1.998 1.876 1.729 1.662 1.615 1.517 1.423 1.383 1.315 1.275 1.184 1.113 1.095 1.083 1.056 .912 .903 .870 .817 Cumulative % 38.115 45.114 50.707 55.320 58.275 60.596 62.893 64.891 66.767 68.496 70.158 71.773 73.291 74.714 76.097 77.412 78.687 79.871 80.984 82.079 83.162 84.218 85.131 86.033 86.904 87.721 Total 19.057 3.500 2.796 2.306 1.478 1.160 1.149 Loadings % of Variance 38.115 6.999 5.593 4.613 2.956 2.321 2.297 Cumulative % 38.115 45.114 50.707 55.320 58.275 60.596 62.893 Total 5.692 5.338 5.138 4.967 4.481 4.068 1.762 Rotation Sums of Squared Loadings % of Variance 11.383 10.677 10.277 9.934 8.962 8.135 3.525
Cumulative

% 11.383 22.060 32.337 42.271 51.233 59.368 62.893

235

27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

.393 .379 .364 .356 .338 .322 .306 .303 .283 .274 .260 .254 .243 .238 .225 .220 .213 .186 .179 .176 .174 .166 .156 .133

.786 .759 .727 .712 .675 .644 .612 .605 .565 .547 .520 .508 .486 .476 .450 .440 .426 .372 .357 .352 .349 .333 .311 .265

88.507 89.266 89.993 90.706 91.381 92.025 92.637 93.242 93.808 94.355 94.875 95.383 95.870 96.346 96.795 97.235 97.661 98.033 98.390 98.742 99.091 99.424 99.735 100.000

Extraction Method: Principal Component Analysis.

236

Appendix K
Rotated Component Matrixa Component 1 REC3 PRO3 REC4 TD4 REC2 IC3 LEADER3 PRO4 AUTO1 IC4 PAY1 PAY4 PAY2 PAY3 PRO1 PRO2 AUTO2 IC2 AUTO4 AUTO3 LEADER4 LEADER2 LEADER1 JS4 IC1 JS1 CHA1 CHA4 .358 .400 .311 .355 .350 .337 .428 .449 .521 .366 .345 .734 .679 .586 .572 .540 .540 .509 .463 .425 .354 .339 .811 .804 .772 .768 .583 .444 .724 .650 .627 .595 .563 .555 .532 .454 .426 .393 .307 .722 .605 .331 .411 .337 .325 .432 .309 .449 .353 .415 .415 .387 .351 .396 .384 .327 .319 .310 .317 .344 2 3 4 5 6 7

237

CHA2 TD3 REC5 JS2 JS3 TD2 TD5 TD1 IC5 JS5 PRO5 AUTO5 WB5 LEADER5 REC1 PAY5 WB3 WB1 WB2 WB4 CHA3 CHA5

.343

.582 .563 .336 .480 .421 .412 .545 .531 .438 .428 .373 .705 .703 .702 .698 .676 .617 .594 .572 .500 .512 .806 .719 .388 .712 .703 .545 .416 -.552 -.486 .597 .374 .331

.354

.307

Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 10 iterations.

238

Appendix L Job Appreciation

Correlations RecPro_9 REC3 RecPro_9 Pearson Correlation Sig. (2-tailed) N REC3 Pearson Correlation Sig. (2-tailed) N PRO3 Pearson Correlation Sig. (2-tailed) N REC4 Pearson Correlation Sig. (2-tailed) N TD4 Pearson Correlation Sig. (2-tailed) N REC2 Pearson Correlation Sig. (2-tailed) N IC3 Pearson Correlation Sig. (2-tailed) N .000 415 .000 415 .000 415 .000 415 .000 415 .000 415 415 .000 415 .000 415 .000 415 .000 415 .807** .000 415 .583** .000 415 .649** .000 415 .638** .000 415 .535** 415 .541** .000 415 1 .000 415 .641** .000 415 .604** .000 415 .525** .000 415 .766** .000 415 .608** .000 415 .613** .000 415 .605** 415 .427** .000 415 1 .000 415 .541** .000 415 .586** .000 415 .537** .000 415 .474** .000 415 .717** .000 415 .511** .000 415 .517** 415 .526** .000 415 1 .000 415 .427** .000 415 .535** .000 415 .552** .000 415 .567** .000 415 .403** .000 415 .836** .000 415 .641** 415 .649** .000 415 1 .000 415 .526** .000 415 .605** .000 415 .638** .000 415 .628** .000 415 .685** .000 415 .545** .000 415 .823** 415 .692** .000 415 1 .000 415 .649** .000 415 .517** .000 415 .613** .000 415 .649** .000 415 .603** .000 415 .589** .000 415 .541** 415 .793** .000 415 1 .000 415 .692** .000 415 .641** .000 415 .511** .000 415 .608** .000 415 .583** .000 415 .599** .000 415 .479** .000 415 .495** 1 .793** PRO3 .823** REC4 .836** TD4 .717** REC2 .766** IC3 .807** LEADER3 PRO4 AUTO1 .820** .788** .718**

239

LEADER3 Pearson Correlation Sig. (2-tailed) N PRO4 Pearson Correlation Sig. (2-tailed) N AUTO1 Pearson Correlation Sig. (2-tailed) N

