Sie sind auf Seite 1von 134

MODULE10

QUALITATIVE STU DIES

INTRODUCTION

In the previous modules, I discussed the different research designs that can re­
spond to descriptive and explanatory questions with a quantitative perspective. As
a rule, they are characterized by the overall purpose of describing and explaining
phenomena through the collection of numerical data. They aim to quantify infor­
mation as a basis for making a generalization.

The quantitative approach is one perspective or paradigm that many researchers


are inclined to adopt.

There are times, however, when the nature of the study or the theory behind it
· calls for qualitative studies. For instance, if the research focus is strategies used in
the implementation of a program or project (i.e., the Minimum Basic Needs Ap­
proach in the Social Reform Agenda), a qualitative study is appropriate. Instead of
quantifying the output or the impact of the program, the focus is on what it has
accomplished in applying the strategies.

Of course, researchers may have a given level of comfort with a particular approach
or perspective (i.e., quantitative or qualitative). The choice of research design may
be shaped by one's aptitude for one mode over the other. Nevertheless, combining
the two perspectives could enrich the researcher's understanding of a given prob­
lem.

In this module, I will discuss popular designs in the conduct of qualitative research.
142 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

OBJECTIVES

At the end of this module, you will be able to:

Define the basic characteristics of qualitative research and differentiate it from


quantitative research;
Identify rules in the formulation of qualitative research; and
Distinguish case studies from the historical approach in the conduct of quali­
tative studies.

1.0 QUALITATIVE STUDIES

Qualitative research studies, as a perspective, is characterized by the following assump­


tions made by the researcher (Cresswell 1994: 145):

Qualitative researchers are concerned primarily with process, rather than out­
comes or products.
Qualitative researchers are interested in meaning-how people make sense of
their lives, their experiences, and their structures of the world.
The qualitative researcher is the primary instrument for data collection and
analysis. Data are mediated through this human instrument, rather than
through inventories, questionnaires, or machines.
Qualitative research involves fieldwork. The researcher physically goes to the
people, setting, site, or institution to observe behavior in its natural setting.
Qualitative research is descriptive in that the researcher is interested in proc­
ess, meaning, and understanding gained through words or pictures.
The process of qualitative research is inductive in that the researcher builds
abstractions, concepts, hypotheses, and theories from details.

Qualitative studies differ markedly from quantitative research in that they focus on
detailing the processes of events, while quantitative studies focus on outcomes or prod­
ucts, which can normally be counted.

Furthermore, qualitative studies emphasize the importance of individuals who share


their experiences, rather than combine their experiences in aggregate form where the
person loses his/her identity. The specific experience of the person or institution is
more important than the figures or scores obtained to measure their performance, as
in quantitative studies.

Qualitative research is similar to survey research since the events being observed oc­
cur in the natural setting. No intervention is made by the researcher. On the other
hand, it differs from surveys since qualitative research is not concerned with obtaining
a sufficient number of cases to represent the population. Rather, qualitative studies are
more concerned with obtaining elements that can best speak of the topic of the inves­
tigation. This involves fieldwork, since the researcher has to witness the event by go­
UP OPEN ing to the site, the setting, or the people, to observe the behavior as it happens or to
UNIVERSITY capture it through other artifacts (such as documents or records).
RESEARCH DESIGN / Qualitative Studies 143

Cresswell emphasizes the tendency of qualitative studies to be inductive. However,


there are instances when qualitative studies may spring from theory and attempt to
show causal or relational patterns. But the burden of proof for its capacity to general­
ize findings to the bigger population is stiffer for qualitative studies.

In general, the design and method of qualitative studies are more flexible and they are
spelled out in more general terms prior to the actual study (Gay 1996: 242).

To summarize, the following points highlight the distinction between quantitative


and qualitative research:

Box 10-1. Comparison of qualitative and quantitive approaches

Quantitative Qualitative

Purpose of the study To describe and explain To detail the processes ap­
events using numerical data plied and draw patterns/
to ultimately be able to con­ insights; to explore/deter­
trol and predict mine data factors influencing
an event using words and
visuals

Role of the participants As subjects or respondents As persons/individuals/


whose views are aggregated groups whose ideas and feel­
into statistical data ings are important; identity is
important

Role of the researcher Manipulates conditions for Observes events in the natu-
experiments/quasi-experi- ral setting, as in surveys (al­
ments though it is not the concern to
quantify, as in the latter case)

2.0 RULES IN FORMULATING QUALITATIVE RESEARCH

Marshall and Rossman (1995: 42) have formulated the following rules for conducting
qualitative studies:

2.1 INFORMATIONAL ADEQUACY

This means that the research design is able to gather sufficient information to
respond to the variables identified in the study. The strategies identified should
be able to pass the question. "Will the strategy elicit the sought after informa­
tion?" (Marshall and Rossman 1995: 42). For instance, is it enough to rely on key
informants to trace the processes applied to implement a program or a project?
Are there written documents that will help in reconstructing these experiences? UP OPEN
The researcher should ask these questions. UNIVERSITY
144 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

2.2 EFFICIENCY

This means that the approaches or strategies adopted ensure that adequate data col­
lected at the least cost in time and effort for the participants. If there are enough writ­
ten materials for reconstructing the processes applied in implementing programs and
projects, these should be exhausted first before key informant interviews are set up.
The latter can be resorted to if there are gaps in the documents reviewed.

2.3 ETHICAL CONSIDERATIONS

This refers to the capability of the research approach to respect the rights of the par­
ticipants in the study. This responds to the questions: "Will the proposed strategy
violate the participants' privacy or unduly disrupt their everyday worlds? Are they
putting themselves in danger or at risk by participating in the study?" (Marshall and
Rossman 1995: 42).

The rules cited above indicate to us that the methodology for collecting information
to make sure that responding to efficiency and adequacy issues will not trample on the
rights of the persons being observed. Definitely, issues about maintaining the balance
between an adequate research design and the rights of the subjects may be affected by
particular approaches selected. This balance will be discussed again in subsequent
sections.

Let me cite for the moment one of the approaches in the study of negative bureau­
cratic behavior in applying for a license to operate a franchise for a taxicab business.
One of the alternative strategies the team selected was deploying a participant ob­
server who will pose as an applicant for a taxicab business. The identity of the bureau­
crats was not divulged because this would mean subjecting the persons involved to
legal action. The objective in presenting the experiences of violation of regular bu­
reaucratic norms was not to identify persons but positions which manifested common
types of corruptive behaviors.

Care was exercised in preparing the report since the persons observed did not even
know they were being studied.

3.0 CASE STUDIES

3.1 FOCUS

One of the most popular qualitative methods in public administration is the conduct
of case study. The case study approach normally entails selection of a particular level
of analytic interest such as a person, group, program, organization, or inter-organiza­
tion (Marshall and Rossman 1995: 42). This is an important decision to make in a case
study.

UP OPEN In public administration, a case study normally focuses on a particular organization.


UNIVERSITY This can be a group in a duly organized public institution such as one unit in an office
RESEARCH DESIGN / Qualitative Studies 145

or department. It may also be a social unit that evolves without the benefit of a
formal process of legitimization, such as an informal network of individuals from
the private sector who may be interested in lobbying for reassessing programs or
projects.

The usual objective of a case study is to capture how an organization has evolved, how
it is constituted, and how it normally performs its regular activities. The major thrust
is to characterize how it operates in reality. But of course, the aspects of the organiza­
tion one will focus on should be spelled out. In other words, the identification of the
level of analysis depends upon the variables of the study and the theoretical basis for
focusing on this set of variables.

A case study may also focus on specific individuals in the social unit, such as the
leader or a service delivery person. The objective could be to disclose how the person
is performing, the specific approaches or techniques adopted in the performance of a
particular role, and the problems or difficulties that may have been encountered in
performing this role. Lessons can then be drawn from the nature or character of lead­
ership or service delivery person.

The focus may also be a particular program-its organizational structure, people, re­
sources, and approaches-and how it operates in a particular context or setting.

3.2 TIMING

It is important to determine whether the case study will focus on what transpired in
the past (retrospective) or will be documented as it occurs at present. This is a choice
that the researcher may have to make. Techniques of data collection can be shaped by
the time frame for the study.

Reconstructing a past event should, however, ensure that the event can be subjected to
recall. The different techniques of data collection that will be adopted should see to it
that this can be done. The issues pertinent to techniques of data collection will be
discussed in a different module.

3.3 NUMBER OF GROUPS/PERSONS

An important decision to make is the selection of the particular individual, group,


institution, program, or event to be studied. Which ones among the many will be
identified and why? A set of criteria can be drawn up to serve as the basis for the
identification of particular cases.

Another concern is determining how many persons or groups will be studied by the
researcher. This can vary from one to many depending on the number available and
the purpose or intention of the researcher.

One group may be highlighted because it is a priority group implementing programs


and projects. In case there are many groups, the focus may be on some groups with the UP OPEN
same characteristics in order that common patterns are identified. UNIVERSITY
146 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

For instance, the first documentation on the implementation of the Minimum Basic
Needs Approach in the Social Reform Agenda was conducted by researchers and prac­
titioners identified by the Department of the Interior and Local Government (DILG).
It was published in their bulletin, called Innovations, in 1996. This research focused on
pilot areas introduced to the MBN approach. Each of the cases discussed the persons
responsible for introducing the approach in pilot provinces, municipalities, and
barangays; the extent of MBN implementation; the involvement of local government
units, nongovernment organizations, and people's organizations in its implementa­
tion; and the problems/issues in the implementation of the approach. Clearly, the pur­
pose of the case reports was descriptive: to recount the dynamics of implementing the
approach in the different localities.

On the other hand, contrasting cases can be helpful in demonstrating a causal or


relational argument, especially if the groups or elements being compared have vary­
ing experiences. For instance, studies of health management conducted in the Area­
Based Child Survival and Development Program implemented in the seven most de­
pressed provinces in the country prior to devolution (see Bautista 1993) contrasted
three most successful and three least performing barangays in terms of level of utiliza­
tion of social development programs packaged-for targeted localities. From the six
cases, comparisons were made to show what factors were demonstrated by the case
areas in terms of:

compliance with the suggested strategies;


resources available in the locality (i.e., health facilities, day care center, man­
power);
preparation of the service delivery workers; and
innovations or initiatives by the local leaders to harness the program in the area.

The research was able to trace the factors related to the utilization of the health serv­
ices. It was able to respond to a relational type of research problem.

It is useful to define the reasons for selecting certain groups or individuals. The re­
searcher must be able explain the rationale for the focus on them.

The persons or groups selected are situated in a context or setting that may be crucial
to the performance of the person or group being studied. Some questions that may be
asked include (see Marshall and Rossman: 1995: 51):

What is unique?
What charactersitics of this setting are compelling and unusual?

Marshall and Rossman (1995: 51) consider the site or setting as important. They stipu­
late that the site is one where:

entry is possible;
there is high probability that a rich mix of the processes, people, programs,
UP OPEN interactions, and structure of interest are present;
UNIVERSITY
RESEARCH DESIGN/ Qualitative Studies 147

the researcher is likely to be able to build trusting relations with the partici­
pants of the study; and
data quality and credibility of the study are reasonably assured.

Do the following activity to bolster our discussion of the case study approach.

ACTIVITY 10-1

Read the article,"Towards the Institutionalization of ABCSDP: Highlights of a Fea­


sibility Assessment," in PJPA XXXVI, 3 (July 1992) to see how the ·cases were
selected as a complement to the survey portion of the study. This study demon­
strates to you that quantitative and qualitative studies can be merged.

4.0 HISTORICAL RESEARCH

4.1 FOCUS

Historical research is the systematic collection and evaluation of data related to past
occurrences in order to describe causes, effects, or trends that may help to explain
present events and anticipate the future (Gay 1996: 185). Unlike case studies, historical
accounts do not focus on one particular moment in time. It follows through an event
over a given period, commencing always with the past. It is a chronology of events
taking note of the important milestones or landmarks in the life span of a social unit or
individual.

It is important for the researcher to determine the particular set of variables or factors
that should be included in the historical account. This should be clear at the outset, as
the research problem is formulated and the variables of the study are laid out. A con­
ceptual or theoretical framework should be able to explain why this focus is necessary,
to provide the rationale for the study.

One of the dissertations prepared by a graduate student at UP CPA makes a historical


analysis of the perspectives of churches on development in the Philippines during the
period 1972-1990, with different periods being characterized thus: the beginning of
the 1970s prior to martial law; 1972-1974, marking the imposition of martial law;
1975-1986 as the period of the dictatorship; and 1987-1990, the process of
redemocratization (Genato-Rebullida 1991). Two of the biggest churches were stud­
ied: the Catholic Church and the Protestant Church. The record of performance was
based on the following theoretical foundation:

UP OPEN
UNIVERSITY
148 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

The Churches' Theoretical Model of Development

To date, the Catholic Church, the National Council of Churches (NCCP), and its Prot­
estant member-churches pursue a line of thought on the issue of development. What is
here called the "churches' theoretical model of development" consists of statements
defining and stipulating the indicators of, and the strategies to pursue and attain de­
velopment in the Philippines. This presents a vision of a "developed Philippines" and
the processes to get there.

The churches' view of development has been draW!J from pastoral statements, state­
ments of concern, and other press releases, as well as decisions and resolutions of their
authoritative policy and decision structures. Due to the nature of these sources, a policy
framework can be used to analyze the churches' position on development.

Policy has been defined as:

...a course of action of a person, group or government within a given environment


providing obstacles and opportunities which policy proposed to utilize and overcome
in an effort to reach a goal or realize an objective or a purpose.

...a purpose, course of action directed toward the accomplishment of some purpose or
goal.

...a purpose, course of action followed by an actor or set of actors in dealing with a
problem as matter of concern (Anderson 1975: 3).

Policy statement is the declaration itself of the proposition taken on the matter; and
policy content refers to the implicit and explicit goals and objectives including guide­
lines or strategies for action (Grindle 1980: 6-10).

The definition and strategies for Philippine development may be viewed as an institu­
tional stance because these were arrived at according to the churches' system of policy
decisionmaking. This signifies the prevailing and collective choice of those placed in
authority, over and above individual decisions or opinions. Disagreements, debates,
and consultations have only sharpened the focus on issues.

With the strong representation of the Roman Catholic Church in the Philippines,
policymaking is identified with the Catholic Bishops Conference of the Philippines
(CBCP). In the case of the NCCP, the general Convention charts the direction until the
next biennium, with an ad-interim Executive Committee to carry out the functions of
the Council. The NCCP member-churches are guided by the principles of collegiality
and commitment to consultation.

Source: Genato-Rebullida 1991. 71-72.


UP OPEN
UNIVERSITY
RESEARCH DESIGN / Qualitative Studies 149

In the previous example, the level of analysis is the organization of two churches. The
critical factors examined is the church's perspective on development-specifically, its
policy and program of work. A conceptual clarification of policy was made and em­
bodied in the plan of the study. The framework also indicates the sources of informa­
tion as the basis for discussing patterns regarding development perspectives.

4.2 ROLE OF THE RESEARCHER

Like case studies, historical research is not subject to the control procedures observed
in experiments and quasi-experiments. Rather, it is based on the study pf events in the
natural setting.

In the historical approach, the researcher goes through the normal steps involved in
the conduct of research, i.e., definition of a problem, formulation of questions to be
answered or hypotheses to be tested, systematic collection of data, evaluation of data,
and production of a verbal synthesis or confirmation/disconfirmation of hypotheses
(Gay 1996: 186).

The researcher must make sure that there is something new to be discovered in apply­
ing the approach or that it will clarify, correct, or expand existing knowledge (Gay
1996: 1187). It should not be undertaken to retell what is already known.

It is time f0r another SAQ!

SAO 10-1

True or False. If the answer is false, fill up the blank with the correct answer.

_____ 1. Qualitative research does not require the researcher to prepare a


research proposal.
_____ 2. In a qualitative study, the researcher is expected to man;p ulate
the events to take place.
_____ 3. In a qualitative study, the identity of the subjects/participants is
very ;mportant.
_____ 4. Case studies are different from surveys in all respects.
_____ 5. Case studies like historical approach may also cover a past event.

See the answers next page.

UP OPEN
UNIVERSITY
1 so PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

ASAQ 10-1

1. False. Even qualitative studies require the preparation of a research proposal.


2. False. Qualitative study focuses on events occurring in the natural setting.
3. True.
4. False. Both study events in the natural setting but differ in terms of presenta­
tion of data. Qualitative study uses narrative statements while survey is concerned
with quantification.
5. True.

ACTIVITY 10-2

Compare the articles of Cordero (see Module 5) and Proserpina Domingo-Tapales et


al. on "Value Profile and Corruption Propensity: Correlates Among Employees in
Two Types of Government Agencies" in the Philippine Journal of Public Adminis­
tration XXXIX, 4 (October 1995) and indicate:

1. What approach was applied in each: quantitative or qualitative research or


both?
2. What specific type of research does each represent?
3. What specific type of research design was applied in each?.

ANSWERS TO ACTIVITY 10-2

·e:iep AJepuOJas
6u!sn 'pa!1dde seM 4Je0Jdde 1eu!pn:i!6uo1 e ':iJed aAqe:iquenb a4:i JO.:J ·Apn:is
a4:i :J.O :iJed aAqe:i!1enb a4:i JO:J. 4Je0Jdde Apn:is aseJ a4:i pa!1dde Apn:is s,oJapJoJ ·t:

·sa!Jepyauaq :ia6Je:i a4:i JO:J. weJ6


oJd a4:i :J.O uoqeJ!ldW! a4:i uo os1e :inq uoqez!ue6Jo a4:i :J.O spadse 1euJa:iu!
a4:i uo Aluo :iou pasnJO:J. :i! se JapeJe4J LI! aAqen1e/\a seM apqJe s,oJapJoJ ·z

·e:iep AJepuoJas uo paseq passnJS!P aJaM sawoJ:ino pue s:ind:ino se :iua:ixa


pa:i!wn e o:i pandde seM 4Je0Jdde aAqe:iquenb v ·seaJe Apn:is aseJ LI! weJ6oJd
a4:i:J.o uoqe:iuawa1dw! a4:i :J.O uo!ssnJS!P a4:i LI! pa!1dde seM 4Je0Jdde aAq
UP OPEN -e:inenb v ·sa4Je0Jdde aAqemuenb pue aAqe:i!1enb 4:ioq pasn apqJe s,oJapJOJ ·i
UNIVERSITY
RESEARCH DESIGN/ Qualitative Studies 151

SUMMARY

This module makes a distinction between quantitative and qualitative research ac­
cording to the purpose of the study, the role of the participants, and the role of the
researcher.

In quantitative research, the purpose of the study is to derive numerical data to re­
spond to descriptive and explanatory research problems. In qualitative studies, the
process is discussed through narrative statements.

Participants in quantitative research are not important as individuals•since data are


aggregated in statistical terms. On the other hand, the identity of the individual or the
group focused on is highlighted in qualitative research.

The researcher's role in setting up the experiments or quasi-experiments is one of


control and manipulation. He/She has a role to play in setting up the experimental
variable. On the other hand, qualitative studies and surveys are similar in that both
study events in the natural setting. They differ in that qualitative research focuses on
process while survey research obtains numerical data to measure a set of variables.

This module also differentiates between case study and historical analysis as two modes
in the conduct of qualitative studies. Case studies focus on a particular level of analysis
(i.e., an individual or a group) and how each operates relative to some variables. This
may focus on past or current occurrences.

On the other hand, historical studies reconstruct the chronology of events in the past
until a given point in time. The variables and theoretical basis of the research problem
serve as the springboard for the identification of what events to focus on and why.

Comparison groups are helpful in contrasting the experiences of groups/individuals/


events to demonstrate relational or causal arguments in case studies. However, the
criteria for the identification of groups should be spelled out. It is not the number of
groups or elements that is important but the rationale for the choice, even if they are
small in number.

This module has only examined the basic structure in setting up the observation-i.e.,
the amount of manipulation that takes place, the timing ofobservation, and the number
of groups to be studied. Other issues in research methodology will be answered in
subsequent sections on sampling, techniques of data collection, and data analysis/in­
terpretation.

UP OPEN
UNIVERSITY
RESEARCH DESIGN/ Qualitative Studies 151

SUMMARY

This module makes a distinction between quantitative and qualitative research ac­
cording to the purpose of the study, the role of the participants, and the role of the
researcher.

In quantitative research, the purpose of the study is to derive numerical data to re­
spond to descriptive and explanatory research problems. In qualitative studies, the
process is discussed through narrative statements.

Participants in quantitative research are not important as individuals•since data are


aggregated in statistical terms. On the other hand, the identity of the individual or the
group focused on is highlighted in qualitative research.

The researcher's role in setting up the experiments or quasi-experiments is one of


control and manipulation. He/She has a role to play in setting up the experimental
variable. On the other hand, qualitative studies and surveys are similar in that both
study events in the natural setting. They differ in that qualitative research focuses on
process while survey research obtains numerical data to measure a set of variables.

This module also differentiates between case study and historical analysis as two modes
in the conduct of qualitative studies. Case studies focus on a particular level of analysis
(i.e., an individual or a group) and how each operates relative to some variables. This
may focus on past or current occurrences.

On the other hand, historical studies reconstruct the chronology of events in the past
until a given point in time. The variables and theoretical basis of the research problem
serve as the springboard for the identification of what events to focus on and why.

Comparison groups are helpful in contrasting the experiences of groups/individuals/


events to demonstrate relational or causal arguments in case studies. However, the
criteria for the identification of groups should be spelled out. It is not the number of
groups or elements that is important but the rationale for the choice, even if they are
small in number.

This module has only examined the basic structure in setting up the observation-i.e.,
the amount of manipulation that takes place, the timing of observation, and the number
of groups to be studied. Other issues in research methodology will be answered in
subsequent sections on sampling, techniques of data collection, and data analysis/in­
terpretation.

UP OPEN
UNIVERSITY
152 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

REFERENCES

Bautista, Victoria A.
1992 Towards the Institutionalization of ABCSDP: Highlights of a Feasibility Assessment.
Philippine Journal of Public Administration (PJPA) XXXVI, 3 (July).
Creswell, John W.
1994 Research Design: Qualitative and Quantitative Approaches. Thousand Oaks: Sage Publi­
cations.
Gay, L.R.
1996 Educational Research: Competencies for Analysis and Application. Columbus, Ohio:
Prentice Hall.
Genato-Rebullida, Ma. Lourdes
1991 Church Development Perspective: Policy Formulation and Implementation. PJPA
XXXV, 1 Oanuary).
Marshall, Catherine and Gretchen B. Rossman
1995 Designing Qualitative Research. Thousand Oaks: Sage Publications.

UP OPEN
UNIVERSITY
SELECTING THE UNIT
OF ANA,LYSIS
AND SAMPLING TECHNIQUES
Dear Student,

One of the methodological issues you have to deal with has to do with determining the
elements from whom to collect the data or information your research requires.

To respond to this basic concern, two modules encompass this unit. Module 11 dis­
cusses the application of probability sampling strategies and the basic concepts that
have to be understood in identifying and selecting the units of analysis.

Module 12 discusses non-prcbability sampling strategies, as well as the issues that


have to be dealt with in sampling and how to respond to them.

To cap this unit, you are required to identify the unit of analysis and sampling scheme
(where applicable) for your own research proposal. Make sure that you accomplish the
SAQs as an aid in applying the concepts you will learn in this unit.
MODULE 11.
SELECTING THE UNIT OF ANALYSIS
AND PROBABILITY SAMPLING
STRATEGIES

INTRODUCTION

In responding to a given research problem, one of the methodological issues we


have to deal with, in addition to identifying the research design, has to do with
collecting information. From whom it will be gathered? Where are they based?
What number of respondents will typify the population we wish to study?

Hence, one of the methodological aspects of the research proposal that we need to
decide on is (a) the unit of analysis and (b) the subset of elements from this unit
that will serve as the respondents of the study.

OBJECTIVES

At the end of this module, you will be able to:

Differentiate the various concepts relevant to the selection of elements of


the study;
Discuss the importance of sampling as an approach;
Characterize the probability Sijmpling strategy and its different types; and
Select which probability sampling strategy to use.
1 58 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

1.0 BASIC CONCEPTS IN DEALING WITH THE ELEMENTS OF THE STUDY

We have discussed how important it is to deal with the issue of external validity in the
formulation of a research design. To reiterate, external validity aims to ensure that the
elements from whom we are drawing information are representative of the population
that we wish to talk about. In the conduct of a study, it is important to determine who
serves as the total number of elements about which we hope to make generalizations.
This total represents the population of the study. For instance, in the study conducted
on the Area Based Child Survival and Development Program (ABCSDP) which tar­
geted the seven most depressed provinces of the Philippines, the population of direct
beneficiaries were the women and children found:in the targeted barangays in these
areas. Hence, the population of interest was not the entire population of the seven
provinces but only some segments of it.

But the group of all women and children in ABCSDP barangays is still too large as a
focus of study. One research option is to select a subset of elements from the popula­
tion, such as the ABCSDP beneficiaries. This subset of elements is the sample. If
there are 10,000 elements in the population in the list of targeted women in the pro­
gram and 1,000 serve as the respondents of the study, the 1,000 represent the sample of
the study.

To be able to define the samples to be drawn for research, it is important to have a list
of names or elements from which the samples will be obtained. This list is called the
sampling frame.

In the assessment of the effectiveness of the ABCSDP, the target for assessment was
both women and children. They were the unit of analysis, or the elements about whom
the study was conducted. They served as the focal groups in the study. However, there
are times when it is not necessary to observe or study the whole unit of analysis to
derive information about its condition or status with respect to the key variables of the
study. Only some of them may be tapped to be the observation units or respondents
of the study. The respondents are the ones who directly provide the information about
the key variables of the study. In the study on ABCSDP, only the mothers were the
respondents of the study as they could already provide information about the children
in the family. In short, the unit of analysis need not always constitute the population
from which the sample respondents will be drawn.

There are instances when the unit of analysis and the observation unit are one and the
same group. For example, in a study of small farmers in an agricultural production
program, the farmers serve both as the unit of analysis and the observation unit.

2.0 THE IMPORTANCE Of·SAMPLING

W hy do we resort to samples as source of information?

UP OPEN I am sure your immediate response is that it is more economical to do so. You save on
UNIVERSITY the cost of materials (since you are cutting down on the amount of paper used for
SAMPLING TECHNIQUES/ Probability Sampling Strategies 1 59

questionnaires, for example) and travel costs to reach your target respondents. You
also cut down on people who will assist in implementing your interview schedule and
in processing the information you have gathered. It is very costly to do all these if you
are studying the entire population.

l
I

',c,

,.·
. ·�o 41
,O• ... ;1
H


' c:.::::J·.·-
•, ·<> K S

.C!f·/,/�. T ! I
jD :

But more important than cost-cutting is obtaining more accurate information by sam­
pling than by trying to reach the entire population. Why do I say this? Consider:

Getting to the population requires an extended period of time such that sys­
tematic changes may have occurred among the elements before we are in a
position to make some generalizations. For example, if a census is slow in ob­
taining information, a new child may be born to a family already visited by a
data collector.
To obtain accurate and reliable information about the population, highly trained
people must be deployed.
T here is also the difficulty of getting good supervisors who will monitor the UP OPEN
performance of the research staff. UNIVERSITY
160 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

3.0 TYPES OF PROBABILITY SAMPLING STRATEGIES

There are two modes of sampling. One mode is called probability sampling. The sec­
ond is non-probability sampling.

Let me discuss the probability sampling approach in this module. Non-probability


sampling will be discussed in the next module.

The basic principle of probability sampling is that elements are randomly selected
from a population. This ensures that bias is avoided in the identification of the ele­
ments. It is an efficient method of selecting eleme{lts which may have varied charac­
teristics, as the process allows for a fair representation of this variability. It also means
that the laws of probability and statistics apply, allowing us to make certain inferences
(if you recall your statistics class).

The randomness of the selection process in probability sampling reminds me of the


popular lottery method of selecting winners of company prizes. If you thought that the
winners were pre-selected, would you continue to send in your entries? You will not
have a chance of winning anyway. You would say the selection is "rigged" and "un­
fair." The lottery method works like probability sampling. The elements to be studied
are identified in an objective way and not on the basis of personal preferences.

Let me talk about the different sampling st;ategies under the probability mode. They
are as follows:

3.1 SIMPLE RANDOM SAMPLING

In this approach, all elements are given an equal chance of being included in the sam­
ple. No one from the population is excluded from the pool.

This could be easily implemented if we have a homogeneous population. For instance,


if we are talking about beneficiaries of the UP Bliss Housing Condominium-Phase II,
we refer to a group of people who belong to a given socioeconomic standing. They are
middle income families since they are mostly employees of the University and affili­
ated with government agencies.

This mode is easily implemented if our population is geographically concentrated. In


the example above, they are located in a compact area adjacent to the UP Diliman
Campus in Quezon City.

It becomes a different matter altogether when we start to talk about all Bliss Program
beneficiaries. They are found in different parts of the country. Simple random sam­
pling is difficult to implement, as this would mean randomly selecting respondents
from a dispersed and large population. What do you think is the implication of this?
Yes, it is going to be very expensive as it means flying or sailing to different parts of the
country.
UP OPEN
UNIVERSITY How can random sampling be implemented?
SAMPLING TECHNIQUES/ Probability Sampling Strategies 161

The lottery method. This is common. Even laymen do this. In lotteries, the names of
players are deposited in a tambiolo. The names of winners are picked at ran.dam by
guests of prize-giving bodies.

In the lottery method of sampling, the names of the entire population are written on
pieces of paper and a given number of names (or sample elements) are drawn at ran­
dom.

However, this approach is quite tedious as it means repeating the listing of the names
in separate pieces of paper. If there are thousands involved, this could take some time.

Consulting the table of random numbers. If you already have a list of names of the
population from which you wish to draw your samples, you can:

1. Number them. If you have a total of 4000, you start with 0001 in the list.
2. Get a table of random numbers.
3. Close your eyes, move your pencil over the table, and then point. See where
your pencil stops.
4. If your pencil stopped on column 6, number 1 (91646), your first sample is
9164. Since you have a population of only 4000, remove the first digit (9) and
your first sample is: 9164.
5. Select the second sample. You may move downward or sideways. (I prefer to go
downward, so my next sample is: 8 919).
6. Proceed until you are able to get the total number of samples you need. If you
do not have enough, you can go back to number one. You may also try column
figures instead of rows.

UP OPEN
UNIVERSITY
Table 11-1. Table of random numbers

Line/ (1) (2) (3) (4 ) (5) (6 ) ( 7) ( 8) (9) (1)) (11) (12) (13) (14)
Col.

1 10480 15011 01536 02011 81647 91646 69179 14194 62590 36207 20969 99570 91291 90700
2 22368 46573 25595 85393 30995 89198 27982 53402 93965 34095 52666 19174 39615 99505
3 24130 48360 22527 97265 76393 64809 15179 24830 49340 32081 30680 19655 63348 58629
4 42167 93093 06243 61680 07856 16376 39440 53537 71341 57004 00849 74917 97758 16379
5 37570 39975 81837 16656 06121 91782 60468 81305 49684 60672 14110 06927 01263 54613
6 77921 06907 11008 42751 27756 53498 18602 70659 90655 15053 21916 81825 44394 42880
7 99562 72905 56420 69994 98872 31016 71194 18738 44013 48840 63213 21069 10634 12952
8 96301 91977 05463 07972 18876 20922 94595 56869 69014 60045 18425 84903 42508 32307
9 89579 14342 63661 10281 17453 18103 57740 84378 25331 12566 58678 44947 05585 56941
10 85475 36857 53342 53988 53060 59533 38867 62300 08158 17983 16439 11458 18593 64952
11 29818 69578 88231 33276 70997 79936 56865 05859 90106 31595 01547 85590 91610 78188
12 63553 40961 48235 03427 49626 69445 18663 72695 52180 20847 12234 90511 33703 90322
13 09429 93969 52636 92737 88974 33488 36320 17617 30015 08272 84115 27150 30613 74952
14 10365 61129 87529 85689 48237 52267 67689 93394 01511 26358 85104 20285 29975 89868
15 07119 97336 71048 08178 77233 13916 47564 81056 97735 85977 29372 74461 28551 90707
16 51085 12765 51821 51259 77452 16308 60756 92144 97735 53900 70960 63990 75601 40719
17 02368 21382 52404 60268 89368 19885 55322 44819 49442 65255 64835 44919 05944 55157
18 01011 54092 33362 94904 31273 04146 18594 23495 01188 85030 5!132 01915 92747 64951
19 52162 53916 46369 58586 23216 14513 83149 51851 71585 64350 94738 17752 35156 35749
20 07056 97628 33787 09998 42698 06691 76988 59193 23495 46104 88916 19509 25625 58104
21 48663 91245 85828 14346 09172 30168 90229 13602 51851 22178 30421 61666 99904 32812
22 54164 58492 22421 74103 47070 25306 76468 26384 59193 06646 21524 15227 96909 44592
23 32639 32363 05597 24200 13363 38005 94342 28728 35806 06912 17012 64161 18296 22851
24 29334 27001 87637 87308 58731 00256 45834 15398 46557 41135 10367 07684 36188 18510
25 02488 33062 28834 07351 19731 92420 60952 61280 50001 67658 32586 86679 50720 94953

Source: Agresti and Finlay 1986.


SAMPLING TECHNIQUES/ Probability Sampling Strategies 163

A basic requirement for this method is a sampling frame for the total population. If
there are 4000 persons in the list, all names should be included in the sampling frame.

It also necessary to consider the sample size. The size can be computed using the
Lynch et al. (1974) Formula (below). The formula gives a 95-percent reliability in
estimating the sample size. It gives you the opportunity to adjust sample size accord­
ing to your resource capability. You may estimate how much you can feasibly accept by
adjusting the sampling error (d) that you are willing to commit. The sampling error
(.025, .05 and .10) simply means the level of error you are willing to consider. An error
of .1 (if multiplied by 100) means a IO-percent margin of error that the samples are
representative of the population.

The Lynch et al. formula is:


NZ 2 • p(l-p)
n=-------
Nd2 + Z 2 . p(l-p)

Where: Z = 1.96 (the value of the normal variable for a reliability level
of 0.95. This means having a 95% reliability in obtaining
the sample size.)
p .50 (the proportion of getting a good sample)
1-p .50 (the proportion of getting a poor sample)
d .025 or .05 or .10 (your choice of sampling error)
N population size
n sample size

Given the above formula:

If: N = 3000
d = .05

Then:
3000 (l.96) 2 X .50 (1 - .50)
n =
3000 (.05)2 + 1.96 2 X .50 (1 - .50)
3000 (3.8416) X .25
3000 (.0025) + 3.8416 X .25

11,524.8 X .25
7.5 + .9604

2881.2
8.46
340.56 or 341
UP OPEN
(Note: Figures in boldface type are constant figures.) UNIVERSITY
164 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

It is also possible to determine the sample size using Table 11-2. T his is based on the
same formula. N stands for the population, n for the sample size, column 1 for a sam­
pling error of .025, column 2 for .05, and column 3 for .1.

Table 11-2. Sample sizes (n) required for population (N) of various sizes
with a desired reliability of 0.95

n n n

N 1 2 3 N 1 2 3 N 1 2 2

50 48 44 33 1050 634 281 88 2050 899 324 92


100 94 79 50 llOO 652 285 88 2100 908 325 92
150 137 108 59 ll50 669 288 89 2150 918 326 92
200 178 132 65 1200 686 291 89 2200 927 327 92
250 216 151 69 1250 702 294 89 2250 935 328 92
300 253 168 73 1300 717 297 89 2300 944 329 92
350 287 183 75 1350 732 299 90 2350 952 330 92
400 320 196 77 1400 747 301 90 2400 960 331 92
450 351 207 79 1450 761 304 90 2450 968 332 92
500 381 217 81 1500 774 306 90 2500 976 333 92
550 409 226 82 1550 787 308 90 2550 983 334 93
600 436 234 83 1600 800 310 91 2600 991 335 93
650 462 241 84 1650 812 312 91 2650 998 336 93
700 487 248 84 1700 824 313 91 2700 1005 336 93
750 5ll 254 85 1750 836 315 91 2750 1012 337 93
800 533 260 86 1800 847 317 91 2800 1018 338 93
850 555 265 86 1850 858 318 91 2850 1025 339 93
900 576 269 87 1900 869 320 91 2900 1031 339 93
950 596 274 87 1950 879 321 92 2950 1038 340 93
1000 615 278 88 2000 889 322 92 3000 1044 341 93

Source: Lynch et al. 1974.


Note: The sample size is derived from the computation of Frank Lynch mentioned in an earlier part of this
module.
T he advantage of applying this formula is that the bigger the population, the smaller
the sample size. Often, there is a tendency for researchers to proportionately increase
the sample size according to the size of the population. In statistical parlance, however,
the sample size decreases in number as the population size increases, although the
reliability level is not reduced.

If you will notice in Table 11-2, the sample size required for a sampling error of .10 is
only 93. On the other hand, for the same sampling error with a population of 2000, the
sample size is 92. For a population of 1000, the sample size is 88.

UP OPEN You see how convenient it is to apply the formula? Try it! To practice the application of
UNIVERSITY the Lynch et al. formula, do Activity 11-1.
SAMPLING TECHNIQUES/ Probability Sampling Strategies 165

ACTIVITY 11-1

Compute the sample size (n) for a population of 5000, using a sampling error of
.10. In your computation, you should be able to get a sample size of 94.

Let us now continue with the second type of probability sampling strategy. This is:

3.2 SYSTEMATIC SAMPLING

This is sampling after every regular interval. This can be undertaken if the features of
the population normally characterize what would be applied to simple random sam­
pling. The key here is to:

Determine the population and prepare the sampling frame. Number each one.
If the population of households in a barangay is 750, start with 001, with the
head of the family as the respondent. Thus:

Juan de la Cruz - 001


Jose Castro - 002
Carlos de la Fuente - 003

Mariano Yulo - 750

Determine the sample size.

If: N = 750
d = .10
n = 85

Determine the interval.


N (population size)
Interval (I)= ------­
n ( sample size)

If: N = 750
n = 85
750
Then I=
85
= 8.8 or 9

The interval is 9.
UP OPEN
UNIVERSITY
166 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Randomly select the first number. If the random start is 456, the next sample is
465, then 474, and so on. When you have exhausted the numbers in the -list,
continue counting from the beginning of the list. Another way is to pick an­
other number as the next random start and then select the elements in the
interval until all the samples are determined.

But why apply the systematic sampling strategy? Well, it is obviously more convenient
than simple random sampling. You identify only one random number to start with.

However, this approach and the first one may not be applicable if it is important to get
representation for particular subgroups of the same population that have different sam­
ple sizes. For example, if we wish to determine the commitment of legislators to social
development issues, it may be difficult to just rely on simple random or systematic
sampling strategy. If it is important to make a distinction between senators (24 in all)
and congressmen (numbering about 250, at most), applying the first two methods will
not assure adequate representation, specially for senators. In fact, you may not even
get any senator in the list, if their names are interspersed with those of congressmen.

If it matters a lot to obtain the views of the senators, not only congressmen, the next
strategy can be implemented.

3.3 STRATIFIED SAMPLING

This entails subdividing the population according to a certain characteristic, then se­
lecting the samples from every subgroup or stratum. This is resorted to when it is
important to get responses per subgroup or stratum.

To be able to implement this, you have to:

Obtain the sampling frame per stratum.


Determine the sample size per stratum.
Apply either the simple random or stratified sampling strategy per stratum.

lq the case of the legislators, this means grouping legislators into senators and con­
gressmen and getting the sample size per group. The sample size of congressmen with
a sampling error of .1 is 69. This process may be summarized in the following manner:

Senators Congressmen
N = 24 N = 250

n = 24 n = 69 (with d of .1)

Stratified sampling is useful if there is a need to differentiate the characteristics or a


heterogeneous population. It becomes complicated if the population with heterogene­
UP OPEN ous characteristics is spread out in various geographical areas. For instance, if an agri­
UNIVERSITY cultural program targets small and medium level farmers in a barangay, stratified sam-
SAMPLING TECHNIQUES/ Probability Sampling Strategies 16 7

piing could easily be applied. This may be difficult to do if the program is implemented in
different parts of the country. It is going to be tedious to obtain a sampling frame per
stratum of the total number of respondents from the different parts of the country.