.820**

.599**

.603**

.628**

.552**

.586**

.641**

.600**

.584**

.000 415 .788**

.000 415 .479**

.000 415 .589**

.000 415 .685**

.000 415 .567**

.000 415 .537**

.000 415 .604** 415 .600**

.000 415 1

.000 415 .506**

.000 415 .718**

.000 415 .495**

.000 415 .541**

.000 415 .545**

.000 415 .403**

.000 415 .474**

.000 415 .525**

.000 415 .584** 415 .506**

.000 415 1

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415 415

**. Correlation is significant at the 0.01 level (2-tailed).

Reliability Statistics Cronbach's Alpha .921 N of Items 9

Fair Remuneration
Correlations Rem_6 Rem_6 Pearson Correlation Sig. (2-tailed) N PAY1 Pearson Correlation Sig. (2-tailed) N PAY4 Pearson Correlation Sig. (2-tailed) N 415 .834** .000 415 .857** .000 415 415 .690** .000 415 415 1 PAY1 .834** .000 415 1 PAY4 .857** .000 415 .690** .000 415 1 PAY2 .824** .000 415 .634** .000 415 .680** .000 415 PAY3 .807** .000 415 .651** .000 415 .698** .000 415 PRO1 .800** .000 415 .577** .000 415 .573** .000 415 PRO2 .755** .000 415 .506** .000 415 .541** .000 415

240

PAY2

Pearson Correlation Sig. (2-tailed) N

.824** .000 415 .807** .000 415 .800** .000 415 .755** .000 415

.634** .000 415 .651** .000 415 .577** .000 415 .506** .000 415

.680** .000 415 .698** .000 415 .573** .000 415 .541** .000 415

.619** .000

.605** .000 415 .500** .000

.497** .000 415 .479** .000 415 .648** .000

415 .619** .000 415 .605** .000 415 .497** .000 415

415 1

PAY3

Pearson Correlation Sig. (2-tailed) N

415 .500** .000 415 .479** .000 415

415 1

PRO1

Pearson Correlation Sig. (2-tailed) N

415 .648** .000 415

415 1

PRO2

Pearson Correlation Sig. (2-tailed) N

415

**. Correlation is significant at the 0.01 level (2-tailed).

Reliability Statistics Cronbach's Alpha .897 N of Items 6

Freedom in Decision Making

Correlations AutoLeader_ AUTO 9 AutoLeader_ Pearson 9 Correlatio n Sig. (2-tailed) N AUTO2 Pearson Correlatio n 415 .706** 415 415 415 .490** 415 .524** 415 .464** 415 .508** 415 415 415 .000 .000 .000 .000 .000 .000 .000 .000 .000 1 2 IC2 AUTO AUTO LEADER LEADER LEADER 4 .814** 3 .792** 4 .806** 2 .709** 1 JS4 IC1

.706** .685*
*

.749** .756* .706*

* *

1 .515*
*

.439** .433* .393*

* *

241

Sig. (2-tailed) N IC2 Pearson Correlatio n Sig. (2-tailed) N AUTO4 Pearson Correlatio n Sig. (2-tailed) N AUTO3 Pearson Correlatio n Sig. (2-tailed) N LEADER4 Pearson Correlatio n Sig. (2-tailed) N LEADER2 Pearson Correlatio n Sig. (2-tailed) N

.000

.000

.000

.000

.000

.000

.000 .000 .000

415 .685**

415 .515**

415 1

415 .494**

415 .466**

415 .438**

415 .409**

415

415

415

.450** .429* .436*

* *

.000

.000

.000

.000

.000

.000

.000 .000 .000

415 .814**

415

415

415 1

415 .676**

415 .717**

415 .490**

415

415

415

.490** .494*
*

.533** .606* .470*

* *

.000

.000 .000

.000

.000

.000

.000 .000 .000

415 .792**

415

415

415 .676**

415 1

415 .565**

415 .460**

415

415

415

.524** .466*
*

.593** .555* .482*

* *

.000

.000 .000

.000

.000

.000

.000 .000 .000

415 .806**

415

415

415 .717**

415 .565**

415 1

415 .542**

415

415

415

.464** .438*
*

.541** .653* .502*

* *

.000

.000 .000

.000

.000

.000

.000 .000 .000

415 .709**

415

415

415 .490**

415 .460**

415 .542**

415 1

415

415

415

.508** .409*
*

.462** .438* .420*

* *

.000

.000 .000

.000

.000

.000

.000 .000 .000

415

415

415

415

415

415

415

415

415

415

242

LEADER1

Pearson Correlatio n Sig. (2-tailed) N

.749**

.439** .450*
*

.533**

.593**

.541**

.462**

1 .469* .521*
* *

.000

.000 .000

.000

.000

.000

.000

.000 .000

415 .756**

415

415

415 .606**

415 .555**

415 .653**

415 .438**

415 .469**

415

415

JS4

Pearson Correlatio n Sig. (2-tailed) N

.433** .429*
*

1 .521*
*

.000

.000 .000

.000

.000

.000

.000

.000

.000

415 .706**

415

415

415 .470**

415 .482**

415 .502**

415 .420**

415

415

415 1

IC1

Pearson Correlatio n Sig. (2-tailed) N

.393** .436*
*

.521** .521*
*

.000

.000 .000

.000

.000

.000

.000

.000 .000

415

415

415

415

415

415

415

415

415

415

**. Correlation is significant at the 0.01 level (2-tailed).

Reliability Statistics Cronbach's Alpha .901 N of Items 9

New Challenging Work

Correlations ChaJs_9 CHA1 ChaJs_9 Pearson Correlation Sig. (2-tailed) N 415 .000 415 .000 415 .000 415 .000 415 .000 415 .000 415 .000 415 .000 415 .000 415 1 .699** CHA4 .668** CHA2 .747** TD3 .611** REC5 .689** JS2 .791** JS3 .710** TD2 .705** TD5 .666**