Two features of the sample make stratified sampling applicable:

if the population is heterogeneous and its characteristics need to be differenti­


ated for varying responses to the factors or variables to be studied; and
the elements are geographically concentrated in a given area.

3.4 CLUSTER SAMPLING

This option entails random selection of groups in a population who could serve as the
respondents of the study or from whom random samples could be drawn. This is best
applied if we are dealing with populations with homogeneous characteristics but who
are geographically dispersed in different parts of the country and getting all of them is
not necessary.

If a program targets small farmers from 20 depressed barangays in lahar-driven areas


in Pampanga, one way to simplify the process of drawing the samples is by randomly
selecting clusters (barangays) from the list. Random or systematic sampling could be
applied to select from the total population of clusters . The clusters that may be ob­
tained will depend on the resources available.

Cluster sampling differs from stratified sampling because all elements in the popula­
tion are considered in the latter but samples are drawn per group. In cluster sampling,
only the randomly selected groups are considered and are the bases for dr!lwing the
respondents. The other groups are excluded. These two could be differentiated in the
following manner:

.__ 1st cluster sample


1st stratum ___.

.__ 2nd cluster sample


Small farmers

Small farmers
m
20 barangays

Medium-level

,,
2nd stratum ___. farmers

UP OPEN
Stratified Sampling Cluster Sampling UNIVERSITY
168 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

It is possible to use a combination of approaches (multistage sampling) in determin­


ing the sample size. Based on our previous discussions on the choices that can be made
in the determination of the sample size, two important issues about the population
have to be considered in determining the sample size:

the characteristics of the population (homogeneous or heterogeneous)


geographic spread (concentrated or dispersed)

Table 11-3 can help us decide what combination of strategies can be applied based on
these two attributes. In this table, if the small and medium scale farmers who are
targeted for the agricultural program are based in two provinces each in Luzon, Visayas,
and Mindanao and established in the most depressed municipality per province, par­
ticularly in poor performing barangays, two attributes are manifested:

geographically dispersed
complex personal attributes

T he following strategies can be applied, given the table:

Stratify provinces according to the island groupings (Luzon, Visayas and


Mindanao).
Cluster sample a province per region, and cluster sample barangays depending
on the budget available.
Stratify small/medium farmers.
Determine the sample size per group.
Randomly select the farmers per group.

Table 11-3. Sampling strategies appropriate to particular features of the population

Personal Attributes Geographic Spread Sampling Strategies

Homogeneous Concentrated Simple or systematic


Dispersed 1. Cluster
2. Simple or systematic

Heterogeneous Concentrated 1. Stratified


2. Simple or systematic

Dispersed 1. Stratified
2. Cluster
3. Simple or systematic

Two levels of stratification are being made here because there is variation in culture in
the three island groupings and at the individual level (the small and medium farmers).
UP OPEN Cluster sampling can also be made at two levels because of the provincial groupings
UNIVERSITY which are homogeneous per region and the barangays under each province.
SAMPLING TECHNIQUES/ Probability Sampling Strategies 169

If there is no listing of small and middle level farmers, adjustments will have to be
made in the sampling approach. It could be a sample of household heads in the cluster
of barangays.

<
The following scheme can summarize the sampling strategies applied in various groupings:

Luzon Province 1 .. Barangay


small farmers

Province 2 .. 1, 2, 3, 4, etc.
Barangay
:
medium farmers

Visayas Province 3 .. 1,2, 3, 4, etc.


Barangay

Province 4 .. 1, 2, 3, 4, etc
Barangay

Mindanao Province 5 .. 1,2, 3, 4, etc.


Barangay

Province 6 .. 1, 2, 3, 4, etc .
Barangay
1, 2, 3, 4, etc. etc.

.. .. .. ..
Stratify Cluster Cluster Stratify Random
by sample a sample a farmers per selection
region province province barangay of farmers
per region per region per stratum

ACTIVITY 11-2

There are many possibilities in the real world for applying sampling strategies. To
give you a real flavor of this, read the article by Bautista on Primary Health Care in
Pasay City, Ph;Upp;ne Journal of PubUc Adm;n;stra6on, XXXIX, 3 July 1995.

It is time now for another SAQ.

UP OPEN
UNIVERSITY
1 70 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

SAO 11-1

Answer the following questions:

1. I need to collect data on a sample of residents of registered homes for


streetchildren to find out what services are offered and how satisfied they are
with these services. I obtain a list of all these homes which I number from 1 to
250. Beginning randomly, I choose every tenth on the list and end up with 25
homes. Then I get a list of all the children of these homes and get the total
number. From the total population of 25 homes, I get a simple random sample
using a sampling error of .1.

a. What sampling strategies did I use?

b. What is the unit of analysis? Who are the respondents/observation units?

2. In one of the big cities in the Philippines, a total of 5000 NGOs are active in
development work. Of the 5000, 10 are multinational funding institutions,
2000 are locally funded NGOs engaged in social development, 2500 are locally
based entities engaged in political advocacy activities, and the rest are in­
volved in a combination of activities of the last two types of organization.

a. What sampling strategies would you recommend if the objective is to de­


termine a profile of the accomplishments and future plans of these NGOs
from the point of view of their respective leaders?

b. Allowing for a sampling error of .1, what is the total sample you will
obtain?

UP OPEN
UNIVERSITY
SAMPLING TECHNIQUES/ Probability Sampling Strategies 171

ASAQ 11-1

ta re:io1
AJeJoApe 1eJq�1oct pue
08 06t :iuawcto1aAap 1e�Jos 4:ioq u� pa6e6ua
Z6 oosz AJeJoApe 1eJq�1oct u� pa6e6ua
Z6 oooz :iuawcto1aAap 1eqos u� pa6e6ua
Ol Ol suoqn:iqsu� 6u�pun1 1euoqeuqinw
u N

·pau �e:iqo aq WM uoqJe1sqes/saJ�A1as uo uoqew101u�


a4:i wo4M wo11 ua1pH4J:iaa1:is a4:i a1e s:i�un uoqeA1asqo/s:iuapuodsa1 a41

· (wa4:i o:i pa1aHo saJ�A1as


a4:i 4l�M uoqJe:}sqes 1�a4:i) ua1pH4J:iaa1:is a4:i pue (pau�e:iqo aq o:i s� pa1a1
-10 saJ�A1as uo uoqew101u� a1a4M) Al�HJE:} a4:i a1e s�sf:1eue 10 s:i�un a41 ·q

·sawo4 sz a4:i wo11 pau�e:iqo :is�1 a4:i wo11 ua1pH4J 10 uoq


-e1nctod 1e:io:i a4:i 10 6u�1dwes wopue1 a1ctw�s seM puoJas a41 ·osz 10 1e:io:i
e wo11 papa1as a1aM sawo4 sz aJu�s 6u�1dwes 1a:isnp seM da:is :is1�t a41 ·e ·i:

SUMMARY

This module emphasized that sampling is a convenient and economical mode of draw­
ing a representative number of cases that will enable the researcher to generalize about
the population. It can also assure accuracy.

It has also been shown that probability sampling, as one mode of sampling, can ensure
an objective process of drawing elements in the population, since each element is given
an equal chance of being selected.

Several options are available in probability sampling (i.e., simple, systematic, strati­
fied, and cluster). The choice of option depends on the characteristics of the elements
(i.e., homogeneous or heterogeneous) and how they are distributed in a given space
(i.e., concentrated or dispersed). A combination of these may be applied if the attributes
of the respondents allow.

On the whole, the methodologies discussed here make possible the identification of
respondents or subjects for quantitative studies.

We have not yet discussed in this module sampliPg strategies that do not apply prob­ UP OPEN
ability principle. This will be discussed in the next module. UNIVERSITY
1 72 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

REFERENCES

Agresti, Alan and Barbara Finley


1986 Statistical Methods for the Social Sciences. San Francisco, California: Dellen Publishing
Co.
Lynch, Frank et al.
1974 Data Gathering by Social Survey: Philippine Social Survey Series No. 2 Trial Edition.
Quezon City: Philippine Social Science Council.

UP OPEN
UNIVERSITY
MODULE12

NON-PROBABILITY SAMPLING
AND SAMPLING ISSUES

INTRODUCTION

T he previous module focused on the different approaches in probability sampling.


It highlighted the importance of making a careful selection of the unit of analysis
to ensure external validity. It stressed the importance of selecting the respondents
or participants, especially when dealing with a big population. Random selection
of the elements from the population is a critical requirement in the probability
sampling approach.

However, there are instances when the principle of probability is difficult to apply.
For instance, if we are dealing with a transient population, random sampling can­
not easily be done. To cite a more concrete example, it is difficult to make an as­
sessment of Light Rail Transit users because they are on the go. It is not easy to
capture their attention. Analyzing the behavior of those who lobby for particular
reforms in Congress raises the same problem, since this is also a very fluid group.

In such instances, the more appropriate sampling strategy is non-probability sam­


pling.

OBJECTIVES

At the end of this module, you will be able to:

Describe non-probability sampling as a mode of selecting elements of the


population who could serve as respondents of the study;
Identify the different approaches in non-probability sampling;
Define the common problems that may be encountered in sampling; and
Identify common measures to respond to these problems.
174 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

1.0 NON-PROBABILITY SAMPLING

By its title alone, non-probability sampling signifies its difference from probability
sampling. It entails the identification of participants in the study based on criteria
spelled out by the researcher, rather than through random selection.

As mentioned in the introduction to this module, this method of sampling is resorted


to when it is difficult to estimate the population of the study because they are mobile
or transitory in a given location.

This mode is also useful in exploratory studies oi.descriptive studies with a qualitative
thrust. In such studies, what is important is to characterize the focus of the study,
rather than to qualify it. Thus, the respondents are chosen on the basis of specific
criteria ( e,g., they are direct participants) formulated by the researcher, rather than
randomly selected. For example, if the purpose of a study is to assess a disaster man­
agement program, then it is necessary to visit the shelters of lahar-driven families of
Pampanga. In understanding the dynamics involved in the introduction of the Mini­
mum Basic Needs approach in the Social Reform Agenda, the persons who became
involved in it can be interviwed.

Non-probability sampling is definitely more economical and easier to implement than


probability sampling. This is because no restrictions on numbers are posed on the
researchers. In the first place, there is no population estimate. Second, the primary
intention of the researcher is to draw qualitative information from the most knowl­
edgeable, instead of counting the number of persons affected by or manifesting a par­
ticular variable. For example, determining the peculiar types of corruption manifested
by a regulatory institution is best based on interviews not of many individuals but of a
few who are both willing to talk and knowledgeable about the topic. Here is where the
Filipino approach to doing research-pakapa-kapa muna (feeling your way through)
before interviews are made-becomes very helpful.

UP OPEN
UNIVERSITY
SAMPLING TECHNIQUES/ Non-Probability Sampling 175

However, this type of sampling cannot be used to estimate the parameters or the
population from sampling statistics (O'Sullivan and Rassel 1995: 126). It is not
possible to apply statistical theory to generalize to the population or to prove the
accuracy of the sample.

2.0 TYPES OF NON-PROBABILITY SAMPLING

2.1 ACCIDENTAL OR CONVENIENCE SAMPLING

This is implemented by seeking out elements who are readily available cir accessible to
respond to our questions. In other words, the first person who comes along who typi­
fies the unit of analysis serves as the respondent of the study.

If the thrust is to obtain respondents from a group of lahar-driven families, conven­


ience sampling is done by having a researcher approach the first household that he/she
comes upon and all others within easy reach.

In determining the attitude of training participants to the teaching of introductory


topics in Filipino, the researcher interviews participants whom he/she finds in the
cafeteria. He/She goes to the same place until the required sample size is attained or
until he/she has interviewed the required number of participants.

In instances where respondents are difficult to find since they are busy, accidental
sampling could be the only recourse. For instance, to find out how passengers assess
the Light Rail Transit, it may be difficult to expect success from systematic random
sampling. While a researcher could estimate the number of riders who usually pass
through during peak periods in a given outlet (let us say the terminal in Baclaran), he/
she may not manage successful interviews during this period (say 8:00 to 9:00 a.m.)
because the target respondents are rushing to work and may not be able to spare him/
her/ the time to respond. At best, the researcher would be left with respondents who
are willing to spare a few minutes for the interview, such as the unemployed or elderly
who are not rushing to work.

2.2 PURPOSIVE SAMPLING

This is implemented with the researcher defining a criterion or set of criteria for de­
termining the respondents of the study. It is the researcher's judgment that becomes
the basis for selecting an element or group that will serve as the unit of analysis.

Purposive sampling is a useful methodology in qualitative or exploratory studies, since


the qualities of the key person or informant are spelled out by the researcher. The
objective is not to have many respondents but to make sure that the person who would
be interviewed will provide a wealth of information. The aim is not to quantify but to
characterize an event being studied. For example, to find out how a program was insti­
tuted in an area, one does not need to interview many program implementors; it is
necessary to interview only the few who paved the way for its introduction in the area. UP OPEN
UNIVERSITY
176 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Purposive sampling can also be the basis for identifying groups that can serve as the
unit of analysis from which a smaller sample of respondents could be drawn for an
interview.

In the study of the Area-Based Child Survival and Development Program, identified
as the unit of analysis to represent the program areas were:

three barangays that were introduced to the program first, had the lowest qual­
ity of life indicators, and were based in a municipality also with the lowest
quality of life indicators per province (in s.even target areas). (Quality of life is
measured by morbidity, mortality, nutritional condition of the children, fe­
male functional literacy rate, and economic profile.); and
barangays that were not affected by open conflict (i.e., where the security of
the researchers could be assured.

Comparison barangays (those not exposed to the program) were drawn from the same
municipality where the pilot barangays were selected. They met the following criteria:

they had the lowest quality of life indicators in the municipality;


they were not physically adjacent to the pilot barangay, thereby avoiding the
contamination effect; and
they did not have a peace and order problem.

From the set of purposively selected barangays, randomly selected decision makers on
health in the family were identified.

In this example, a combination of probability and non-probability sampling strategies


was applied. The purposive sampling of barangays was comparable to cluster sampling
with the exception of the pre-identification of the groups by the researcher. However,
the final elements of the study were randomly selected. Thus, the findings of the
study could be generalized to the population, unlike in a situation where the respond­
ents were drawn through accidental sampling.

In short, there are advantages to using purposive sampling. Real conditions could con­
strict the application of random sampling strategies as in the study of the ABCSDP
where the targeted provinces were areas in Mindanao where armed conflict occasion­
ally erupted. In this case, the use of purposive sampling meant the security of the
interviewers was assured.

2.3 QUOTA SAMPLING

This is somewhat similar to stratified sampling except that the selection of the ele­
ments per stratum is not done through the application of random sampling strategy.
Quota sampling entails grouping elements according to certain characteristics and
ensuring that each group is represented.

UP OPEN In the case of Light Rail Transit (LRT) commuters, accidental sampling can be re­
UNIVERSITY fined by having respondents drawn by sex, since males and females could have differ-
SAMPLING TECHNIQUES/ Non-Probability Sampling 177

ent problems or issues to raise. For instance, females may want to say something about
congestion as a basic issue. Age could also be a stratifying factor since older people
may find the LRT inaccessible due to the need to go up the stairs.

After having estimated the population by gender and by age using data obtained from
those selling the tickets, the ideal size per group can be determined using the formula
or the table. Convenience sampling can be done thereafter, if random sampling is not
feasible.

Quota sampling is very helpful especially if the sampling frame is not available per
group or stratum. It refines the application of convenience sampling .since there is
conscious intent on the part of the researcher to view the probable differences of every
stratum or group with regard to the critical variables of the study.

2.4 SNOWBALL OR REFERRAL SAMPLING

This involves having a respondent refer other people who are in a position to answer
some of the questions of the researcher.

This is particularly helpful in the study of highly sensitive topics where the identity of
respondents is difficult to divulge or may even be unknown to many. In other words, if
the sampling frame can not be provided and the topic has security implications, a
researcher could obtain referrals from the first respondent to other respondents who
may be willing to talk. For example, it may be difficult for a researcher to gain access to
the mechanisms or processes of corruptive behavior. Thus, clients who are willing to
share their experience may also refer the researcher to friends who they think can
express their views on the topic.

The deficiency of the approach, however, though is that the referral could have the
same characteristics as the first respondent. Referrals could be engaged in a type of
activity such as the respondent is engaged in. If there is no list of participants since
only an informal network of interested persons meet on the issue, the snowball tech­
nique may lead to the identification of peers with the same commitments. However,
this could be balanced by a conscious intent on the part of the researcher to create a
different listing to represent other stakeholders with a different point of view.

3.0 POSSIBLE PROBLEMS REGARDING THE SAMPLING FRAME

We have seen that non-probability sampling is sometimes resorted to because of the


difficulty of defining the sampling frame. Let me now talk about other problems that
may be encountered in the formulation of the sampling frame. Some of the usual prob­
lems encountered by those who are engaged in probability sampling are the following:

1. Missing elements. Sometimes, sampling frames may not contain an updated list
of the population. There may be "missing elements" �n the sense that the names of
persons who should be in the list are not recorded. It is therefore important for a UP OPEN
researcher to check when the listing was updated. UNIVERSITY
178 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

In evaluation studies of programs and projects, a popular recourse is to draw from


a list of residents or to rely on a spot map of households in the barangay. However,
it is possible that this listing is not updated.

2. Intrusive elements. These are elements that may be included in the list but which
no longer have a right to be there. For example, the name of a family may be
included in the sampling frame for a a barangay although they have already mi­
grated to another locality.

3. Double listing. Errors in the sampling frame may also be experienced if the name
of a respondent is double-listed. For example, ihe name of a single female who gets
married may be listed both under the family of origin and the family of procrea­
tion.

4.0 HOW TO RESOLVE SAMPLING FRAME ISSUES

How do we respond to these problems with sampling frames? Some of the possible
solutions include:

1. Correct the listing. This would mean doing a census of the population involved.
This is a costly solution. However, if a big percentage of the sampling frame is
being doubted, this could be a necessity.

I recall how important this approach was when we conducted the study on ABCSDP.
In one of the barangays in Tawi-tawi, the area research leader wondered why the
sample of households, which was determined through systematic sampling, was
bigger than the total population of households. It turned out that the population
size was purposely increased for electoral purposes. Thus, the area leader had to do
an actual survey of the households based in the barangay before the samples could
be drawn.

2. Report the problem. Another possibility is to report the problem without correct­
ing the sampling frame. The report should indicate what percent of the population
is affected by the problem.

3. Get a substitute. A double listing could be substituted by another randomly se­


lected element. An intrusive element could also be replaced by an element that is
not intrusive. However, whatever approach the researcher opts to use should be
accounted for in the report.

Finally, let us not discount the fact that if there is a serious flaw in the sampling frame
and it cannot be depended upon, non-probability sampling is an alternative. Bear in
mind, however, that a study based on non-probability sampling will not be able to
make generalizations about the population.

UP OPEN
UNIVERSITY
SAMPLING TECHNIQUES/ Non-Probability Sampling 1 79

5.0 COMBINING PROBABILITY AND NON-PROBABILITY SAMPLING

In instances where probability sampling cannot be applied in various groups because


of the sensitivity of a topic, non-probability sampling may be resorted to. However, it
is important to define the criteria for preferring certain groups over others. Then, the
final elements of the study could be based on probability sampling strategy.

I recall a study in 1976 on voters' assessment of the referendum as a means of consult­


ing the people on certain issues under the dictatorship. Since barangay captains might
not be willing to have the study conducted in their areas, the research team of the UP
College of Public Administration decided to negotiate with barangay: captains who
were willing to have their areas covered. At the same time, the team saw to it that
Luzon, Visayas and Mindanao, and both urban and rural areas in each of these islands,
were represented. The final selection of the respondents was made through systematic
sampling of houses where heads of families were identified as respondents.

If this approach is followed, it is still possible to apply probability sampling statistics


since the selection of final elements upholds the principle of objectivity.

SAO 12-1

Identify the sampling strategy (i.e., probability or non-probability) applied in the


following cases. You may also indicate the specific types under each sampling
strategy resorted to:

1. A congresswoman reports that letters to her office have indicated that 4 out of
5 are in favor of stricter gun controls. She concludes that 80% of her constitu­
ents would like her to endorse restrictions on ownership of guns.

2. A total of 12 programs were implemented to train Department of Health repre­


sentatives to the Local Health Board on the value of the community-based
approach. The 12 programs had a total of 360 participants. To assess the
utility of the program to their respective areas, 75 respondents were obtained
by picking every fifth participant in the list, beginning with a random start.

3. A researcher would like to make an assessment of the contributions of NGOs to


the formulation of social welfare policies in the Philippine legislature. Since
contributions of NGOs in this field are undocumented as interactions occur
around coffee tables and mass actions, the first source of information was one
identified by the Secretary of the Department of Social Welfare and Develop­
ment and the chairperson of the committee on social welfare.
180 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

ASAQ 12-1

1. This is a non-probability sampling strategy since the samples are not ran­
domly identified. The specific type used is convenience sampling.
2. This is probability sampling, using a systematic sampling approach.
3. This is non-probability sampling, using the snowball technique.

ACTIVITY 12-1

Be ready to identify your unit of analysis and sampling strategy in your research
proposal.

SUMMARY

This module has indicated the different options available in applying non-probability
sampling. While it is an approach that prevents one from generalizing to a population,
a researcher may resort to it when the real situation prevents probability sampling and
when the sampling frame is not available. It is also resorted to when the thrust of the
study is qualitative in nature. This approach is less costly than probability sampling.

Four modes of non-probability sampling are discussed here. One is convenience sam­
pling, characterized by selection of elements according to availability of the respond­
ents.

Another is purposive sampling. It is based on criteria specified by the researcher.

Quota sampling, another approach, is similar to stratified sampling in that the target
population is classified according to attributes of the population, which requires dis­
tinct representations per group.

Snowballing is another mode. It facilitates identification of samples known to an ear­


lier respondent. This mode is useful if the topic is sensitive and requires willingness
on the part of the elements to participate in the study.

In probability sampling, problems may be encountered regarding the sampling frame


UP OPEN such as: missing elements, intrusive elements, and double listing. These could be re­
UNIVERSITY solved by correcting the listing, reporting the problems, or getting a substitute.
SAMPLING TECHNIQUES/ Non-Probability Sampling 181

Probability and non-probability sampling strategies may be considered in some in­


stances. Identifying the final elements or respondents of the study through prob­
ability sampling procedures enables the researcher to generalize findings to the
population.

REFERENCES

Bautista, Victoria
1992 Towards the Institutionalization of ABCSDP. Highlights of a Feasibility Assessment.
Philippine Journal of Public Administration XXXVI, 3 (July).
Judd, Charles M., Eliot R. Smith and Louise Kidder
1991 Research Methods in Social Relations. Fort Worth: Holt, Rinehart and Winston.
O'Sullivan, Elizabethann and Gary R. Rassel
1994 Reserch Methods for Public Administration. z nc1 Edition. White Plains, New York:
Longman Publishers.

UP OPEN
UNIVERSITY
TECHNIQUES
OF DATA COLLECTION
Dear Student,

An important step in research is the selection of the appropriate technique and the
corresponding tool for gathering data. This unit addresses this concern.

Four modules comprise this unit. Three of these (Modules 13, 14 and 15) cover pri­
mary techniques of data collection. These are the techniques in which the researcher
has an active role to play in gathering information from those who have knowledge or
actual experience of the factor or variables being studied. Module 13 focuses on direct
observation. Module 14 discusses interviews, questionnaires, and indirect methods of
gathering information. Module 15 provides basic instructions on how to formulate
continuous types of scale for measurement.

Module 16 gives you an overview of and the techniques for utilizing data obtained
from reports prepared by other entities. This module focuses on secondary techniques
of data collection.

After completing this unit, you should be able to formulate your own instrument for
data collection for your own research proposal.
MODULE 13
PRIMARY TECHNIQUES:
DIRECT OBSERVATION

INTRODUCTION

Two important aspects of the methodological requirements of a research problem.


have so far been discussed. They are:

defining the research design; and


determining the unit of analysis and whether all or some of the elements of
the unit of analysis will be dra wn to serve as the respondents or subjects of
the study.

A third aspect of the methodological component of the research process is dealing


with the various techniques for gathering information. A basic concern is what ap­
proaches and tools will be adopted in order to facilitate the conduct of observation.

Two broad types will be discussed under techniques of data collection. These are
the primary techniques and secondary techniques.

This module focuses on one primary technique-direct observation. Subsequent


modules will discuss other primary techniques and secondary techniques.

OBJECTIVES

At the end of this module, you will be able to:

Define primary techniques for gathering information;


Describe direct observation as a primary technique;
Distinguish bet ween structured and unstructured observations;
188 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Define the relationship of the researcher with the persons observed;


Determine the implications of participant observation and detached observa­
tion; and
Enumerate the important guidelines for keeping a diary.

1.0 PRIMARY TECHNIQUES OF DATA COLLECTION

One of the basic approaches to gathering data may be labeled as primary techniques.
They are characterized by the researcher's direct interaction with the persons who
can provide information on the topic through intewiews, questionnaires, and focus
group discussions. Information could also be derived through direct observation
where the persons or entities being studied do not necessarily interact with the re­
searcher at all times. They may either be conscious or unaware that information is
being collected about their behavior by the researchers.

Direct observation is discussed in the following sections.

2.0 DIRECT OBSERVATION DEFINED

One of the basic approaches to gathering information is direct observation. The re­
searcher witnesses the event in the natural setting and can thus give a firsthand ac­
count of the event. The observation is not mediated by other persons who have wit­
nessed the event, as is normally the case in interviews and questionnaires where the
personal experience of the respondents is communicated to the researcher.

For example, in the study of graft and corruption in a regulatory institution, the UP
CPA team deployed a research assistant who posed as an applicant to witness what
happens in processing a paper for a franchise to run a taxicab business. This enables
the researcher to determine firsthand what specific bureaucrats expected the applicant
to pay beyond the official fees set by the agency. This technique of gathering informa­
tion was resorted to because it would have been difficult to ask questions about the
"going rate" in processing a paper, apart from the normal fees which should be paid.

What are the strengths of direct observation as a method?

First, in the application of this method, there is less distortion on recall a� the event
is witnessed as it happens. When interviews are conducted to make respondents re­
count what has transpired, let us say as a result of a program, the account hinges on
their recollection of the events, which may be limited or incomplete or even inaccu­
rate. In direct observation, the event unfolds before the researcher's eyes. He/She
witnesses the event him/herself.

UP OPEN
UNIVERSITY
DATA COLLECTION/ Direct Observation 189

Another strength of direct observation is its applicability to studying individuals or


groups who are not able to communicate orally or in writing. For instance, the method
may be used to find out about the responsiveness and sensitivity of caretakers of an
institution for the deaf and blind. This method is also appropriate for a study of how
caretakers deal with infants or babies in health facilities unless researchers will de­
pend on the personal assessment of workers of their own behavior or those of their
peers, which may be biased.

Direct observation may also be implemented in instances where there is resistance


to interviews and/or questionnaires. This may be experienced when the topic at hand
has moral implications or could affect the prestige or reputation of the person or insti­
tution being studied. An example is the study of graft and corruption in which both
bureaucrats and clients may be unwilling to participate as respondents.

However, researchers have an important role to play in protecting the identity of the
subjects involved. They have a moral obligation to maintain the anonymity of the
subjects. They are not there to tell on these behaviors to authorities.

There are areas or experiences not accessible to direct observation. It may be easy
to witness, for instance, whether a family is able to maintain environmental sanitation.
(I recall one researcher who, instead of asking whether a rural household has a toilet,
would request to use the bathroom.) However, there are some private or sensitive ex­
periences that may not be open to the researcher to witness. For instance, one cannot
find out by direct observation what family planning techniques are adopted by mar­
ried couples of reproductive age.

Direct observation also does not provide access to sensitive positions or functions. For
example, in the study of graft and corruption, only over-the-counter transactions may
be observed directly. Backroom transactions and corruption at the higher levels of
government are not so readily seen or observed.

Another weakness of direct observation is that it is limited to the "here and now."
Information can be gathered only on areas or events that can be directly witnessed by
the observer. Direct observation techniques cannot capture the historical context of
the event. Neither can they anticipate the future since they are confined to the current
situation.

Also, direct observation may not be able to capture the inner feelings of subjects
being studied, unless the subjects of the study express them in words or act them out.
Still, the interpretation of actions and expressions requires cultural sensitivity on the
part of the researcher. Researchers have to be knowledgeable of the meaning of out­
ward behavioral manifestations of feelings.

UP OPEN
UNIVERSITY
190 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

3.0 TYPES OF DIRECT OBSERVATION

Direct observations may be structured or unstructured.

3.1 STRUCTURED OBSERVATION

In structured observations, the variables or factors for the study are more delineated.
The researcher has a clear idea of what factors to study and in what context or event
this will be observed.

The role of the researcher is to ascertain some key points or issues in conducting the
observation such as:

variables
participants to be observed or the artifacts to be assessed
event to be observed
how the variable (act) is measured, or the operational definition

For example, in assessing a program one administrative issue that must be ascer­
tained is the demeanor of the leader. Is the leader one who encourages involve­
ment in decision making? In this research problem, the following key issues need
to be addressed:

Identification of the variable to be subjected to observation: participativeness


of the leader in the organization
Identification of the participant: The researcher proceeds by spelling out who
serves as the unit of analysis (or the participants to be directly observed). In
this case, it is the leader of the institution or entity.
Selection of the critical event when participativeness could be witnessed: Sup­
pose there is one critical period in the organization where most decisions are
made. This could be in the meeting in December where there is a review of the
policies of the organization. Hence, the event to be observed is the meeting in
December when policies are reviewed.
Formulation of the operational definition or the scheme for measuring the
variable: Some categories for analysis can be defined by the researcher to de­
scribe the quality of management manifested by the leader, such as:
Top-down-The leader decides on what policies to pass without any
deliberation with the staff.
Consultative-The leader discusses issues and draws comments on possi­
ble solutions that could be adopted, but eventually makes the choice on
the courses of action to be adopted.
Participative-The leader facilitates the discussion of issues, draws opin­
ions on how to solve issues, and obtains the staff's opinion on the courses
of action to take.

Direct observation may also be combined with other techniques of data collection,
UP OPEN particularly interviews and focus group discussions. In this case data on some vari­
UNIVERSITY ables derived by structured direct observation could also be subjected to quantitative
190 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

3.0 TYPES OF DIRECT OBSERVATION

Direct observations may be structured or unstructured.

3.1 STRUCTURED OBSERVATION

In structured observations, the variables or factors for the study are more delineated.
The researcher has a clear idea of what factors to study and in what context or event
this will be observed.

The role of the researcher is to ascertain some key points or issues in conducting the
observation such as:

variables
participants to be observed or the artifacts to be assessed
event to be observed
how the variable (act) is measured, or the operational definition

For example, in assessing a program one administrative issue that must be ascer­
tained is the demeanor of the leader. Is the leader one who encourages involve­
ment in decision making? In this research problem, the following key issues need
to be addressed:

Identification of the variable to be subjected to observation: participativeness


of the leader in the organization
Identification of the participant: The researcher proceeds by spelling out who
serves as the unit of analysis (or the participants to be directly observed). In
this case, it is the leader of the institution or entity.
Selection of the critical event when participativeness could be witnessed: Sup­
pose there is one critical period in the organization where most decisions are
made. This could be in the meeting in December where there is a review of the
policies of the organization. Hence, the event to be observed is the meeting in
December when policies are reviewed.
Formulation of the operational definition or the scheme for measuring the
variable: Some categories for analysis can be defined by the researcher to de­
scribe the quality of management manifested by the leader, such as:
Top-down-The leader decides on what policies to pass without any
deliberation with the staff.
Consultative-The leader discusses issues and draws comments on possi­
ble solutions that could be adopted, but eventually makes the choice on
the courses of action to be adopted.
Participative-The leader facilitates the discussion of issues, draws opin­
ions on how to solve issues, and obtains the staff's opinion on the courses
of action to take.

Direct observation may also be combined with other techniques of data collection,
UP OPEN particularly interviews and focus group discussions. In this case data on some vari­
UNIVERSITY ables derived by structured direct observation could also be subjected to quantitative
DATA COLLECTION/ Direct Observation 191

measure. For example, in a study .on primary health care, some variables depended on
what was observed by the interviewers. The interview schedule included observational
items such as (Bautista 1989: 168-169):

Manner of maintaining animals, if any:


» All/some fenced in or caged
» All/some tied to posts
» Free to roam around in the yard
» Free to roam around in the house

Presence of stagnant water in the yard


»No
» Yes

Presence of animal waste and/or garbage in the yard


»No
» Yes

In general, artifacts or physical entities are easier to observe than people. With people,
there are ethical considerations involved especially if observation infringes on their
right to privacy.

3.2 UNSTRUCTURED OBSERVATION

Unstructured observation is more flexible than structured observation. Due to the


lack of specificity in the variables for the study, the researcher captures as much as he/
she can in the setting and the event being witnessed. This is particularly helpful when
the study is still in the exploratory phase and the researcher has not yet decided on the
variables to be investigated.

The focus of observation includes:

setting
participants
social behavior manifested
frequency or duration of the behavior

An unstructured observation can be the starting point in the conduct of structured


observation. The former can provide insights that may help the researcher spell out
what specific variables to measure. A researcher who is not yet familiar with the set­
ting of the study will find getting the "feel" of the setting ( the Filipino term is
pakikiramdam) very important. It is foolhardy to get into a situation one is not familiar
with. Unstructured observation enables one to find out more about a situation-at UP OPEN
least enough about it to enable one to decide what to focus on. UNIVERSITY
192 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

4.0 ROLE OF THE OBSERVER

Two important issues confront the researcher conducting direct observation. These
are:

How to relate with the subjects or participants of the study: Will one reveal
one's purpose or keep it concealed?
Will one be a participant observer or a detached observer?

Revealing one's purpose to the subjects or groups who will be covered by the observa­
tion gives the researcher the advantage of gaining access to more information than if
he/she would get by just looking around, without assistance from those being observed.
When a team from the UP CPA undertook the study of corruption in a regulatory
institution, some persons who had a credible reputation knew about the purpose of the
study and provided memoranda issued by top policy makers to protect the interest of
some business enterprises. Thus, the researchers found out that a ban on dilapidated
taxicabs was issued because the market was already saturated by a particular brand of
cars.

However, revealing the purpose of the observation may create reactivity on the part of
the subjects of the observation. Being aware that they are being observed could lead
them to put their best foot forward, so to speak. This is a normal reaction. They would
not wish to have their reputation compromised. For instance, if executives knew that
the research is about participativeness, then they may try to exude a behavior that
would depict this. Hence, concealing the purpose of one's observation may be the best
recourse if the behavior of the subjects of the study is likely to be affected by the
knowledge that they are being studied.

Researchers also have to make a decision on whether or not they will get involved in
UP OPEN the activities being undertaken by the subjects or participants of the study. Serving as
UNIVERSITY a detached observer has the advantage of making the observer maintain objectivity.
DATA COLLECTION/ Direct Observation 193

Being uninvolved in the activities of the social unit being observed may prevent one
from identifying with the feelings and experiences of those engaged in the activity.
There is also no need to learn new roles. Being involved in the activities of the obser­
vation unit may exert pressure on the observer to perform well in the roles he/she is
assuming in the organization. This could intervene with recording information.

On the other hand, serving as a participant observer may enable one to have firsthand
knowledge of the activities to be pursued. There is an opportunity to "learn and feel"
what participants are going through. For example, posing as an applicant for a fran­
chise to operate a taxicab may enable the researcher to directly witness how bureau­
crats send the message if they expect additional money to be paid. This may not be
directly witnessed if the observer is only observing "from a distance."·

5.0 KEEPING A DIARY

If interviewers use interview schedules, observers make use of the diary for recording
their observations, specially for unstructured observations. Structured types of obser­
vation have well-delineated variables that can be incorporated in a checklist by the
observer. This is the instrument used in direct observation.

In the diary, record as much information as you can. But make sure to distinguish
your observation from the interpretation you are making of what you see or hear.
Make markings in the diary of the observations you have made vis-a-vis your insights
or interpretations of the facts you have obtained. The observations must also be as
specific as possible; avoid generalized statements. Instead of making judgments that
a person is happy or sad, the notes can indicate how the person looked and behaved.
An example of two types of observations are presented below. This is based on M. Q.
Patton's book (Qualitative Evaluation and Research Methods, 1990), as cited by Gay (1996:
225).

Vague: The next student who came in to take the test was very poorly dressed.

Detailed: The next student who came into the room was wearing clothes quite
different from [those of] the three students who'd been in previously. The three
previous students looked like they'd been groomed before they came to the
test. Their hair was combed, their clothes were clean and pressed, the colors of
their clothes matched, and their clothes were in good condition. This new
student had on pants that were soiled with a hole or tear in one knee and a
threadbare seat. The flannel shirt was wrinkled with one tail tucked into the
pants and the other tail hanging out. Hair was disheveled and the boy's hands
looked like he'd been playing in the engine of a car.

Notice the difference between the first and second account. The first one made a judg­
ment or interpretation. The second described what was seen in terms of the appedr­ UP OPEN
ance of the students who came into the room. UNIVERSITY
194 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Another helpful tip: Do not make yourself too conspicuous when recording or jot­
ting down information. If recording on the spot could affect the performance of the
subjects of the study, record the activities later but do so immediately after your obser­
vation is over so you do not forget what transpired.

You may use additional paraphernalia such as a tape recorder, camera, or video cam­
era, to help record what transpired.

It is time to evaluate how much you have learned from this module.

SAO 13-1

In the following situations, indicate what type of direct observation (structured or


unstructured) you would recommend and why. Will you conceal or reveal the pur­
pose and will you serve as a participant or as a detached observer?

1. A researcher is interested in determining whether drug addicts placed on reha­


bilitation are responding favorably to the activities undertaken by the institu­
tion, which include: spiritual advocacy, socials, in-house seminar on drug abuse,
and training for income generating projects.

structured unstructured
reveal purpose �� conceal purpose
participant observer detached observer

2. A researcher won a bid to assess how local executives in Metro Manila manage
their meetings in local development councils. The researcher is not yet certain
about what factors to focus on in the study and is kind of feeling his way
through the process.

structured unstructured
reveal purpose �� conceal purpose
participant observer detached observer

3. Many complaints had been aired about the Bureau of Internal Revenue in
Quezon City for helping companies to avoid paying a big amount if they gave
a certain fee to an examiner to fix information in the computer. Examiners do
this by making it appear that they earned less from their sales. You have
access to a company that is willing to make you pose as their representative
for you to learn more about this "negotiation" with the BIR.

structured unstructured
reveal purpose �� conceal purpose
participant observer detached observer
UP OPEN
UNIVERSITY
A1.1S�3J\1Nn
N3d0 dn

"J\Je�p uMo moA u� suoqeAJasqo moA uMop


al�JM ·wooJ a4l ap�su� a1doad a4l JO JO�Ae4aq a4l uo smoJ pue salnu�w Sl JOJ
uMop l�S ·4l�M JP�HWP:J OOl lOU aJe noA lP4l aJ�HO JnoA u� l�un Je1nJqJed e aAJasqo

t-£t All/\IlJV

laJPJ noA p�p MOH

·a1npow s�4l Ma�/\a� - s ue4l ssa1


·os-os - ·sld s
·poo9 - ·sld L-9
jpoo9 AJaA - ·sld 6-8

:lo6 noA H
·slu�od aaJ4l JO lPlOl e se4 wal� 4Jea 'sn41 ·JaMsue paJJOJ 4Jea JOJ lU�od auo a/\�9

1. I would apply structured observation. I will conceal the purpose from the
participants because their behavior would be affected by their knowledge that
their response to the activities is being evaluated. It is not necessary to serve
as a participant observer since information may be obtained without going
througb the different activities.
2. I would' conduct an unstructured observation first since I am not yet sure of
what to do in this case. I will be a detached observer initially. I will talk about
my purpose if it will not affect the local executive.
3. Structured observation can be undertaken since I know what to observe. The
purpose can be concealed since revealing it may lead to reactivity if they
know my purpose. I can be a participant observer since the company is allow­
ing me to do that.

t-£t ovsv

S6 l UO!lEAJasqo l)aJ!O / NOIJ.)3110) VJ.VO


196 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

SUMMARY

This module focuses on direct observation as a mode of gathering information. Direct


observation is witnessing events as they happen without mediation of the participants
or subjects of the study.