243

CHA1

Pearson Correlation Sig. (2-tailed) N

.699**

.569**

.588**

.356**

.412**

.485**

.395**

.379**

.245**

.000 415 .668** 415 .569**

.000 415 1

.000 415 .487**

.000 415 .324**

.000 415 .422**

.000 415 .535**

.000 415 .342**

.000 415 .302**

.000 415 .284**

CHA4

Pearson Correlation Sig. (2-tailed) N

.000 415 .747**

.000 415 .588** 415 .487**

.000 415 1

.000 415 .336**

.000 415 .419**

.000 415 .515**

.000 415 .430**

.000 415 .457**

.000 415 .432**

CHA2

Pearson Correlation Sig. (2-tailed) N

.000 415 .611**

.000 415 .356**

.000 415 .324** 415 .336**

.000 415 1

.000 415 .386**

.000 415 .392**

.000 415 .343**

.000 415 .365**

.000 415 .374**

TD3

Pearson Correlation Sig. (2-tailed) N

.000 415 .689**

.000 415 .412**

.000 415 .422**

.000 415 .419** 415 .386**

.000 415 1

.000 415 .546**

.000 415 .436**

.000 415 .395**

.000 415 .359**

REC5

Pearson Correlation Sig. (2-tailed) N

.000 415 .791**

.000 415 .485**

.000 415 .535**

.000 415 .515**

.000 415 .392** 415 .546**

.000 415 1

.000 415 .624**

.000 415 .440**

.000 415 .444**

JS2

Pearson Correlation Sig. (2-tailed) N

.000 415 .710**

.000 415 .395**

.000 415 .342**

.000 415 .430**

.000 415 .343**

.000 415 .436** 415 .624**

.000 415 1

.000 415 .491**

.000 415 .416**

JS3

Pearson Correlation Sig. (2-tailed) N

.000 415 .705**

.000 415 .379**

.000 415 .302**

.000 415 .457**

.000 415 .365**

.000 415 .395**

.000 415 .440** 415 .491**

.000 415 1

.000 415 .552**

TD2

Pearson Correlation Sig. (2-tailed) N

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415 415

.000 415

244

TD5

Pearson Correlation Sig. (2-tailed) N

.666**

.245**

.284**

.432**

.374**

.359**

.444**

.416**

.552**

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415 415

**. Correlation is significant at the 0.01 level (2-tailed).

Reliability Statistics Cronbach's Alpha .869 N of Items 9

Development
Correlations Dev_9 Dev_9 Pearson Correlation Sig. (2-tailed) N TD1 Pearson Correlation Sig. (2-tailed) N IC5 Pearson Correlation Sig. (2-tailed) N JS5 Pearson Correlation Sig. (2-tailed) N PRO5 Pearson Correlation .000 415 .732** .000 415 .479** .000 415 .492** 415 .445** .000 415 1 .000 415 .483** .000 415 .413** .000 415 .474** .000 415 .337** .000 415 .386** .000 415 .710** .000 415 .455** 415 .476** .000 415 1 .000 415 .445** .000 415 .502** .000 415 .487** .000 415 .308** .000 415 .427** .000 415 .279** .000 415 .758** 415 .378** .000 415 1 .000 415 .476** .000 415 .492** .000 415 .575** .000 415 .512** .000 415 .633** .000 415 .235** .000 415 .373** 415 .655** .000 415 1 .000 415 .378** .000 415 .455** .000 415 .479** .000 415 .393** .000 415 .340** .000 415 .312** .000 415 .395** .000 415 .302** 1 TD1 .655** IC5 .758** JS5 .710** PRO5 AUTO5 WB5 LEADER5 REC1 .732** .762** .719** .694** .548** PAY5 .575**

245

Sig. (2-tailed) N AUTO5 Pearson Correlation Sig. (2-tailed) N WB5 Pearson Correlation Sig. (2-tailed) N LEADER5 Pearson Correlation Sig. (2-tailed) N REC1 Pearson Correlation Sig. (2-tailed) N PAY5 Pearson Correlation Sig. (2-tailed) N

.000 415 .762**

.000 415 .393**

.000 415 .575**

.000 415 .502** 415 .483**

.000 415 1

.000 415 .601**

.000 415 .502**

.000 415 .338**

.000 415 .300**

.000 415 .719**

.000 415 .340**

.000 415 .512**

.000 415 .487**

.000 415 .413** 415 .601**

.000 415 1

.000 415 .418**

.000 415 .312**

.000 415 .320**

.000 415 .694**

.000 415 .312**

.000 415 .633**

.000 415 .308**

.000 415 .474**

.000 415 .502** 415 .418**

.000 415 1

.000 415 .179**

.000 415 .410**

.000 415 .548**

.000 415 .395**

.000 415 .235**

.000 415 .427**

.000 415 .337**

.000 415 .338**

.000 415 .312** 415 .179**

.000 415 1

.000 415 .153**

.000 415 .575**

.000 415 .302**

.000 415 .373**

.000 415 .279**

.000 415 .386**

.000 415 .300**

.000 415 .320**

.000 415 .410** 415 .153**

.002 415 1

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.000 415

.002 415 415

**. Correlation is significant at the 0.01 level (2-tailed).

Reliability Statistics Cronbach's Alpha .857 N of Items 9

246

Workplace Flexibility
Correlations Wlb_4 Wlb_4 Pearson Correlation Sig. (2-tailed) N WB3 Pearson Correlation Sig. (2-tailed) N WB1 Pearson Correlation Sig. (2-tailed) N WB2 Pearson Correlation Sig. (2-tailed) N WB4 Pearson Correlation Sig. (2-tailed) N 415 .817** .000 415 .877** .000 415 .862** .000 415 .841** .000 415 415 .599** .000 415 .579** .000 415 .587** .000 415 415 .721** .000 415 .654** .000 415 415 .631** .000 415 415 1 WB3 .817** .000 415 1 WB1 .877** .000 415 .599** .000 415 1 WB2 .862** .000 415 .579** .000 415 .721** .000 415 1 WB4 .841** .000 415 .587** .000 415 .654** .000 415 .631** .000 415 1

**. Correlation is significant at the 0.01 level (2-tailed).

Reliability Statistics Cronbach's Alpha .871 N of Items 4

Decision to Stay
Correlations Ret_5 Ret_5 Pearson Correlation Sig. (2-tailed) 1 RET1 .806** .000 RET2 .792** .000 RET3 .806** .000 RET4 .851** .000 RET5 .674** .000

247

N RET1 Pearson Correlation Sig. (2-tailed) N RET2 Pearson Correlation Sig. (2-tailed) N RET3 Pearson Correlation Sig. (2-tailed) N RET4 Pearson Correlation Sig. (2-tailed) N RET5 Pearson Correlation Sig. (2-tailed) N

415 .806** .000 415 .792** .000 415 .806** .000 415 .851** .000 415 .674** .000 415

415 1

415 .541** .000

415 .549** .000 415 .611** .000

415 .710** .000 415 .579** .000 415 .592** .000

415 .390** .000 415 .414** .000 415 .385** .000 415 .453** .000

415 .541** .000 415 .549** .000 415 .710** .000 415 .390** .000 415

415 1

415 .611** .000 415 .579** .000 415 .414** .000 415

415 1

415 .592** .000 415 .385** .000 415

415 1

415 .453** .000 415

415 1

415

**. Correlation is significant at the 0.01 level (2-tailed).

Reliability Statistics Cronbach's Alpha .844 N of Items 5

248

Appendix M Job Appreciation

Descriptive Statistics N Statistic RecPro_9 Valid N (listwise) 415 415 Minimum Maximum Statistic 1.00 Statistic 4.67 Mean Statistic 3.2011 Std. Deviation Statistic .71100 Skewness Kurtosis

Statistic Std. Error Statistic Std. Error -.807 .120 .436 .239

Fair Remuneration
Descriptive Statistics N Statistic Rem_6 Valid N (listwise) 415 415 Minimum Maximum Statistic 1.00 Statistic 4.67 Mean Statistic 2.9064 Std. Deviation Statistic .77016 Skewness Kurtosis

Statistic Std. Error Statistic Std. Error -.395 .120 -.498 .239

Freedom in Decision Making

Descriptive Statistics N Statistic AutoLeader_9 Valid N (listwise) 415 415 Minimum Maximum Statistic 1.00 Statistic 4.56 Mean Statistic 3.2439 Std. Deviation Statistic .66036 Skewness Kurtosis

Statistic Std. Error Statistic Std. Error -.726 .120 .080 .239

New Challenging work

Descriptive Statistics N Statistic ChaJs_9 Valid N (listwise) 415 415 Minimum Maximum Statistic 1.00 Statistic 4.56 Mean Statistic 3.3068 Std. Deviation Statistic .63283 Skewness Kurtosis

Statistic Std. Error Statistic Std. Error -.526 .120 -.125 .239

249

Development
Descriptive Statistics N Statistic Dev_9 Valid N (listwise) 415 415 Minimum Maximum Statistic 1.33 Statistic 4.78 Mean Statistic 3.0450 Std. Deviation Statistic .62840 Skewness Kurtosis

Statistic Std. Error Statistic Std. Error -.009 .120 -.138 .239

Workplace Flexibility
Descriptive Statistics N Statistic Wlb_4 Valid N (listwise) 415 415 Minimum Maximum Statistic 1.00 Statistic 4.75 Mean Statistic 3.1054 Std. Deviation Statistic .82635 Skewness Kurtosis

Statistic Std. Error Statistic Std. Error -.698 .120 -.076 .239

Decision to stay
Descriptive Statistics N Statistic Ret_5 Valid N (listwise) 415 415 Minimum Maximum Statistic 1.00 Statistic 4.60 Mean Statistic 2.9055 Std. Deviation Statistic .74532 Skewness Kurtosis