Its advantages include: less distortion on recall, applicability in studying individuals/


groups who are not able to communicate orally or by writing, and applicability to
events where there is resistance to interviews and questionnaires.

Its limitations include: inapplicability to events that are private, restriction to current
events only (not past or prospective occurrences), �nd inability to capture inner feel­
mgs.

There are two types of observation. One is structured observation where explicit vari­
ables are identified. The second is unstructured since the variables are not yet defined
when observation takes place. The first type of observation can be applied in quantita­
tive studies as well as qualitative studies. The second is mainly for qualitative studies.
The researcher is confronted with the issue of whether to conceal or reveal the purpose
of the study and whether to be a participant or a detached observer. These choices
depend on the situation at hand.

Finally, a diary is an important instrument in recording information, especially in


unstructured types of observation, to make sure that the researcher will not forget
important insights captured in a given setting.

REFERENCES

Bautista, Victoria
1989 How effective is the PHC Strategy: Highlights of the Results of a Survey. Philippine
Journal of Public Administration XXXIII 2 (April).
Creswell, John W
· 1994 Research Design: Qualitative and Quantitative Approaches. Thousand Oaks: Sage Publi­
cations.
Gay, L.R.
1996 Educational Research: Competencies for Analysis and Application. Columbus, Ohio: Prentice
Hall.
Judd, Charles, Eliot R. Smith and Louise Kidder
1991 Research Methods in Social Relations. Forth Worth: Holt, Rinehart and Winston.
O'Sullivan, Elizabethann and Gary R. Rassel
1994 Research Methods for Public Administration. 2 11d Edition. White Plains, New York:
Longman Publishers.

UP OPEN
UNIVERSITY
MODULE14

INTERVIEWS, QUESTIONNAIRES,
AND INDIRECT METHODS

INTRODUCTION

The previous module focused on direct observation as a method for gathering


information. The present module introduces other primary techniques of data col­
lection. They are grouped together here because they are characterized by respond­
ents relating information about their experiences to the researcher. The experi­
ence is seen through the eyes of the respondent, not the researcher.

OBJECTIVES

At the end of this module, you will be able to:

Differentiate between interviews and questionnaires;


Point out the strengths and weaknesses of telephone interviews;
List the rules in the formulation of interview schedules and questionnaires;
Explain how to implement pretests prior to fielding questionnaires and
interview schedules;
Distinguish indirect methods from interviews 94
and questionnaires; and
Identify the different indirect methods that can be applied in public ad-
ministration research.
198 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

1.0 INTERVIEWS AND QUESTIONNAIRES DEFINED

Interviews and questionnaires are very common techniques of data collection. They
both entail drawing information from respondents.

The difference lies in the fact that interviews encompass a set of questions raised through
person-to-person or telephone interaction between interviewers and respondents. The
interviewers record the responses derived from the respondents.

A questionnaire, on the other hand, is an instrument that embodies a set of questions


asked in a standardized manner to each respondent:That is, the questions are raised in
the same words and in the same order to each respondent. Also, the questionnaire
method differs from interviews because it is the respondent who fills out the instru­
ment.

There are two types of interviews.

One is the standardized interview. This is characterized by a set of questions formu­


lated in a standardized way, as in questionnaires. It utilizes an instrument called an
interview schedule. This type of instrument is used in well-structured types of re­
search problems where the variables are well delineated. This is applicable to quanti­
tative types of research problem.

The second type is the unstructured interview. This involves asking a set of questions
derived by the interviewer from a list of key topics as a guide. This allows the inter­
viewer the freedom to pursue unanticipated topics or directions as they arise. This is
particularly helpful in qualitative studies, especially those of an exploratory nature.
The instrument formulated here is an interview guide.

A variant of the unstructured interview is the focus group discussion or FGD. This is
often undertaken by an interviewer with a small number of people who share a com­
mon interest (Gay 1996: 224). They meet with the interviewer for some hours and
respond to open-ended questions. Participants are able to listen to what each one has
to say, which gives each an opportunity to reflect on the opinions of the others. The
Mindanao Training and Resource Center (1995) suggests that the total number of par­
ticipants in an FGD should not exceed 15 and that sessions should not last more than
three hours. This method was applied in an assessment of one of the programs on
health (Bautista 1996). A brief note on this assessment is given below, since this is an
emerging methodology.

Purpose of the study: To make an assessment of the effects and impact of


Partnership for Community Health Development (PCHD) and how these are
influenced by administrative inputs, leadership commitment to a participa­
tory approach, and the extent and level of citizen participation.
UP OPEN
UNIVERSITY
DATA COLLECTION/ Interviews, Questionnaires, and Indirect Methods 199

Techniques of data collection: FGD was conducted in each to cover each of


the 64 barangays in seven provinces, with the assistance of a research team
with at least three members. One of the three in the team served as the facilitator
while two acted as documentors.

Key informants were organized into two focus groups. The first included the
barangay leaders, community volunteer health workers, and residents. They
were asked about the health condition of the barangay, the facilities and serv­
ice providers available, the perceived improvements with the introduction of
PCHD, and the difficulties encountered with PCHD.

The second focus group was composed of active PCHD participants or those
knowledgeable about PCHD implementation, such as the community organ­
izer, the proponent NGO, and the Rural Health Unit staff. They discussed
such factors as: background information on the barangay, the chronology of
events that paved the way for the introduction of PCHD, zeroing in on the
nature and extent of community participation, and the general assessment of
PCHD.

In general, both interviews and questionnaires have an edge over direct-observation


since past events can be reconstructed by selecting persons who can recollect what
already transpired. This cannot be done in direct observation.

Furthermore, future plans can be ascertained using interviews and questionnaires.

Both approaches can also explore people's feelings about certain events or persons. It
is difficult to ascertain these in direct observation. Subjects will either say what they
feel or express it through body language. Intepreting the latter can be tricky. For
instance, does a frown mean the respondent agrees with what another respondent is
saying, or that she disagrees?

2.0 COMPARING INTERVIEWS AND QUESTIONNAIRES

2.1 INTERVIEWS

Let me first talk about interviews. One of the advantages of interviews over question­
naires is the relative ease by which the cooperation of respondents can be obtained.
This is because the person-to-person interaction can easily convince respondents of
the importance of the research. Follow-up statements can be made to explain the pur­
pose of the study, which may encourage the respondents to cooperate. For Filipinos
who are keen on the personalized approach in interaction, interviews provide a power­
ful method for obtaining the cooperation of respondents. In contrast, the fallout rate
in questionnaires is high because of the inability to convince the respondents of the UP OPEN
value of the research. (This explains how questionnaires end up in the garbage can.) UNIVERSITY
200 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Second, interviews can be applied to those who cannot read and/or write. In assess­
ing programs and projects catering to the marginalized sector of the population whose
beneficiaries have not had formal, or even informal schooling, interviews are more
applicable than questionnaires.

A third strength is the opportunity to clarify questions in an interview. This is not


possible in questionnaires. When respondents have questions about certain items in
the instrument, they are not likely to call up or trace you.

Fourth, interviews enable the researcher to understand the context under which a
given set of questions is being answered. For instance, does the husband raise com­
ments to certain questions which the wife will then consider instead of conveying her
own opinion? In questionnaires, this cannot be ascertained. Also, in interviews you
may be assured that there is no interference as your queries are being answered. You
can see whether the person in the right mood, and come some other time when he/she
is in the right disposition. In short, you are able to determine influences of the envi­
ronment on the mood of the respondent.

A fifth advantage to using interviews is the opportunity to ascertain the identity of


the respondent. This is not possible in questionnaires. Among key officials of govern­
ment and nongovernment organizations who may be targeted to serve as respondents,
there is a tendency to delegate the responsibility of answering questionnaires to their
staff. If the personal opinion of the key official is the one sought after in the question­
naire, the validity of the research is compromised because the response is being made
by another person.

Sixth, an interview is particularly powerful in raising open-ended questions. These


are questions that do not call for fixed answers. Examples of such questions are:

UP OPEN What problems have you encountered in implementing this program in your area?
UNIVERSITY What do you think are the possible solutions to each problem?
DATA COLLECTION/ Interviews, Questionnaires, and Indirect Methods 201

These kinds of questions need probing. You have to continue asking until the re­
spondent is able to exhaust all the possible answers to your question. This is difficult
to do in questionnaires. It is normal for respondents to follow the path of least
resisistance. That is, they usually give the first answer that comes to their minds and
rarely elaborate on it.

In questionnaires, fixed alternative types of question are more applicable. This is char­
acterized by raising categories after a question is raised which the respondent can
easily mark. For example, to the question, What problems do you encounter in the imple­
mentation of this program in this area?, the respondent will simply tick some of the fol­
lowing items that respond to this query:

--- Lack of resources


--- Lack of commitment among frontline workers
___ Lack of interest among target beneficiaries
___ Others (Specify)

The disadvantage of fixed alternative questions is that they do not give you a sense of
the processes or dynamics involved. In contrast, in an interview a respondant can give
the story behind a response such as "lack of resources." The respondent may say:

There is lack of resources since those from the central office are so slow in releasing the funds
to us here in the region. This is because the criteria for releasing these funds have not yet
been fully set up-especially with respect to who should be given priority among the target
beneficiaries of this program.

2.2 QUESTIONNAIRES

Let me now talk about questionnaires. The advantage of questionnaires over inter­
views is that they are less expensive. This is because interviewers need not be hired
for every instrument that is fielded. The instruments can be delivered all at the same
time either by mail or by messenger to the offices of the target respondents. They are
less costly in time and resources compared to the conduct of interviews. In fact, you
even have to allocate some funding for training interviewers before they are deployed
to areas where the samples have been drawn.

UP OPEN
UNIVERSITY
202 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Also, questionnaires could have a wider coverage in terms of respondents because it is


not as expensive to administer as interviews.

For some embarrassing questions that may be difficult to raise in interviews, question­
naires could be the answer. In this approach, the anonymity of the respondent is
maintained. For instance, with a questionnaire it may be easier to pose (and get an­
swers to) questions on premarital sex among young adults to determine the extent of
family planning program coverage.

Another advantage of using questionnaires is the opportunity given to the respondents to


answer the questionnaire at their convenience. Th.ere is less pressure on their part to
answer the queries that have to be answered immediately when there are interviewers.

3.0 TELEPHONE INTERVIEWS

Let me now make a few comments about telephone interviews as a special approach to
gathering information.

Compared to person-to-person interviews, a telephone interview is less expensive.


T here is no need to hire many interviewers, which is required in personal interviews if
you wish to complete the study within a short time.

Telephone interviews also have a wider coverage of respondents because the inter­
viewer can have more interviews at one point in time compared to an interviewer who
has to personally seek out the respondents.

With telephone interviews, too, the researcher does not encounter difficulties posed
by the setting where personal interviews have to be conducted. I recall one of my
students in research methods who related his encounter with street corner men (istambay)
who were unemployed, as his group was assessing the Zonal Improvement Program of the
National Housing Authority. Before he could seek out his interviewees he had to sit down
to some bottles of beer when he was invited by the group. Also, in households that do
not keep their dogs in cages, you run the risk of being bitten.

With telephone interviews, it is also easier on the part of the supervisor to oversee
how the interviews are being conducted since the interviewers are all located in one
place.

However, there are limitations to the use of telephone interviews. First, it is applica­
ble only to a given class of people-that is, those who can afford telephones. It is also
limited by area, since not all of those who can afford it have access to telephone serv­
ices. Hence, telephone interviews can be limited in terms of the sector we wish to
generalize to.

Second, it is very difficult to get the cooperation of the respondents as they may
UP OPEN doubt your intent. In some areas in the Philippines where pranksters are common­
UNIVERSITY place, respondents may not take your call seriously.
DATA COLLECTION / Interviews, Questionnaires, and Indirect Methods 203

Also, as with questionnaires, you will not be able to establish the condition under
which a response is being made. Is the person in the right mood for providing an­
swers to your questions? This could be a perennial issue.

More importantly, is this the person who should be able to answer your questions? If
you wish to talk to the mother in the family but it is a housemaid who answers, then
this invalidates the responses to your queries. It is difficult to ascertain the identity
of your respondents in telephone interviews. You should be aware of this, unless your
calls are directed to anyone who answers the phone, as when you are finding out whether
people are watching government-run programs on television and if so, which pro­
grams.

4.0 RULES IN THE FORMULATION OF QUESTIONS

The formulation of the interview schedule and the questionnaire is a major responsi­
bility of the researcher. These are the basic instruments that he/she utilizes in gather­
ing information on the variables of the study. To ensure the validity and reliability of
the instrument, the researcher c�n be guided by some important rules for framing it.
These are:

1. The purpose must be clearly stated in the instrument formulated.

The objective in conducting the study should be clearly stated in the letter to the
respondent in questionnaires. This is orally expressed by the interviewer in inter­
views.

There are times when the real purpose is not openly stated but "softened" through the
use of more general statements. For example a researcher need not say outright that
graft and corruption practices is the objective of his/her study. It could be stated that
the purpose is to witness the kinds of relationships established between bureaucrat
and clients. However, the researcher has a moral obligation to protect the security of
the respondents under study.

2. Directions on how to answer the questions must be clear.

As a rule, instructions on how to fill out questionnaires must be given to the respond­
ents. Are they going to check the boxes? Will they put an X on their answers? Will
they check only one category or more than one of them? Will they rank the categories?
These should be made clear to them.

In interview schedules, the instructions for the interviewer should also be clear. How­
ever, questions raised in interviews are different from questionnaires. Questions in
interview schedules can be raised in a conversational way. You can practice questions
orally before they are written down.

UP OPEN
UNIVERSITY
204 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

3. The questions must be clear and unequivocal, not ambiguous.

As far as possible, the questions must be specific in time, place and/or frame of refer­
ence in terms of the key variables to be measured. We have mentioned that a simple
variable like income could be misleading if the researcher does not clarify the time
frame and the referent for the term. Is it income per month or per year? What income
is referred to? Is it the family income or the income of the head of the family? If it is
family income, who constitutes the family? Does this include other kin staying in the
household? Will you adopt the National Statistics Office's definition of the family,
which is: all the members in the same household sharing the food cooked in the same kitchen.
If so, you pose a question about family income thus:

W hat is the monthly income of your family? (The family is refe1Ted to as all persons living
in this house who share food cooked in the same kitchen.)

4. A question should cover only one topic at a time.

As a rule, the researcher should avoid raising double-barreled questions. Covering two
issues at a time in one question can be confusing to the respondent. An example of a
double-barreled question is:

In the following scale, how would you rate the performance of the Barangay Health Work­
ers and the government health workers?

This item is difficult to answer because two sets of workers are referred to in the state­
ment. The question should be asked for each set of workers.

5. Respondents must be in a position to answer the questions.

Make sure that you are addressing questions to individuals who will be able to answer
them. For instance, if you are asking about health practices for children in the family,
it is important to know who in the household is the key decision maker on health. It
is difficult to assume that it is mainly the mother. Sometimes, it could be a grand­
mother who takes care of the children and is knowledgeable about immunizations and
other health care given to the children.

6. Avoid influencing respondents to answer in a particular way.

In other words, don't raise leading questions. The questions raised should not make
respondents think that you have preferences in the way they should answer. For in­
stance, asking the question, You are in favor of this program, aren't you?, indicates your
expectation that they give an affirmation of the program. This question can be posed
in a more neutral manner.

7. Avoid posing embarrassing questions.

UP OPEN Be sure that the questions will not make respondents feel uncomfortable because you
UNIVERSITY have indicated some standards of classifying them. Educational attainment is a vari-
206 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

9. T he items in the instrument must cover all the variables incorporated in the
research proposal.

It may help to formulate a matrix to show what items in the instrument respond to
these variables such as:

Variables Item number in the instrument

1
2
3

10. Do not forget to translate the instrument into the dialect the respondents are
familiar with.

The best way to establish rapport and obtain cooperation from respondents is to use
language that they are familiar with. This means that the instruments should be suit­
able to the level of understanding of the target respondents. They have to be translated
and must use words that the target respondents would understand. Let us remember
that our motive is not to impress them by using sophisticated language. The language
that we use should make them feel at ease and should be readily understood.

5.0 PRETESTING THE INSTRUMENT

Before they are disseminated questionnaires, should be pretested with a group with
the same characteristics as the sample. A pretest is a dry-run of the instrument to be
used. Pretesting can indicate whether the questions are understood by the respond­
ents. It can also help determine whether the questions are posed in such a way that
respondents are not embarrassed or turned off.

To whom should pretests be administered? A possible option could be the persons


from the same locale where respondents will be chosen. This is a convenient way of
doing it since you are proceeding to the same place where it is going to be imple­
mented and therefore can become familiar with the place. However, the samples should
be drawn from individuals who are not likely to serve as substitutes in case the sam­
ples to be drawn are not available.

Another option is to select from a group thar is outside of the sampling frame but
with the same characteristics. Thus, if the target is a cluster barangay pre-identified
another barangay included in the program coverage can be the source of pretest cases.

Symptoms of faulty structuring of questions in the instrument are:

Don't know responses. These could be indicative of questions th�t are vague,
embarrassing, or inapplicable to the respondent. If the first two ;are the rea­
UP OPEN sons, then the questions need to be rephrased. If the answer to the question is
UNIVERSITY "not applicable," it is a valid response.
DATA COLLECTION/ Interviews, Questionnaires, and Indirect Methods 207

Refusal to answer. This could be attributed to the inadequacy of the explana­


tion for conducting the study, lack of trust in the researcher, and the respond­
ent being indisposed. The last one is a valid response and a subsequent sched­
ule can be set. The first two require further explanation of the importance of
the research and the identity of the researcher.

Irrelevant answers. These may indicate that the question is vague or embar­
rassing. Modifications can be made to make it more specific if it is vague. If it
is embarrassing, it has to be rephrased.

6.0 INDIRECT METHODS

The previous sections dealt with interviews and questionnaires. They are normally
constructed with a set of questions that are usually straightforward in intent. Each
question posed is as specific and pointed as possible. Each is expected to be direct and to
the point. There is a saying for this: What you see is what you get.

However, there are times when respondents are not willing to reveal their feelings or
experiences. They cannot be drawn out by asking a direct question. I recall how diffi­
cult it was to ask direct questions about whether or not voters participated in the 1976
referendum during the martial law regime. A team who planned the research realized
that it would not be able to obtain accurate responses because respondents were afraid
to be candid.

For situations like this, an indirect method could be an option. This aims at deter­
mining experiences that respondents are not willing to reveal especially if they are
fearful of the implications of their response on their security or reputation. Psycholo­
gists use this methodology in order to determine the unconscious experiences of the
subjects of the study-such as their fears and aspirations, which they may not recognize.

There are two modes of the indirect method. One is projective and the other is struc­
tured indirect method.

The projective technique is characterized by the presentation of a series of ambigu­


ous stimuli. The assumption is that the subjects being observed put structure into the
ambiguity that is presented and unknowingly reveal themselves. Psychological tests
like the Rorschach Ink Blot Test and the Thematic Apperception Test are examples of
this mode. In the first case, the ambiguous stimuli are ink blots. In the second test, the
stimuli are a series of pictures. Respondents are asked to say what they see in the ink
blots. In the second test, they are asked to tell stories. Training is required in order to
interpret the meaning of the responses in these tests, which are psychological in na­
ture. You may consult with or hire experts for training in utilizing and interpreting
tests of this nature in case you need assistance in dealing with this sort of research
problem.

More applicable tests in public administration that are relatively easier to implement UP OPEN
include: UNIVERSITY
208 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Word Association Test. In this, the respondents are presented with a word
and asked to respond with the first word that comes to mind. For example,
office personnel may be asked about "boss." Asking directly how they feel about
the boss may be difficult, and may require this indirect approach. The first word
that comes to their mind could indicate how they see their boss.

The answers could be: "tyrant," "caring," "ruthless," "uncaring," "pro-peo­


ple," "understanding," etc. The dominant qualities can be determined and re­
lated to relevant concepts or theories in public administration, where these
terms could be analyzed or classified. It ,is possible that a new terminology is
generated from the insight of the researcher.