Statistic Std. Error Statistic Std. Error -.479 .120 -.103 .239

250

Appendix N Job Appreciation

Fair Remuneration

251

Freedom in Decision Making

New Challenging work

252

Development

Workplace Flexibility

253

Decision to stay

254

Appendix O
Residuals Statisticsa Minimum Predicted Value Std. Predicted Value Standard Error of Predicted Value Adjusted Predicted Value Residual Std. Residual Stud. Residual Deleted Residual Stud. Deleted Residual Mahal. Distance Cook's Distance Centered Leverage Value a. Dependent Variable: Ret_sq .9828 -.58107 -3.368 -3.422 -.59958 -3.468 .388 .000 .001 2.0372 .77904 4.516 4.584 .80259 4.703 20.466 .091 .050 1.5788 .00000 .000 .000 -.00018 .000 5.985 .003 .015 .19362 .17123 .993 1.003 .17478 1.007 4.142 .007 .010 409 409 409 409 409 409 409 409 409 .9833 -3.075 .010 Maximum 2.0359 2.362 .040 Mean 1.5786 .000 .022 Std. Deviation .19357 1.000 .006 N 409 409 409

255

256

Appendix P All aggregated employees

Correlations Ret_sq Ret_sq Pearson Correlation Sig. (2-tailed) N RecPro_sq Pearson Correlation Sig. (2-tailed) N Rem_sq Pearson Correlation Sig. (2-tailed) N AutoLeader_sq Pearson Correlation Sig. (2-tailed) N ChaJs_sq Pearson Correlation Sig. (2-tailed) N Dev_9 Pearson Correlation Sig. (2-tailed) N Wlb_sq Pearson Correlation Sig. (2-tailed) N .000 409 .000 409 .000 409 .000 409 .000 409 .000 409 409 .000 409 .519** .000 409 .619** .000 409 .491** .000 409 .639** .000 409 .509** 409 .348** .000 409 1 .000 409 .420** .000 409 .430** .000 409 .349** .000 409 .415** 409 .421** .000 409 1 .000 409 .348** .000 409 .634** .000 409 .812** .000 409 .657** 409 .772** .000 409 1 .000 409 .421** .000 409 .509** .000 409 .635** .000 409 .834** 409 .596** .000 409 1 .000 409 .772** .000 409 .415** .000 409 .639** .000 409 .662** 409 .728** .000 409 1 .000 409 .596** .000 409 .657** .000 409 .349** .000 409 .491** 409 .686** .000 409 1 .000 409 .728** .000 409 .834** .000 409 .812** .000 409 .430** .000 409 .619** 1 RecPro_sq Rem_sq AutoLeader_sq ChaJs_sq .686** .662** .635** .634** Dev_9 .420** Wlb_sq .519**

**. Correlation is significant at the 0.01 level (2-tailed).

257

Software engineers
Correlations Ret_sq Ret_sq Pearson Correlation Sig. (2-tailed) N RecPro_sq Pearson Correlation Sig. (2-tailed) N Rem_sq Pearson Correlation Sig. (2-tailed) N AutoLeader_sq Pearson Correlation Sig. (2-tailed) N ChaJs_sq Pearson Correlation Sig. (2-tailed) N Dev_9 Pearson Correlation Sig. (2-tailed) N 311 311 311 311 311 311 311 .000 .000 .000 .000 .000 .000 311 .397** 311 .389** 311 .304** 311 .381** 311 .394** 311 1 311 .278** .000 .000 .000 .000 .000 .000 311 .644** 311 .815** 311 .669** 311 .784** 311 1 311 .394** 311 .516** .000 .000 .000 .000 .000 .000 311 .650** 311 .838** 311 .634** 311 1 311 .784** 311 .381** 311 .674** .000 .000 .000 .000 .000 .000 311 .681** 311 .746** 311 1 311 .634** 311 .669** 311 .304** 311 .464** .000 .000 .000 .000 .000 .000 311 .686** 311 1 311 .746** 311 .838** 311 .815** 311 .389** 311 .618** .000 .000 .000 .000 .000 .000 1 RecPro_sq .686** Rem_sq .681** AutoLeader_sq .650** ChaJs_sq .644** Dev_9 Wlb_sq .397** .497**

258

Wlb_sq

Pearson Correlation Sig. (2-tailed) N

.497**

.618**

.464**

.674**

.516**

.278**

.000

.000

.000

.000

.000

.000

311

311

311

311

311

311

311

**. Correlation is significant at the 0.01 level (2-tailed).

Project Leaders
Correlations Ret_sq Ret_sq Pearson Correlation Sig. (2-tailed) N RecPro_sq Pearson Correlation Sig. (2-tailed) N Rem_sq Pearson Correlation Sig. (2-tailed) N AutoLeader_sq Pearson Correlation Sig. (2-tailed) N ChaJs_sq Pearson Correlation Sig. (2-tailed) .000 .000 .000 .000 .013 .001 69 .506** 69 .757** 69 .680** 69 .630** 69 1 69 .299* 69 .407** .000 .000 .000 .000 .004 .000 69 .522** 69 .776** 69 .492** 69 1 69 69 69 .501** .000 .000 .000 .000 .000 .000 69 .613** 69 .692** 69 1 69 .492** 69 69 69 .611** .000 .000 .000 .000 .000 .000 69 .678** 69 1 69 .692** 69 .776** 69 69 69 .608** .000 .000 .000 .000 .000 .000 1 RecPro_sq .678** Rem_sq .613** AutoLeader_sq .522** ChaJs_sq Dev_9 Wlb_sq .539**

.506** .418**

.757** .420**

.680** .424**

.630** .343**

259

N Dev_9 Pearson Correlation Sig. (2-tailed) N Wlb_sq Pearson Correlation Sig. (2-tailed) N

69 .418**

69 .420**

69 .424**

69 .343**

69 .299*

69 1

69 .476**

.000

.000

.000

.004

.013

.000

69 .539**

69 .608**

69 .611**

69 .501**

69

69

69 1

.407** .476**

.000

.000

.000

.000

.001

.000

69

69

69

69

69

69

69

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

Assistant Managers
Correlations Ret_sq RecPro_sq Rem_sq AutoLeader_sq ChaJs_sq Dev_9 Ret_sq Pearson Correlation Sig. (2-tailed) N RecPro_sq Pearson Correlation Sig. (2-tailed) N Rem_sq Pearson Correlation Sig. (2-tailed) N AutoLeader_sq Pearson Correlation Sig. (2-tailed) .000 .000 .006 .000 .000 .000 .001 29 .856** .000 29 .926** 29 .500** .006 29 1 .003 29 .916** .000 29 .801** .011 29 .660** .000 29 .592** 29 .621** .000 29 1 .000 29 .500** .000 29 .531** .000 29 .631** .000 29 .463* 29 .788** .000 29 1 .001 29 .621** .000 29 .926** .000 29 .889** .000 29 .836** .000 29 .670** 1 .788** .592** .856** .836** .775** Wlb_sq .722**

260

N ChaJs_sq Pearson Correlation Sig. (2-tailed) N Dev_9 Pearson Correlation Sig. (2-tailed) N Wlb_sq Pearson Correlation Sig. (2-tailed) N

29 .836**

29 .889**

29 .531**

29 .916**

29 1

29 .887**

29 .690**

.000 29 .775**

.000 29 .836**

.003 29 .631**

.000 29 .801** 29 .887**

.000 29 1

.000 29 .754**

.000 29 .722**

.000 29 .670**

.000 29 .463*

.000 29 .660**

.000 29 .690** 29 .754**

.000 29 1

.000 29

.000 29

.011 29

.000 29

.000 29

.000 29 29

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

261

Appendix Q T-Tests results Development

Group Statistics age_group Dev_9 age between 20 - 29 age between 30-44 N 127 277 Mean 2.9458 3.0890 Std. Deviation .60614 .62081 Std. Error Mean .05379 .03730

Independent Samples Test Levene's Test for Equality of Variances t-test for Equality of Means 95% Confidence Interval of the Difference Sig. F Dev_9 Equal variances assumed Equal variances not assumed -2.189 249.949 .030 -.14329 .06545 -.27221 -.01438 .236 Sig. t df 402 Mean Std. Error Lower -.27312 Upper -.01347

(2-tailed) Difference Difference .031 -.14329 .06604

.627 -2.170

Job Appreciation
Group Statistics age_group RecPro_sq age between 20 - 29 age between 30-44 N 127 277 Mean 1.6953 1.6828 Std. Deviation .26435 .24927 Std. Error Mean .02346 .01498

262

Independent Samples Test Levene's Test for Equality of Variances t-test for Equality of Means 95% Confidence Interval of the Difference Sig. F RecPro_sq Equal variances assumed Equal variances not assumed .449 232.070 .654 .01251 .02783 -.04233 .06734 .003 Sig. t df 402 Mean Std. Error Upper .06604

(2-tailed) Difference Difference Lower .646 .01251 .02723 -.04102

.956 .459

Fair Remuneration
Group Statistics age_group Rem_sq age between 20 - 29 age between 30-44 N 127 277 Mean 1.5713 1.5848 Std. Deviation .27618 .26140 Std. Error Mean .02451 .01571

Independent Samples Test Levene's Test for Equality of Variances t-test for Equality of Means 95% Confidence Interval of the Difference Sig. F Rem_sq Equal variances assumed Equal variances not assumed -.462 232.823 .645 -.01343 .02911 -.07078 .04392 .019 Sig. t df 402 Mean Std. Error Lower Upper .04263

(2-tailed) Difference Difference .638 -.01343

.890 -.471

.02852 -.06950

263

Freedom in Decision Making

Group Statistics age_group AutoLeader_sq age between 20 - 29 age between 30-44 N 127 277 Mean 1.6978 1.6969 Std. Deviation .24967 .24290 Std. Error Mean .02215 .01459

Independent Samples Test Levene's Test for Equality of Variances t-test for Equality of Means 95% Confidence Interval of the Difference Sig. F AutoLeader_sq Equal variances assumed Equal variances not assumed .034 238.573 .973 .00089 .02653 -.05137 .05315 .000 Sig. t df 402 Mean Std. Error Upper .05251

(2-tailed) Difference Difference Lower .973 .00089 .02626 -.05073

.984 .034

New Challenging work

Group Statistics age_group ChaJs_sq age between 20 - 29 age between 30-44 N 127 277 Mean 1.7228 1.7180 Std. Deviation .24380 .23815 Std. Error Mean .02163 .01431