Sentence Completion Test. This is the same in structure as the first type
except that the stimuli are phrases and the response of the respondent is to
complete each into a sentence. For example, instead of just presenting a word,
like the example in the first case, the phrase could be:

My boss is ------------
My office is

Projective Question. This is actually a violation of the rules for the formula­
tion of questions in standardized instruments for interviews and questionnaires.
Earlier, I said that framing the questions in structured problems should be as
specific as possible. In formulating projective questions, the intention is to
make the instrument ambiguous so that the respondent can express his/her
own perception or feeling about the situation. In line with the earlier example
I raised, a projective question can be:

How do you feel about your boss?


What do you think about your office?

In addition to projective methods, a method often used in standardized instru­


ments (i.e., questionnaires) is the structured indirect method. In this case,
fixed alternative answers follow some questions that are posed by the researcher.
However, there is no objective basis for determining the veracity of these
alternatives.

I recall one question we raised in the team that studied people's reactions re­
garding the 1976 referendum conducted by the Marcos administration. In­
stead of asking directly what they personally feel about the referendum, the
itern raised was:

UP OPEN
UNIVERSITY
DATA COLLECTION / Interviews, Questionnaires, and Indirect Methods 209

What do you think the majority of the community members feel about the
referendum as a mechanism for political participation?

--- Strongly disagree


--- Disagree
--- Not sure/ambivalent
___ Agree
___ Strongly agree

SAO 14-1

Fill in the blanks.

1. The instrument in structured interviews is called ________


2. The instrument in unstructured interviews is _________
3. The instrument used in documenting observations in unstructured research
problems is called _________
4. When standardized questions are formulated and the instrument is mailed to
the respondents, the method applied is ________
5. Conducting an interview involving a group of persons is called

6. This type of question is more powerful in raising probing questions:

7. Name one difficulty in conducting telephone interviews:

8. Before fielding your instruments, it is important to _________


9. Raising a projective question to policemen who are often accused of collect­
i'ng protection money from public transportation drivers by asking about
how they th;nk the;r peers are do;ng on the job is an example of an
________ method of data collection.
10. When I ask a question like: "How do you rate your resource speakers and
facilitators of this training program?" I am violating the principle in
formulating questions in an instrument?

UP OPEN
UNIVERSITY
210 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

ASAQ 14-1

1. Interview schedule
2. Interview guide
3. Diary
4. Questionnaire
5. Focus group discussion
6. Open-ended questions
7. You are dealing with a special group of people.
8. Pretest
9. Indirect
10. Avoid raising double-barreled questions

9-10 pts. - Very Good


7- 8 pts. - Good
5- 6 pts. - Fair
< 5 pts. - Review

SUMMARY

Interviews rely on information obtained by the researchers through person-to-person


interaction or telephone calls. The researchers in this case fill out the instrument,
called an interview schedule. Questionnaires, on the other hand, are filled out by the
respondents themselves. Like interview schedules, questionnaires are raised in a stand­
ardized way.

Interviews can also be used in unstructured or qualitative studies. In this case, the
instrument-the interview guide-is more flexible and less standardized than ques­
tionnaires and interview schedules.

Both interviews and questionnaires have the advantage over direct observation since
you can reconstruct past experiences and anticipate the future. Inner feelings may also
be ascertained.

Interviews have an edge over questionnaires for: the ease of cooperation a researcher
can obtain from respondents; its applicability to those who cannot read and/or write;
the opportunity for respondents to ask questions; the opportunity for the researcher to
understand the context of the investigation; the opportunity for the researcher to es­
tablish the identity of the respondent; and the occasion it offers for raising open-ended
UP OPEN questions.
UNIVERSITY
DATA COLLECTION/ Interviews, Questionnaires, and Indirect Methods 211

On the other hand, questionnaires have the advantage of being less expensive, allow­
ing a wider coverage at a single point in time, assuring the anonymity of the respond­
ents, and giving respondents an opportunity to respond at their convenience.

Telephone interviews as a special kind of interview are less expensive than person-to­
person interviews; they can have a wider coverage at a single point in time; they will
not endanger interviewers due to difficulties posed by travelling; and they can facili­
tate supervision of the interviewers. However, they can only reach a particular sector
of people; they make it more difficult to obtain cooperation; and they provide no
means to establish the environmental context of the observation or to ascertain the
identity of the respondents.

T here are rules to assure the validity and reliability of the instruments that are to be
formulated. One important rule is to pretest the instrument to make sure that it is
clearly understood by the respondents.

For questions that are difficult to raise because they are embarrassing and will not get
straightforward responses from respondents, projective techniques and structured in­
direct methods can be raised.

REFERENCES
Bautista, Victoria
1996 Partnership for Community Health Development Appraisal Report-Year III. Quezon
City: UP College of Public Administration.
Gay, L. R.
1996 Educational Research: Competencies for Analysis and Application. Columbus, Ohio:
Prentice Hall.
Judd, Charles, Eliot R. Smith and Louise Kidder
1991 Research Methods in Social Relations. Forth Worth: Holt, Rinehart and Winston.
Mindanao Training and Resource Center
1995 Training Package on Community Organizing--Participatory Action Research. Davao
City: MTRC.
O'Sullivan, Elizabethann & Gary R. Rassel
1994 Research Methods for Public Administration. 2nct Edition. While Plains, New York:
Longman Publishers.

UP OPEN
UNIVERSITY
MODULE15

SCALING TECHNIQUES

INTRODUCTION

One way to enhance the measurement of a variable is to determine the magnitude


of its manifestation. Do you remember that we talked about the different scales of
measurement in order to operationally define a variable? If you will recall, one of
the scales for measurement is an interval scale. In Module 3, I said that the scores
are arbitrarily formulated. It is the researcher who defines the range of scores a
given variable can be assigned.

In this module, I will discuss some scaling techniques and some of the basic rules
in formulating them.

This topic is related to that of the pre,.rious module. Scales are usually incorpo­
rated in standardized instruments (i.e., interview schedules, questionnaires, and
checklists for structured observation).

OBJECTIVES

At the end of this module, you will be able to:

Define scale and scaling;


List the rules in formulating scales;
Apply the techniques for formulating a rating scale; and
Apply the techniques for formulating a multiple-item scale.
214 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

1.0 SCALES AND SCALING

It is important for a researcher to make a decision regarding the level of measurement


he/she will adopt because this will influence the techniques for data analysis-in par-
. ticular, what statistical tools, if any, will be used to summarize the data. For example,
with empowerment having become a commonplace objective of programs and projects
of government and nongovernment organizations, "level of participation" has become
the focus of study of academicians, researchers and practitioners in the Philippine
setting. This may be measured using various scales, from the nominal to the fixed ratio
scale.

A researcher formulates a scale for every variable of a study. A scale is "...a set of cat­
egories or a range of scores on a variable" CJudd et al. 1991: 146). "The process of
assigning scores to objects to yield a measure" CJ udd et al. 1991: 146) of a variable is
called scaling.

Among the four levels of measurement, the interval scale requires the most creativity
on the part of the researcher in defining the range by which a given variable is scored.
Nominal scales only mRke qualitative distinctions and therefore do not require scor­
ing. Ordinal scales have elements that are ranked. It is interval scaling that requires
the researcher to assign values along various points in the scale. This means that the
researcher must conceptualize how these values should be assigned.

Fixed-ratio scales are based on standardized values. The researcher does not assign
arbitrary scores. For instance, the number of community activities that a family is
engaged in at the purok level in a barangay is measured by a fixed ratio scale. The
researcher does not formulate the scores. The highest possible score a resident would
have is the number of activities actually implemented in the locality. The researcher
does not assign these values. However, to the question, "How do you personally assess
your level of participation in your cornmunity on the following scale?", scores must be
assigned, thus:

Very lowlI-- -----------+--------+------1 Very high

1 2 3 4 5

It is the researcher who does this. There are rules that can guide a researcher in formu­
lating this scoring system to make sure that the scale is valid and reliable.

It should also be borne in mind that scaling is based on a given theoretical foundation.
There is a conceptual basis for the construction of a scale. The scale aims to operation­
ally define a variable that fits the theoretical argument of the study.

UP OPEN
UNIVERSITY
DATA COLLECTION / Scaling Techniques 21 5

2.0 RULES IN THE FORMULATION OF SCALES

A researcher must follow the rules in the formulation of scales. The first rule is to
make sure that the items formulated are related to the variable being studied. If the
concern is assessing the level of performance of the volunteer health worker in terms
of mobilizing the community, the item should refer to mobilization and not to her role
in giving curative care. Also, if the effectiveness of the resource person in a capability
building program is being evaluated, the items should be with reference to the per­
son's instructional capability and not how the speaker dresses.

Another rule is the principle of unidimensionality. As with interview schedules and


questionnaires, scales should avoid raising double-barreled questions. Each item in a
scale should focus on one attribute at a time.

Furthermore, the items that are constructed should be able to differentiate people
along different points in the continuum, from one extreme to the other. A midpoint
position should also be taken into consideration. For example, some people do not
have an attitude regarding certain issues. Thus, they should be able to choose "neu­
tral" in the scale provided. For example:

Question: What do you think about the issue of computerizing electoral results?

Strongly Neutral Strongly

I I I I
Disagree Disagree (No Opinion) Agree Agree

2 3 4 5

The sample scale shows a middle position for those with no opinion or those who have
a neutral stand regarding the issue at halld. To present only the two poles will not elicit
all the possible opinions respondents could have on a given issue.

In scoring items in a scale, remember to assign equal units from one point of the
continuum to another. The division of the points must be equidistant throughout. In
the scale item presented previously, if there is a difference of one unit from the first to
the second point in the scale, there should be also one unit between the second and
third category. It cannot be 1 for the first and then .5 for the second, and so forth, as in
the example below:

*
Strongly Neutral Strongly
Disagree Disagree (No Opinion) Agree Agree

1 2 2.5 4 5
UP OPEN
UNIVERSITY
216 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

3.0 RATING SCALE

I will discuss two types of scales in this module. One is a rating scale. The second is a
multiple-item scale. Let me discuss rating scale in this section.

A rating scale consists of an item indicating the characteristic or trait of a person,


object, or event being evaluated. Assessment is usually made by a rater who places any
element being assessed at some point along a continuum or in one of an ordered set of
categories to which scores are assigned. The rater may be another person reviewing
the performance of the element. Or the respondent could rate him/herself. This scale
is often used to determine a person's feelings, beliefs, and attitudes.

Researchers should select the elements well. The validity of the research could be
affected by poor selection of raters-that is, those who are not in a position to evaluate
the elements in the research in focusing on. For instance, it may not be appropriate to
obtain the opinion oflocal executives regarding nongovernment organizations (NGOs)
interfacing with them in local development councils, especially if they are newly in­
stalled executives who have not personally encountered NGO representatives.

T here are many options in constructing rating scales.

3.1 GRAPHIC RATING SCALE

In this option, the rater indicates his/her assessment of the element being evaluated by
placing a marking (a check, an X, whatever) at the appropriate point along a con­
tinuum. Descriptions are provided for the two end-points of the scale, but not for
intermediate positions. For example, we may ask NGO representatives in the local
development council to rate their satisfaction with the local chiefexecutive's openness
in accepting the ideas of NGOs. They may be asked the question:

How would you rate your mayor in terms of his openness in


accepting your ideas in local development council meeting�?
(Encircle the point that depicts your assessment of the mayor.)

Very Very
Satisfactory Unsatisfactory

A score can be given for every point in this continuum. To obtain the mean, the scores
of every respondent can be summed up and divided by the total number of respond­
ents A summary profile is therefore obtained about the mayor as assessed by a group of
NGOs. If the Very Satisfactory assessment is given 7 points, a mean of 6 is high com­
UP OPEN pared with another mayor who gets a mean of 2.5.
UNIVERSITY
DATA COLLECTION / Scaling Techniques 21 7

3.2 ITEMIZED RATING SCALE

This is made up of a set of categories that are organized according to scale positions on
a continuum. A corresponding score is given for each category. The specific qualities
of each point in the continuum are spelled out.

For example, NGOs are asked to rate their mayor using an itemized rating scale:

How would you generally rate your mayor in terms of his openness
in accepting your ideas in local development council meetings?
(Check the statement that corresponds to your answer.)

Score
____ Very satisfactory (5)
---- Satisfactory (4)
---- Neutral/ambivalent (3)
____ Unsatisfactory (2)
____ Very unsatisfactory (1)

In this example, scores may be indicated in the researchers' code book but they do not
appear in the interview schedule or questionnaire where this scale could be incorpo­
rated.

3.3 COMPARATIVE RATING SCALE

This requires the rater to compare an element with a group with known characteristics.
One way of doing this is to rank-order the element with respect to a few others. Ranking
is used "only when the investigator is concerned with a limited group" (Judd 1991:
152). If there is a bigger group against which it is being compared, its position in
relation to the top performing or poor performing elements in the group could be
determined.

To follow through the same example provided above, a comparative rating scale using
the rank-order mode is as follows:

Of the past two mayors we had in this municipality, how would you rank Mayor
Sanchez in relation to Mayor Samonte and Mayor de la Cruz in terms of their
openness in accepting ideas of NGOs in local development council meetings?

Indicate their rank order in the following list with 1 as the most open:

Rank Name of Mayor


1
2
3 UP OPEN
UNIVERSITY
218 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Suppose the NGOs who have been asked to assess the performance of this mayor have
a track record of working with other mayors in all municipalities in the province,
where the municipalities number 35 in all. Then it is possible to construct the item
such that the reference point is not individuals but the group of mayors. Rank-order­
ing will be difficult to do in this case. Instead, reference to the entire group may be
useful. For example:

How would you rate Mayor Sanchez in relation to the other mayors in the
province in terms of his openness to the ideas of NGO representatives in local
development council meetings? Circle the nUJI1ber that responds to your an­
swer:
1 2 3
Belongs to the top Belongs to the top Below the top
10 % 20% 20%

On the whole, we have seen different ways of constructing rating scales. However, they
have one major deficiency. This is their reliance on a single item to assess a particular
attribute. As a rule, validity and reliability increase if there are more items to a test.
Multiple items can easily show whether there is consistency in the behavior of the
element being observed. Furthermore, having several items could provide more char­
acterization of the attribute being assessed. For this reason, a multiple-item scale may
prove to be a better alternative. This is discussed in the next section.

4.0 MULTIPLE-ITEM SCALES

A multiple-item scale consists of a series of items in order to depict a particular at­


tribute. The rater (or judge) expresses agreement or disagreement with each of these
items to which specific scores are assigned. A composite score for all items can be
obtained, showing the overall profile of the element being evaluated.

Three techniques will be discussed here as examples of this type of scale. These are
differential scale, summated scale, and semantic differential scale.

4.1 DIFFERENTIAL SCALE

This type of scale consists of both favorable and unfavorable items concerning a per­
son, event, concept, or object to which a rater is asked to react. The rater marks only
the items with which h/she agrees. A score is obtained by computing the mean of the
scale value of the items to which the rater responds. The score indicates the rater's
position of favorable or unfavorable attitude, with a continuum of 1 to 11 points.

How is a differential scale constructed? How is the scale value obtained? The follow­
ing are the basic steps in formulating a differential scale:
UP OPEN
UNIVERSITY
DATA COLLECTION/ Scaling Techniques 219

1. Items that depict an attitude or an attribute relevant to a particular issue are for­
mulated. The items should be formulated so that they are distributed over various
points on the dimension underlying the attitude or attribute.

For example, I may be formulating scale items to study the participativeness of


local executives in relation to NGOs. The first step would be to develop a set of
items or statements with some referring to very low levels of participativeness,
some to very high levels, and some to various points in between.

The following are some sample statements:

Trial Items

1. Does not involve NGOs in making decisions, but invites them to attend
local development council meetings.
2. Does not consider their presence essential in local development council
meetings
3. Invites them to participate in socials of the local development council.

l
20. Gives NGOs an opportunity to share their views and draws consensus from
them together with the other members, on the policy decision that will be
made to resolve an issue.

2. Next, a panel of judges is asked to rate the statements on a scale from 1 to 11,
where 6 serves as the neutral position.

For instance, experts on participatory management are asked to group the previ­
ous items on their scale position from 1 to 11 points and the following results are
obtained from five experts:

Trial Items Judges' Assessment of the Scale Position of the Items

A B C D E
No. I 2 1 2 2 2
No. 2 1 1 1 1 1
No. 3 8 2 5 8 1

No. 20 10 11 11 10 11

UP OPEN
UNIVERSITY
220 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

3. The scale value of an item is assigned by obtaining the mean of the position to
which it is assessed by a group of judges. Only the statements where there is most
consistency in rating among the judges are considered in the scale.

Considering the previous examples, only three of the four items show consistency
in terms of the items that can be considered in the scale. These are:

Item No. 1 with a scale value of 1.8 after getting the mean from the
sum of 2, 1, 2, 2 and 2 divided by 5.
Item No. 2 with a scale value of 1
Item No. 3 (Excluded because of lack"of consistency)
Item No. 20 with a scale value of 10.6

4. Final selection is made by taking items that are spread out evenly along the scale
from one point of the continuum to the other.

For example, the following scale values are obtained from a list of items:

Item No. 1- 1.8


Item No. 2--.JL 1
Item No. 3- Excluded
Item No. 4- 3
Item No. 5- 8
Item No. 6- 8.2
Item No. 7- 2.1

Betw�en Item No. 5 and No. 6, the first can be selected for the final scale since it
shows number eight position. Between Item No. 1 and No. 7, I would have chosen
Item No. 7 since it varies from position number 2 only by a difference of 0.1. Item
number 1 differs by 0.2 points from the scale val1_1e of 2.

However, it takes creativity on the part of the researcher to design the different
items so that each could represent the different points in the continuum to depict
the variable being evaluated.

The selection of the items to be incorporated in the scale is based on the mean of
the scale value of the trial items ticked by the respondents. The mean rating of
each rater can be added, and the sum divided by the number of raters. An example
of how the mean rating of the mayor is determined follows:

( 9 + 8 + 10 + 10 + 10 + 12 ) -=- 6 = 9.8

The mean score of the mayor is close to the highest possible score of 11, which
UP OPEN shows a favorable performance on this assessment.
UNIVERSITY
DATA COLLECTION / Scaling Techniques 221

The differential scale is sometimes called a Thurstone Scale, after the person who
initiated it.

The table is an example of this scale (Judd et al. 1991:159).

Table 15-1. Example of a Thurstone Scale

Scale Value Item

1.2 I believe the church is a powerful agency for promoting both individual and
social righteousness.
2.2 I like to go to church for I get something worthwhile to think about and it
keeps my mind filled with right thoughts.
3.3 I enjoy my church because there is a spirit of friendliness there.
4.5 I believe in what the church teaches but with mental reservations.
6.7 I believe in sincerity and goodness without any church ceremonies.
7.5 I think too much money is being spent on the church for the benefit that is
being derived.
9.2 I think the church seeks to impose a lot of worn-out dogmas and medieval
superstitions.
10.4 The church represents shallowness, hypocrisy, and prejudice.
11.0 I think the church is a parasite on society.

Note: On the actual questionnaire, the items would appear in a random order, not
ordered by scale value as shown here.

4.2 SUMMATED SCALE

This is sometimes called the Likert Scale, after the person who conceptualized it. This
is a very popular scale because it is relatively easy to formulate, compared with the
Thurstone Scale.

It consists of several items showing either a.favorable or unfavorable response to the


person, event, concept, or object being assessed. The rater responds to all items by
indicating a certain degree of agreement or disagreement along a continuum, usually
ranging from 1 to 5 points. The total score is obtained bysumming up the rater's item
scores. Obtaining the mean for all items in the scale could also be one wayof summa­
rizing the assessment of the element evaluated.

How is the summated scale constructed?

1. Items are assembled according to the attitude investigated-that is, whether


favorable or unfavorable.

Suppose we wish to evaluate the effectiveness of the health worker assigned in a


given area. Some of the attributes that depict qualityof service are: equity, appro­
priateness of the service, adequacy, timeliness, rapport with clients, etc. On the UP OPEN
UNIVERSITY
222 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

basis of these attributes, items can be formulated to demonstrate the favorable


features of an effective frontline worker.

Some trial items may include the following. (Respondents are asked to check the
appropriate column with respect to a person being rated.)

Trial Items Strongly Agree Neutral Disagree Strongly


Agree (A) (N) (D) Disagree
(SA) (SD)
Delivers service
on time.
Treats clients
fairly.
Delivers services
that are responsive
to my needs.
Shows concern °
when I talk to
him/her.
Is fluent
in speaking
the dialect.

2. These items are administered to a group that is representative of those to whom


the scale will be administered.

In the preceding example, the trial items can be administered to other barangays
that are not covered by the same health workers to whom the instrument will be
administered.

3. Participants are made to indicate either agreement or disagreement to all the items.

4. The total score is obtained by adding item scores. The mean may also be used to
determine the profile of the one evaluated.

If there is a total of 15 trial items, the highest possible score is 75.

5. The items that are finally selected are those that show a difference between the
high scorers and the low scorers in the pretest.

In the preceding case, the pretest items may be administered to raters to purposively
assess selected high performing health workers against a poor performing one.
They may be identified with the assistance of key informants.
UP OPEN
UNIVERSITY The consistency of the ratings given by the pretest raters on the trial items for
DATA COLLECTION / Scaling Techniques 223

these two persons are then compared.

The items that are finally selected are those that show the noted top performer to
be consistently high in ratings and the poor performer, consistently low.

Irrelevant items yield variable ratings for the two persons evaluated.

In the same example, hypothetical data of the raters' assessment of these two per­
formers show the following:

Trial Items Scores of High Performer Scores of Poor Performer


(30 Raters) (30 Raters)

SA A N D SD SA A N D SD

Item 1 28 2 3 27
Item 2 25 5 1 4 24
Item 3 26 4 1 29
Item 4 27 2 1 3 27

Item 5 5 3 10 3 9 5 6 10 7 2

In this example, you may discard number 5 in the final scale since the two ratees
have not been consistently marked as high performer and poor performer.

You may reverse the presentation of the items to make sure that respondents are
not answering mechanically. If you are starting on a positive note such as: "Deliv­
ers service on time," you may reverse the quality for the next item, such as, "Does
not treat clients objectively." Make sure not to forget to reverse the scoring system.
If you score "strongly agree" as 1 for the. first item "strongly disagree" for the
second item is circled as 5 by respondent but this is scored as 1 by the researcher.

Suppose you have many ratees for a given assessment (let us say it is self-assess­
ment of many health workers on how they treat their clients), the following box

UP OPEN
UNIVERSITY
224 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

shows the process of computing the discriminative power of the items:

Box 15-1. Computing the discriminative power of a scale item

1. Arrange respondents (judges) on the basis of their scores for all items.
2. Identify the top and bottom 25% of the respondents. If there are 120 pretest raters, 25% of
this is 30.
3. For any given statement (or item), the discriminative power (DP) is the weighted mean
response of the top 25% minus the weighted mean response of the bottom 25% of the re­
spondents. Let us suppose number 22 is being ratea. The DP is computed as follows:

(a) (b) (c) (d) (e)


Group Number Itern Responses Weighted Total Weighted
in Group (Item 22 Score) (Response Value Mean
X Frequency) ( d /b )

SA A N D SD
1 2 3 4 5

Top 25% 30 0 2 10 9 9 0 + 4 + 30 + 36 + 45 = 115 115/30 = 3.8 3


Low 25 % 30 6 24 0 0 0 6 + 48 + 0 + 0 + 0 = 54 54/30 = 1.80

Discriminative Power (DP) = Weighted mean of top 25% -


Weighted mean of low 25 %
(3.83 - 1.80 = 2.03)

4. The DP Index must be computed for each of the possible scale items, and those
items with the largest DP indices will be selected. A DP of at least 1.00 should be
obtained by the items.

As the final scale is formulated, the researcher must devise a way to summarize the
information that will be collected from the raters. The following demonstrates
how the responses to the items in a summated scale can provide a composite score.

UP OPEN
UNIVERSITY
DATA COLLECTION/ Scaling Techniques 225

Suppose you have the following data on the raters for a scale. What is the rating given
by 30 raters for the person rated? Let us suppose there are 30 raters and you obtain a
score for each column which shows 1, excellent; 2, very good; 3, good; 4, fair; and S
poor, the frequency count for each column per item is the assessment of the raters.

Number of Raters Answering

1 2 3 4 s
Item a 25 s
Item b 10 10 10
Item c 15 10 s
Item d 15 15
Item e 30

Multiply the number of raters who answered every number in Item a by the weight of
the response per column. Column 1 has a weight of 1; column 2, has 2 and so on. Thus,

Item a: 25 X 1 = 25; Sx 2=10


Item b: 10 X 1 = 10; 10 X 2 = 20; 10 X 3 = 30
Item c: 15 X 1 = 15; 10 X 2 = 20; 5 X 3 = 15
Item d: 15 X 1 = 15; 15 X 2 = 30
Item e: 30 X 1 = 30

You may now sum up the weighted scores per item:

Item a: 25 + 10 = 35
Item b: 10 + 20 + 30 = 60
Item c: 15 + 20 + 15 = so
Item d: 15 + 30 = 45
Item e: 30 = 30

The total weighted score is 220 for all raters.

Divide the total weighted score by the number of raters to get the mean score per rater.
This is 220 + 30 = 7.3. If the highest possible score is S (or excellent for all items),
getting 7.3 is in between excellent and very good.

If you wish to get the mean score per item by each rater, divide the mean score per rater
by the total number of items. Hence, 7.3 + 5 = 1.46.

Considering the data in the foregoing example, the ratee shows a cl-0se to excellent
performance since the mean score per item is 1.46.
UP OPEN
UNIVERSITY
226 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

4.3 SEMANTIC DIFFERENTIAL SCALE

This is actually the easiest to formulate among the three. It is like a summated scale
since every respondent is expected to answer all items. However, it also differs from
the summated scale since it consists of several scales instead of one scale composed of
several items (like the first two examples). Each scale is a bi-polar adjective pair on a
seven-point continuum. The total score for all of the scales is derived although each
scale may be assessed independently.

Formulating this scale is relatively easy. It entails:

1. The selection of the concepts to be rated

2. The selection of the appropriate adjective pairs to depict the concept or concepts

Criteria for the selection include:

Relevance to the concept or its appropriateness to the concept. This would


mean understanding the theoretical basis of the concept to be able to prepare
adjective-pairs.
Factor representativeness which means sufficient sampling to characterize each
of the concepts ·studied.

For the concept of"effectiveness" and "concern for community participation" some
adjective pairs could be:

(Effectiveness)

Competent . .. .. . . .. Incompetent
-------
Fair . .
------- . . . . Unfair
Fast . .
-------. . . . Slow
Uncaring . . . . . . Caring
. .. .. .. .. .
-------
Incapable .
------- Capabie

(Openness to citizen participation-how they relate with the community) •

Open . . .
. .... . .. Dogmatic
-------
Motivating . . .
-------. . . Stifling
Poor listener .. .. . . . .. Good listener
Dictatorial .. .. .. .. . ..
-------
Liberal
-------

3. Scores can be analyzed for differences between concepts (effectiveness and


participativeness), between scales (technical preparation and rapport with clients),
and the overall components of the scale.

The mode for analyzing the scores in the final item is similar to a summated scale.
UP OPEN However, in differential scales, you may break the scoring by concepts or by scales
UNIVERSITY
DATA COLLECTION/ Scaling Techniques 227

per concept, depending on the purpose of your research.

The mean scores of scales comparing effectiveness and participativeness may show
a high score for effectiveness (6.7) but only 4.5 for ability to relate with the com­
munity.

In the effectiveness dimension, the weakest spot of the ratee could be "being car­
ing" and "fair" where ratings could be below 4. On other hand, the ratee scores
higher than 5 on the other items, indicating technical preparedness.

Now, try the SAQ to see how well you understand the discussion.

SAO 15-1

Answer the following questions:

1. Is there anything wrong with the following statement? Yes or no? If yes,
explain your answer.

How would you rate your mayor and your vice mayor in terms of their
responsiveness to the needs of the municipality they are managing? Check
the number which corresponds to your assessment.

Very satisfactory1--2--3--4--5Very Unsatisfactory

2. What type of scale is illustrated in number 1? A rating scale or a multi-item


scale?

3. What type of scale is shown in the following case? Differential, summated or


semantic differential?

Please rate the- resource person of this program on each of the behaviors
listed below. Circle one of the numbers from 1-5. 1 stands for Excellent; 2 is
Very Good; 3 is Satisfactory, 4 is Fair and 5 is Poor.

a. Explains the objectives/expectations of the program. 1 2 3 4 5


b. Presents the topic systematically 1 2 3 4 5
c. Speaks clearly. 1 2 3 4 5
d. Demonstrates thorough knowledge of the subject matter. 1 2 3 4 5
e. Invites questions from participants. 1 2 3 4 5

UP OPEN
UNIVERSITY
228 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

4. The following are some attributes of the resource person from the Human
Resource Development Office of a department to be rated by clients. What
kind of scale is illustrated here?

a. Idealistic 1-2-3-4-5-6-7 Practical


b. Lazy 1-2-3-4-5-6-7 Industrious
C. Teaches well 1-2-3-4-5-6-7 Teaches poorly
d. Good speaker 1-2-3-4-5-6-7 Lousy speaker

5. Given number 4, if 20 raters put marks for the youngest resource person of this
program per item, what can you say about the assessment of the participants?
Make sure that you compute your scores consistently. (Note: A high score is
for favorable attributes. You have to reverse computations for certain items.)

1 2 3 4 5 6 7
a. Item a - 15 5
b. Item b - 10 10
c. Item c - 15 5
d. Item d - 20

ASAQ 15-1

1. Yes. Avoid raising double-barreled questions.


2. Rating scale.
3. Summated or Likert scale
4. Semantic differential scale
5. Make sure to reverse the score for a and b.

Scoring is as follows:
a. 15 x 6= 90; 5 x 7= 35 Sum = 125
b. 10 x 6= 60; 10 x 7= 70 Sum = 130
c. 5 x 6= 30; 15 x 7= 105 Sum = 135
d. d. 20 x 7= 140 Sum = 140
Total= 530

Mean score per rater= 530 7 20= 26.5


Mean score per item= 26.5 7 4= 6.625

The performance of the resource person is close to perfect on the four items. Of
the highest possible score of 28 which could be given by a rater, a score of 26.5
was obtained. On a scale of 1 to 7, with 7 as the highest point on the continuum,
the resource speaker rated 6.625.
UP OPEN
UNIVERSITY
DATA COLLECTION / Scaling Techniques 229

SUMMARY

This module has primarily focused on the rules for formulating an interval scale-a
level of measurement characterized by arbitrary assignment of scores by the researcher.
Some guidelines in constructing this type of scale include: rel�vance of the items in
the scale to the variable of the study; unidimensional character of the items; differen­
tiation of people along the different points in a continuum; and equal points being
assigned from one point of the continuum to another.

Two types of scales can be formulated by a researcher. One is a rating scale made up
only of a single item. The other one is made up of several items an<l:is called a multi­
item scale. Between the two, the second assures greater reliability and validity since a
repeated measure of the same factor is made possible. Also, the complexities of the
attribute can be captured by several items incorporated in the test.

Some examples of each approach are provided, together with options for formulating
each type of scale.

REFERENCES

Bautista, Victoria
1 996 Partnership for Community Health Development Appraisal Report-Year III. Quezon
City: U.P. College of Public Administration.
Gay, L. R.
1996 Educational Research: CompetenciesforAnalysis and Application. Columbus, Ohio: Prentice
Hall.
Judd, Charles, Eliot R. Smith and Louise Kidder
1991 Research Methods in Social Relations. Forth Worth: Holt, Rinehart and Winston.
O'Sullivan, Elizabethann & Gary R. Rassel
1994 Research Methods for Public Administration. 2nd Edition. White Plains,N ew York:
Longman Publishers.

UP OPEN
UNIVERSITY
MODULE16

SECONDARY TECHNIQUES

INTRODUCTION

The previous modules focused on the different primary techniques of data collec­
tion. These techniques enable researchers to witness for themselves how an event
occurs by direct observation or by conducting interviews or fielding questionnaires
to get the testimonies of persons who experienced or are experiencing the event. We
have also seen how interval scales for enhancing the quantitative measurement of a
variable can be incorporated in interview schedules and questionnaires.

However, there are sources of information already collected that researches may be
able to use in answering their research questions. These sources of information
will be discussed in this module.

OBJECTIVES

At the end of this module, you will be able to:

Characterize secondary techniques of gathering information;


Explain how the researcher can maximize the utility of secondary tech­
niques; and
Define the different sources of information under secondary techniques,
suc:1 as statistical records, personal documents, and narrative documents.
232 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

1.0 SECONDARY TECHNIQUES DESCRIBED

Secondary techniques of data collection include methods of gathering information


obtained by other entities. Unlike in primary techniques where the researcher collects
the information directly from the persons who have knowledge of the events or factors
being studied, secondary techniques depend on the researcher's ability to obtain from
other institutions or individuals information they keep.

There are many such sources of information. These include statistical records such as
.
census and vital statistics collected regularly; personal records of personnel of public
offices and other agencies; and narrative documents such as annual reports, memo-
randa, office circulars, and other agency reports.

What are the strengths of secondary documents?

First and foremost is that they are economical as the materials have already been
collected by others. The researcher spends only for moving around to retrieve these
materials and for duplicating the materials for personal use.

Second, the materials can be retrieved anytime at the convenience of the researcher,
for as long as the gatekeepers of the information will give the researcher access to the
information. On the o�her hand, in primary techniques the timing of the investigation
is very important since information is gathered according to the availability of the
persons who will be providing the information. Also, information is collected at the
critical moment when the event is occurring or has just occurred. For instance, if we
wish to study how a program was introduced in a given area, enough time should be
allotted by the researcher to make sure that the respondent is able to remember how it
all started. This is not the case with secondary techniques since the required informa­
tion has already been collected by others and the researcher need not establish rapport
with the direct sources of information. This is the problem of another data collector.

Third, secondary techniques do not result in or lead to reactivity on the part of the
units of analysis. This is because they are not directly contacted by the researcher to
provide the information. In fact, information that is difficult to disclose in a person­
to-person dialogue such as an interview, can be obtained through secondary '5ources.
For instance, information about family income may be difficult to get in inter­
views. This information, however, is available in the personal record of the re­
spondent.

2.0 ROLE OF THE RESEARCHER

A basic responsibility of the researcher in dealing with secondary sources of informa­


tion is to establish rapport with the "gatekeepers of the information." While there is
less pressure in dealing with respondents or the subjects of the study as the data needed
have already been collected from them, it is important for the researcher to explain
well to those who have collected the information how he/she will use the information.
UP OPEN Much depends on the capability of the researcher to convince the bearers of informa­
UNIVERSITY tion. Oftentimes, a formal letter of request is required by the bearers of in formation.
_
DATA COLLECTION / Secondary Techniques 233

The letter of request can indicate: (1) what specific data are needed, (2) the purpose of
the study, and (3) the significance of such an investigation to the institution or com­
munity.

The researcher should also be sensitive to the reliability of the information collected.
S/he should always assume a questioning mind regarding the accuracy of the informa­
tion. The researcher who is relying on secondary records should ask how the informa­
tion was derived. The reputation of the institution or persons gathering the informa­
tion plays a critical role in ascertaining whether the data collected can be relied upon.
I distinctly recall that many had doubts about the status of the debt burden during the
martial law regime. Many people also doubted the official reports regard.ing the unem­
ployment rate. At the time, people questioned the trustworthiness of tne information
provided by the regime. In particular, information on performance and accomplish­
ments was taken with a grain of salt.

3.0 CENSUS DATA AND STATISTICAL RECORDS

One of the secondary sources of information is census data. These are often collected
regularly by institutions and have the advantage of enabling the conduct of trend stud­
ies. They are often aggregated regularly to summarize massive data drawn from a popu­
lation. They cover information pertaining to institutions or a given population being
served by their programs or projects.

Institutions normally keep a profile of their personnel, financial resources, institu­


tional outputs, and impact data. For instance, personnel offices are able to release the
total number of employed and the type of positions they occupy. Institutions also keep
track of how much resources they have, especially funding, and how these are dis­
bursed. Outputs, such as agricultural, health, or social welfare services, are often gath­
ered regularly and incorporated in annual reports.

Macro-data on government resources and expenditures are also kept and reported regu­
larly by the Department of Budget and Management. Thus, it would be easy to report
on trends and pattern� in how much money had been spent for particular sectors in
development (e.g., economic, social, development administration, military, and debt
burden). Social development expenditures could also be disaggregated for health, edu­
cation, housing, and social welfare expenditures.

The National Statistics Office regularly compiles and summarizes census data. It con­
ducts, for instance (NSO 1994):

Census of Population and Housing, a decennial undertaking to obtain an in­


ventory of the total population and housing units in the country. It gathers
information on the size and distribution of the population as well as informa­
tion about its demographic, social, economic, and cultural characteristics.

Census of Agriculture and Fisheries, a complete count of all farm operators in UP OPEN
the Philippines. The basic characteristics of each farm, such as size, location, UNIVERSITY
234 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

tenure status, source of irrigation, crops planted, livestock/poultry raised, agri­


cultural practices, etc.are recorded.

Census of Establishments, a comprehensive collection and compilation of sta­


tistical information pertaining to the economic activities of establishments in
the country such as agriculture, forestry and fishery, mining and quarrying,
manufacturing, electricity, gas and water; construction, wholesale and retail
trade, transportation, communication and storage, financing, insurance, real
estate and business services, and community, social and personal services. This
is conducted every five years.

Census of Buildings, an inventory of buildings in the urban areas of the coun­


try, their characteristics, usage and the economic activities engaged in by the
occupants.

The National Statistics Office also conducts surveys on the socio-economic status of
households (i.e., labor force, health, demographics, family income/expenditures; over­
seas Filipinos; household-operated activities; and functional literacy, education and
mass media) and industry and price (i.e., weekly and monthly price surveys; monthly,
quarterly and annual surveys of establishments; annual survey of construction projects
of local government; quarterly survey of employment and compensation of local gov­
ernments; and quarterly survey on employment and compensation in the national
government). Statistical data cover sampled respondents from a population.

The National Statistics Office also compiles administrative-based data on:

foreign trade, from import and export documents submitted by importers and
exporters to the Bureau of Customs;
domestic trade, based on the flow of commodities within and among the dif­
ferent regions of the country;
private building construction, based on data generated from building permits
issues; and
vital statistics, based on data generated from birth, death, and marriage certifi­
cates issued.

Statistical and census data are useful in responding to quantitative research prob­
lems and do not require much cost on the part of the researcher. Because they are
collected regularly, it is relatively easy to derive patterns and study these according to
particular perspectives or modes defined by the researcher. For instance, for a study on
social development commitment of various regimes (Bautista 1994 ), the data were based
on expenditures of various presidential terms in postwar Philippines reported in the
Philippine Statistical Yearbook published by the National Statistical Coordination Board.
Comparing social and economic development sectors, the data in Table 16-1 show that
lack of parity between social and economic development were experienced during the
two decades under Marcos.

UP OPEN Earlier years (under Garcia and Macapagal) already indicated the resolve to balance
UNIVERSITY both economic and social development. Following the Marcos administration, there
DATA COLLECTION/ Secondary Techniques 235

was nearly an equal consideration for both social and economic development, as dem­
onstrated during the terms of Aquino and Ramos. However, the overall expenditures
for the two sectors were remarkably lower because of the high percentage of the total
expenditures channeled to debt service, a consequence of excessive borrowings that
started under Marcos.

Table 16-1. Percentage Expenditures by the National Government

Presidential terms Economic sector Social sector Debt service

Garcia
(1958-1961) 32.8 36.0 5.5
Macapagal
(1962-1965) 31.6 36.1 6.2
Marcos
(1966-1975) 44.7 24.1 5.2
Marcos
(1976-1985) 34.2 19.5 13.6
Aquino
(1986-1991) 19.4 19.3 40.9
Ramos
(1992-1993) 21.4 21.0 37.3

Source: Bautista (1994)

In a study on the costs and benefits of tourism, Saguil (1993: 12-13) focused on tourist
receipts and viewed these as percentage of Gross National Product for the years 1986
to 1990.