264

Independent Samples Test Levene's Test for Equality of Variances t-test for Equality of Means 95% Confidence Interval of the Difference Sig. F ChaJs_sq Equal variances assumed Equal variances not assumed .183 239.448 .855 .00474 .02594 -.04636 .05583 .591 Sig. .442 t .184 df 402 Mean Std. Error Lower -.04581 Upper .05529

(2-tailed) Difference Difference .854 .00474 .02571

Workplace Flexibility
Group Statistics age_group Wlb_sq age between 20 - 29 age between 30-44 N 127 277 Mean 1.6772 1.6405 Std. Deviation .29233 .29226 Std. Error Mean .02594 .01756

Independent Samples Test Levene's Test for Equality of Variances t-test for Equality of Means 95% Confidence Interval of the Difference Sig. F Wlb_sq Equal variances assumed Equal variances not assumed 1.169 244.503 .243 .03663 .03132 -.02507 .09833 .148 Sig. t df 402 Mean Std. Error Lower -.02495 Upper .09820

(2-tailed) Difference Difference .243 .03663 .03132

.701 1.169

265

Decision to stay
Group Statistics age_group Ret_sq age between 20 - 29 age between 30-44 N 127 277 Mean 1.5563 1.5884 Std. Deviation .26846 .25573 Std. Error Mean .02382 .01537

Independent Samples Test Levene's Test for Equality of Variances t-test for Equality of Means 95% Confidence Interval of the Difference Sig. F Ret_sq Equal variances assumed Equal variances not assumed -1.134 234.147 .258 -.03215 .02835 -.08800 .02370 .153 Sig. t df 402 Mean Std. Error Lower -.08688 Upper .02258

(2-tailed) Difference Difference .249 -.03215 .02784

.696 -1.155

266

Appendix R ANOVA Results Job Appreciation

Descriptives RecPro_sq 95% Confidence Interval for Mean N software engineer project leader assistant manager Total 311 69 29 409 Mean 1.6798 1.6966 1.7421 1.6870 Std. Deviation Std. Error Lower Bound Upper Bound Minimum Maximum .25718 .23212 .25235 .25273 .01458 .02794 .04686 .01250 1.6511 1.6408 1.6462 1.6625 1.7085 1.7523 1.8381 1.7116 1.00 1.03 1.00 1.00 2.42 2.12 2.00 2.42

Test of Homogeneity of Variances RecPro_sq Levene Statistic 2.126 df1 2 df2 406 Sig. .121

ANOVA RecPro_sq Sum of Squares Between Groups Within Groups Total .111 25.950 26.061 df 2 406 408 Mean Square .055 .064 F .866 Sig. .422

267

Multiple Comparisons RecPro_sq Tukey HSD Mean (I) position software engineer (J) position project leader assistant manager project leader software engineer assistant manager assistant manager software engineer project leader Difference (I-J) -.01678 -.06235 .01678 -.04557 .06235 .04557 Std. Error .03364 .04909 .03364 .05595 .04909 .05595 Sig. .872 .413 .872 .694 .413 .694 95% Confidence Interval Lower Bound -.0959 -.1778 -.0624 -.1772 -.0531 -.0860 Upper Bound .0624 .0531 .0959 .0860 .1778 .1772

Fair Remuneration
Descriptives Rem_sq 95% Confidence Interval for Mean N software engineer project leader assistant manager Total 311 69 29 409 Mean 1.5807 1.5592 1.6522 1.5821 Std. Deviation Std. Error Lower Bound Upper Bound Minimum Maximum .26525 .26908 .25472 .26536 .01504 .03239 .04730 .01312 1.5511 1.4945 1.5553 1.5563 1.6103 1.6238 1.7491 1.6079 1.00 1.04 1.00 1.00 2.42 2.00 2.00 2.42

Test of Homogeneity of Variances Rem_sq Levene Statistic .643 df1 2 df2 406 Sig. .526

268

ANOVA Rem_sq Sum of Squares Between Groups Within Groups Total .179 28.551 28.731 df 2 406 408 Mean Square .090 .070 F 1.275 Sig. .281

Multiple Comparisons Rem_sq Tukey HSD Mean (I) position software engineer (J) position project leader assistant manager project leader software engineer assistant manager assistant manager software engineer project leader Difference (I-J) .02148 -.07151 -.02148 -.09299 .07151 .09299 Std. Error .03529 .05149 .03529 .05869 .05149 .05869 Sig. .815 .348 .815 .253 .348 .253 95% Confidence Interval Lower Bound -.0615 -.1926 -.1045 -.2310 -.0496 -.0451 Upper Bound .1045 .0496 .0615 .0451 .1926 .2310

Freedom in Decision Making

Descriptives AutoLeader_sq 95% Confidence Interval for Mean N software engineer project leader assistant manager Total 311 69 29 409 Mean 1.6838 1.7402 1.7400 1.6973 Std. Deviation Std. Error Lower Bound Upper Bound Minimum Maximum .24320 .23391 .26476 .24382 .01379 .02816 .04916 .01206 1.6567 1.6841 1.6393 1.6736 1.7109 1.7964 1.8408 1.7210 1.06 1.03 1.00 1.00 2.33 2.18 2.06 2.33

269

Test of Homogeneity of Variances AutoLeader_sq Levene Statistic .248 df1 2 df2 406 Sig. .781

ANOVA AutoLeader_sq Sum of Squares Between Groups Within Groups Total .237 24.019 24.256 df 2 406 408 Mean Square .118 .059 F 2.003 Sig. .136

Multiple Comparisons AutoLeader_sq Tukey HSD Mean (I) position software engineer (J) position project leader assistant manager project leader software engineer assistant manager assistant manager software engineer project leader Difference (I-J) -.05645 -.05625 .05645 .00020 .05625 -.00020 Std. Error .03237 .04723 .03237 .05383 .04723 .05383 Sig. .190 .459 .190 1.000 .459 1.000 95% Confidence Interval Lower Bound -.1326 -.1673 -.0197 -.1264 -.0548 -.1268 Upper Bound .0197 .0548 .1326 .1268 .1673 .1264

270

New Challenging work

Descriptives ChaJs_sq 95% Confidence Interval for Mean N software engineer project leader assistant manager Total 311 69 29 409 Mean 1.7147 1.7374 1.7397 1.7203 Std. Deviation Std. Error Lower Bound Upper Bound Minimum Maximum .24478 .20631 .25558 .23917 .01388 .02484 .04746 .01183 1.6874 1.6879 1.6425 1.6971 1.7420 1.7870 1.8369 1.7436 1.06 1.24 1.00 1.00 2.33 2.12 2.12 2.33

Test of Homogeneity of Variances ChaJs_sq Levene Statistic 2.581 df1 2 df2 406 Sig. .077

ANOVA ChaJs_sq Sum of Squares Between Groups Within Groups Total .041 23.298 23.339 df 2 406 408 Mean Square .020 .057 F .356 Sig. .701

271

Multiple Comparisons ChaJs_sq Tukey HSD Mean (I) position software engineer (J) position project leader assistant manager project leader software engineer assistant manager assistant manager software engineer project leader Difference (I-J) -.02270 -.02502 .02270 -.00232 .02502 .00232 Std. Error .03188 .04651 .03188 .05301 .04651 .05301 Sig. .757 .853 .757 .999 .853 .999 95% Confidence Interval Lower Bound -.0977 -.1344 -.0523 -.1270 -.0844 -.1224 Upper Bound .0523 .0844 .0977 .1224 .1344 .1270

Development
Descriptives Dev_9 95% Confidence Interval for Mean N software engineer project leader assistant manager Total 311 69 29 409 Mean 3.0100 3.1256 3.2375 3.0456 Std. Deviation Std. Error Lower Bound Upper Bound Minimum Maximum .60804 .61538 .68088 .61685 .03448 .07408 .12644 .03050 2.9422 2.9778 2.9786 2.9857 3.0778 3.2734 3.4965 3.1056 1.67 1.44 1.44 1.44 4.78 4.44 4.00 4.78

Test of Homogeneity of Variances Dev_9 Levene Statistic .304 df1 2 df2 406 Sig. .738

272

ANOVA Dev_9 Sum of Squares Between Groups Within Groups Total 1.904 153.343 155.247 df 2 406 408 Mean Square .952 .378 F 2.521 Sig. .082

Multiple Comparisons Dev_9 Tukey HSD Mean (I) position software engineer (J) position project leader assistant manager project leader software engineer assistant manager assistant manager software engineer project leader Difference (I-J) -.11560 -.22754 .11560 -.11194 .22754 .11194 Std. Error .08178 .11932 .08178 .13601 .11932 .13601 Sig. .335 .138 .335 .689 .138 .689 95% Confidence Interval Lower Bound -.3080 -.5082 -.0768 -.4319 -.0531 -.2080 Upper Bound .0768 .0531 .3080 .2080 .5082 .4319

Decision to stay
Descriptives Ret_sq 95% Confidence Interval for Mean N software engineer project leader assistant manager Total 311 69 29 409 Mean 1.5894 1.5254 1.5896 1.5786 Std. Deviation Std. Error Lower Bound Upper Bound Minimum Maximum .26121 .23498 .27445 .25844 .01481 .02829 .05096 .01278 1.5602 1.4689 1.4852 1.5535 1.6185 1.5818 1.6940 1.6037 1.00 1.00 1.00 1.00 2.37 2.00 2.00 2.37

273

Test of Homogeneity of Variances Ret_sq Levene Statistic .362 df1 2 df2 406 Sig. .696

ANOVA Ret_sq Sum of Squares Between Groups Within Groups Total .235 27.015 27.250 df 2 406 408 Mean Square .118 .067 F 1.767 Sig. .172

Multiple Comparisons Ret_sq Tukey HSD Mean (I) position software engineer (J) position project leader assistant manager project leader software engineer assistant manager assistant manager software engineer project leader Difference (I-J) .06402 -.00019 -.06402 -.06420 .00019 .06420 Std. Error .03433 .05008 .03433 .05709 .05008 .05709 Sig. .150 1.000 .150 .499 1.000 .499 95% Confidence Interval Lower Bound -.0167 -.1180 -.1448 -.1985 -.1176 -.0701 Upper Bound .1448 .1176 .0167 .0701 .1180 .1985

274