Saguil observed that there was a decline in the payments of foreign tourists for the
goods and services they acquire in the country by foreign currency, which amounted
to f.31% of the country's GNP per year from 1986 to 1990. Although tourist receipts
declined from 2.15% of the GNP in 1986 to 1% of the GNP in 1990, the amount earned
was noted by Saguil to be a significant portion of the country's foreign currency re­
serves needed to pay for the country's imports and foreign debts. See Table 16-2:

Table 16-2. Philippine Tourist Receipts and GNP

Year GNP in $ Million Tourist Receipts TR as %0/GNP


(M) (TR) in $M

1986 30130 647.l 2.15


1987 34638 458.0 1.32
1988 39010 405.0 1.04
1989 44250 469.0 1.06
1990 46580 466.0 1.00
UP OPEN
Source: Saguil 1993. UNIVERSITY
236 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Saguil noted that tourism had the number 1 slot in 1989, followed by garments and
electronics. In 1990, it became second to electronics, like garments.

It is the researcher's responsibility to define new hypotheses not covered in a second­


ary source of information. The researcher is always burdened with the issue of: Am I
contributing anything new? Is the topic I am going to cover sufficiently tackled in the
data or report I am reading? Is there anything new that I can contribute which has not
been written about?

Since statistical data are often presented regularly, the researcher will have to deter­
mine in what way the information can be recast according to the purpose of the
study. For instance, since the definition of the region has been changed, a trend study
can be undertaken by reconstructing the data collected in the past according to the
components of the region based on the current definition. To be able to obtain longitu­
dinal patterns for the Autonomous Region of Muslim Mindanao, data derived for the
provinces that comprise it can be recomputed from previous accounts to have a sum­
mary profile for the region over a period of time.

4.0 PERSONAL RECORDS

Personal records serve as a source of raw data about the profile of people. They are
compiled regularly about each person and can be aggregated by a researcher to serve a
particular research purpose. Personal records include personal data sheets collected by
a personnel office such as those on income, absences, training programs attended,
educational profile, etc. These may also include legal documents on certain individu­
als or institutions for regulatory purposes such as applications for licenses for a fran­
chise to operate establishments (i.e., restaurants, hospitals, taxicabs). These may also
cover documents on individuals or entities with legal cases and income tax returns.

The information collected from certain groups of people can be used for both qualita­
tive and quantitative research problems. Some data sheets may be the source of insights
on patterns and sensitive information, which may he difficult to obtain in interviews
and questionnaire. For instance, completed cases filed witi1 the Sandiganbayan may
be studied to determine specific practices of corruptive behaviors.

The same source may also be utilized for quantitative queries, such as reviewing the
dominant types of cases that have been convicted in a given period.

For quantitative types of research problems, the responsibilities of the researcher in­
clude:

the need to clarify the variables of the study;


the need to spell out the categories of analysis for every variable of the study;
and
the need to define whether the document will be sampled or will be drawn for
UP OPEN the entire population.
UNIVERSITY
DATA COLLECTION/ Secondary Techniques 237

I recall my study on gender and mobility at the UP Diliman campus (Bautista 1990-91).
The study was based on the personal documents kept by the Human Resources Devel­
opment Office on faculty and staff with a regular appointment. The study covered:

demographic variables such as sex, age, civil status, current rank, educational
attainment and sectoral affiliation (i.e., faculty, research or administrative)
other variables such as rate of mobility (the number of steps a person has moved
using the Department of Budget and Management scale); research opportuni­
ties; and other incentives made available to the staff (i.e., trips abroad)

One of the highlights of the findings shows the lack of significant difference among
male and female faculty administrators in terms of educational preparation. However,
in rank, females are significantly lower than males: an average of 22.9 for the former
and 23.9 for the latter.

5.0 NARRATIVE DOCUMENTS

Narrative documents include those that are regularly prepared by agencies or specific
individuals within an institution, including development plans and annual reports.
The latter provide useful information on what institutions would like to accomplish
(plans) and what they are able to accomplish on a yearly basis (annual reports).

Narrative documents regularly prepared by specific individuals include minutes of


meetings, diaries of community organizers, and personal diaries, among others.

There may also be documents that are prepared occasionally by institutions or desig­
nated persons that may provide insights about specific events, such as special reports to
assess programs or projects, investigative reports of special cases that have deviated from
normal activities (i.e., reports on disaster, etc.), and reports on occasional conferences.

There are narrative documents that are open to public consumption like annual re­
ports. However, there is also private information. This type of information requires
sensitivity on the part of the researcher in eliciting the willingness of the keeper of the
information to disclose the material gathered. Personal diaries, for example, are a helpful
source of insights as events are recorded immediately by the writer. However, only
those who are gifted with skill in writing down their thoughts and feelings keep dia­
ries. They may not be representative of the population we wish to study.

Regular reports could provide trends or a longitudinal perspective about certain fea­
tures of the organization or events regularly featured. They are particularly useful in
qualitative studies as they provide accounts of certain events from the perspective of
the i_nstitution or individuals preparing these documents.

The role of the researcher who makes use of narrative documents is to:

clarify what variables or factors will be focused on; and UP OPEN


define the pattern peculiar to the material or set of materials obtained. UNIVERSITY
238 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

A case in point is the study of Carino (1995) on population policy in the Philippines.
In this study, Carino breaks down the periods of population perspective of govern­
ment into five, namely:

Family planning as independent activities of private organizations (prior to


1969);
The start of family planning as government policy (1969-1973)
Family planning and population as part of total development (1974- 1986)
Family planning subsumed under maternal and child health (1987 to 1992)
Policy balancing population concerns with that of resource and development
(1992 to the present)

These periods were identified using various documents such as: State of the Nation
Addresses of Presidents, population plan, accomplishment reports, population policy
statements, academic assessments, and presidential policy directives. These were also
confirmed by interviews of academicians and key executives of the Commission on
Population.

Danilo Reyes also relied substantially on books, monographs, and the Philippine Jour­
nal of Public Administration published by the UP College of Public Administration in
order to characterize the evolution and growth of the discipline of public administra­
tion in the Philippines as the pioneer institution in the study of public administration
in the country and the key institution in organizing the Association of Schools of
Public Administration in the Philippines. Reyes used the source materials produced
by the institution as bases for the phases of the institution's growth he identified in his
study. He labelled these phases as:

The Beginnings (1952-1956)


Foundations and Growth (1957-1972)
The Transition Years (1973-1981)
Activist Public Administration: Reform and Reflection (1982-present)

The researcher must also be resourceful to be able to obtain narrative reports that are
kept on file. The researcher can raise probing questions to make sure that the materi­
als that could provide a wealth of information are available and can be utilized.

It is the researcher's responsibility to make sure that the narrative materials collected
provide a comprehensive view of the factors he/she wishes to study, rather than just a
one-sided point of view of the topic. For example, in writing about the processes ap­
plied in the introduction of a program or a project, a researcher can balance the reports
written by the frontline workers with those by representatives of community leaders
who have had the opportunity to work with the program.

It is time for another SAQ!

UP OPEN
UNIVERSITY
DATA COLLECTION / Secondary Techniques 239

SAO 16-1

Fill in the blanks with the correct answer.

1. Between primary and secondary techniques, techniques pose less


reactivity on the unit of analysis.
2. In secondary techniques, it is important to establish rapport with the ____
of the information.
3. Among the different secondary data sources, enable the conduct
of trend studies showing quantitative data.
4. as type of secondary data source can be helpful in providing infor
mation about the individual which may be difficult to ask in interviews.
5. Annual reports, diaries and minutes of meetiRgs are examples of _____
as a type of secondary data source.

ASAQ 16-1

s:iuawnJop aAqeJJPN ·s
s+uawnJop 1euosJad ·,
SplOJaJ lPJqsqe+S £ .£
sJadaa�a+e9 ·z
AJepuoJas ·1

If you get less than 3 points, review the module

ACTIVITY 16-1

Prepare and complete your tools for gathering data.

UP OPEN
UNIVERSITY
240 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

SUMMARY

Secondary techniques make one rely on sources of information already collected by


other institutions and individuals.

Secondary information is economical; can be retrieved at the convenience of the re­


searcher for as long as the gatekeepers of the information make them accessible; and
do not stimulate reactivity on the part of the unit of analysis.

It is the researcher's responsibility to establish rapport with the gatekeepers of infor­


mation and be sensitive to the accuracy of the infoxmation collected by other entities.

Statistical data, which are one source of information, can be useful for quantitative
studies conducted over a period of time. The researcher should be able to determine
what new hypotheses can be answered by the set of data he/she is going to use. In some
instances, some of the data can be recast in order to suit one's research focus.

The use of personal records, another source of information, can respond to both quan­
titative and qualitative research problems. Personal records contain information about
individuals who may be embarrassed to answer questions in interviews and question­
naires. For quantitative purposes, it is important for the researcher to clarify the vari­
ables of the study, the categories to be adopted in classifying the information, and
whether or not the data sheets will be sampled.

Narrative documents are excellent sources of insights about processes and histories of
the implementation of programs and/or projects and the operation of the institution.
They are composed of documents that are regularly or occasionally prepared. Some­
times, they are for public consumption. At other times they are private and therefore
pose a challenge to the researcher in convincing the gatekeepers to make them accessi­
ble. However, the data qm be questioned for their representativeness of the population
as they are often written only by those who have the facility with articulating their
views and feelings.

The researcher should be able to clarify what variables will be focused on and derive
the patterns reflected in the narrative documents reviewed.

UP OPEN
UNIVERSITY
DATA COLLECTION / Secondary Techniques 241

REFERENCES

Bautista, Victoria
1994 Trends and Patterns in Social Development Efforts of the Philippine Government.
Public Administration Bulletin Vol. 3, 3-4 (September).
1990-91 Excerpts from the Preliminary Report on the Status of Women in an Educational Setting:
T he Case of the University of the Philippines (Diliman). Review of Women's Studies,
Vol. 1, 2.
Carino, Ledivina
1995 Population Policy of the Philippines. Philippine Journal ofPublicAdministration. XXXIX,
1 (January).
National Statistics Office
1994 National Statistics Office. Manila: Management, Planning and Coordinative Staff.
Reyes, Danilo
1995 Life Begins at Forty: An Inquiry on Administrative T heory in the Philippines and t
he Structure of Scientific Revelations. Public Administrtion by the Year 2000. Proserpina
D. Tapales and Nestor N. Pilar, eds. Quezon City: U.P. College of Public Administra
tion.
Saguil, Dan
1993 T he Benefits and Costs of Philippine Tourism Industry. Public Administration Bulletin
Vol. 2, 2 (June).

UP OPEN
UNIVERSITY
PROCESSING, DATA ANALYSIS
AND INTERPRETATION
Dear Student,

This unit, consisting of two modules, caps the entire course.

Module 17 discusses the basic principles of data collection, processing, analysis


and interpretation of data.

Module 18 gives you basic instructions on how to write the report.

Included as an annex to this manual is a set of instructions on how to apply the


Microstatistics program to facilitate the processing of information you will derive
from your research. The program is particularly helpful if you have a quantitative
research undertaking. In fact, this is what I encourage you to do in order to make
you practice what you have learned in PM 299.1. It is presumed here that you
already have a lot of practice in the qualitative approach in your other subjects.
This will also build your confidence in using quantitative techniques for decision
making purposes.

Completion of these modules should enable you to implement your proposal. This
will also enable you to process and analyze your data in preparation for writing the
report.

I hope you had an enriching semester. Try to complete your research according to
schedule.

God bless.
MODULE 17
DATA COLLECTION, PROCESSING,
ANALYSIS, AND INTERPRETATION

INTRODUCTION

In the previous modules, we discussed three important aspects of the methodo­


logical part of a research proposal: research design, unit of analysis, and techniques
of data collection. A last component in the methodological aspect of research work
is defining the tools to be used for data analysis and interpretation. Since this has
been discussed in a previous course, I will no longer discuss this separately. I will
simply review this with you in this module, as I highlight what you have to re­
member as you implement your proposal and process the information you have
gathered.

OBJECTIVES

In this module, you will be able to:

Process research data;


Review the different tools for data analysis; and
Review how to analyze and interpret data, especially for explanatory types
of studies.

1.0 DATA COLLECTION

Having prepared the research proposal and received the approval of your profes­
sor, tutor or rese2-rch superior, you are now ready to proceed to data collection. As
you do so, bear in mind the following guidelines:
248 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

1. Pretest your instrument before going to the field to implement it. The corrections
on the instrument should be incorporated before it is reproduced.
2. Know the number of respondents you plan to reach and reproduce enough copies
for them. Make extra copies in case you commit errors.
3. Have a general knowledge of the kind of respondents you will be interacting with.
Be sensitive to their expectations and needs. Make sure you have translated the
instrument to the dialect of your respondents.
4. Communicate with the key officials or officers of the institution or community
you are planning to study. They should know how long you will be in the area and
what you expect their constituents to do. It may be necessary to write a formal
letter detailing the:

purpose of the study


persons to be contacted and the duration of the visit
materials needed, if any
endorsement/approval of the person or institution supervising your work
importance of your study to their institution/community

5. Review the instruments you will be using before you go to the site. If you are
involving other interviewers or researchers, make sure you have trained them well
before deploying them.
6. Do not impose on your respondents/subjects beyond having them answer your
instrument. Do not expect them to give you snacks or anything of the sort. Make
sure you visit them at a time that is convenient to them. If you are visiting them at
home, do not go at an unholy hour, such as during mealtime or siesta.
7. Record any deviation from your proposal. Deviations should also be mentioned in
the report. An example of a deviation is when your samples are not willing to
participate and what you did to resolve this (e.g., getting a replacement).
8. Document the problems you encounter (i.e., resistance on the part of your respond­
ents to answer some questions and why) in collecting·data so that other research­
ers may learn from your experience.
9. Thank your respondents.
1q. Acknowledge officials/officers who gave you permission to gather information.
11. Fulfill your promise to the institution/community. If you promised to give them a
copy of the report, make sure that you do so.

2.0 PROCESSING DATA

Once you have collected all the data you need, then you can move on to the processing
of data. This entails transforming your data into a form that will facilitate data analysis
and interpretation. Some steps in data processing are:

categorization/culling patterns
coding
application of appropriate statistical tools
UP OPEN
UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 249

2.1 CATEGORIZATION

Categorization is the formulation of categories for nominal variables, such as defining


your respondents' gender (categorized into male and female), religious profile (grouped
into Catholic, Protestant, Iglesia ni Kristo, etc.), and educational profile (categorized
into those who obtained some elementary education, graduated from elementary school,
obtained some high school education, graduated from high school, etc.).

Normally, categories are incorporated in the interview schedule or the questionnaire if


you are dealing with fixed alternative types of questions. This can be done prior to
data collection.

However, if you are dealing with open-ended questions, categorization is undertaken


after data collection, and is therefore an important phase in processing information.

This can be done by:

1. Formulating a list of responses from all the respondents of the study

For example, in a study assessing the reasons for non-participation in a referen­


dum on Charter change, the following data were obtained from 20 non-partici­
pants in a barangay.

Respondent Reasons for Not Voting (Item 40)


Number
1 I don't believe it will do anything.
2 I don't know my precinct. I had been looking for it the whole day.
3 No answer
4 It's a waste of time. I don't agree with it.
5 It will not do anything anyway.
6 A bigpalabas. I don't have faith it will do something good.
7 Paralyzed na ako.
8 I gave birth to a baby boy on the same day.
9 Nobody could attend to my five-month-old baby.
10 I was down with flu.
11 I did not know the polling place.
12 I had an asthma attack.
13 I do not have faith in the process.
14 I am physically handicapped-blind.
15 I was hospitalized on the same day.
16 I went to the province on an emergency trip. My father had a heart attack.
17 I did not know where to cast my vote.
18 I don't have faith in the process.
19 I do not agree with it.
20 I am not in favor of the process.

UP OPEN
UNIVERSITY
250 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

2. Developing a set of categories from the list of responses

From the preceding list of responses, I formulated the following categories :

Lack of faith in the process (8 respondents-I, 4, 5, 6, 13, 18, 19 and 20)


Lack of unfamiliarity with the precinct (3 respondents-2, 17 anq 11)
Physical incapacity (6 respondents-7, 8, 10, 12,14 and 15)
Other reasons (2 respondents-9 and 16)
No answer (1 respondent-3)

Make sure that the categories formulated are:

Mutually exclusive-A response can be subsumed in only one category.


Inclusive-All responses can be subsumed in the categories formulated. If, for
instance, a "No response" category is not included in the earlier list of re­
sponses on reasons for non-involvement in the referendum, the respondent
who did not answer the question can not be accounted for in the listing.
For a fixed ratio or interval scale variable, categories can be formulated accord­
ing to cut-off scores for certain variables from reputabfe sources. For instance,
school children are classified by Department of Education Culture and Sports
thus:
Elementary (7-12 years old)
Secondary (13-16 years old)

In case there is no conceptual basis for the formulation of categories, these can be
formulated according to the following formula:
Highest Score - Lowest Score + 1
Interval =
Number of Classes

For example, in a question about the total number of activities a respondent is engaged
in, the community with a range of scores from 1 to 9, and where the desired number of
classes or category is 3, the interval is:
9- 1 + 1
Interval =
3

Thus, the lower limit for the first category is 1. For the second, it is 1 plus 3 (the
interval), which makes 4. And then, 4 plus 3 is 7 for the third category. The categories
are as follows:

1-3 Low
4-6 Medium
7-9 High

The labels could be formulated by the researcher if there is no conceptual basis for
UP OPEN grouping the scores.
UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 251

Do Activity 17-1 to practice categorization.

ACTIVITY 17-1

The following are some managerial problems encountered in the past y�ar by bu­
reau directors/service chiefs of a department. Classify the responses by formulationg
categories.

Respondents Responses

A Lack of funds to implement projects


B Internal politics and intrigues
C Unclear delineation of authority and responsibility
D Lack of qualified and trained personnel
E Poor systems and procedures
Interference in recruitment, selection and appointment of personnel
F Factionalism among the staff
G Poor system of communication
Unclear agency objectives/policies
Poor control procedures

See some suggested categories below!

Answers to Activity 17-1

C 'I 'H '3 'J - sa1n J JO kgJEl) ·t:


9 ':J 'o '8 - 1 auuosJad ·c
\f - 6u�pun:J ·i

2.2 CODING

Coding entails transforming responses into symbols, usually numerical, especially if


the information to be processed is for computer operations.

Why is coding an important step in processing data? UP OPEN


UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 251

Do Activity 17-1 to practice categorization.

ACTIVITY 17-1

The following are some managerial problems encountered in the past year by bu­
reau directors/service chiefs of a department. Classify the responses by formulationg
categories.

Respondents Responses

A Lack of funds to implement projects


B Internal politics and intrigues
C Unclear delineation of authority and responsibility
D Lack of qualified and trained personnel
E Poor systems and procedures
Interference in recruitment, selection and appointment of personnel
F Factionalism among the staff
G Poor system of communication
Unclear agency objectives/policies
Poor control procedures

See some suggested categories below!

Answers to Activity 17-1

r 'I 'H '3 '] - sa1n1 jO AWED "£


9 'j 'o '8 -1auuos1ad ·z
v - 5u�punj ·1.

2.2 CODING

Coding entails transforming responses into symbols, usually numerical, especially if


the information to be processed is for computer operations.

Why is coding an important step in processing data? UP OPEN


UNIVERSITY
252 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

First, coding allows you to economize on the space to be used in storing data. Imag­
ine that you are dealing with thousands of respondents and you have to store raw data
embodied in instruments in your filing cabinets. T hey will take up a lot of space! On
the other hand, a standard coding sheet can accommodate 80 variables/factors and 30
respondents. Figure 17-1 is an example of a coding sheet.

Second, for manual operations, having a coding sheet facilitates cross-tabulations or


reading two variables simultaneously. For instance, we may be relating sex (reflected
in Column 5-coded as 1 for male and 2 for female) and citizen participation in the
community (found in Column 27-coded as 1 for involved and O for uninvolved). To
make the tabulation process easier, the researcher lnay fold the coding sheet and view
these two variables side by side. Figure 17-2 is an example.

If data are not coded, the researcher will go through each instrument to locate the
response of every respondent in the instrument to be able to tabulate the information.

Figure 17-1. Coding sheet

C •·•·••.. -•

UP OPEN
UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 253

Figure 17-2. Preparing coding sheets for manual tabulation

identification J J
sex citizc� participation

column number - I 2 3 4 5 27

- () 0 () I I ()

codes () () {) 2 2 I

[ __ 0 () () 3 I ()

Take note that in Figure 17-2 we have put the columns being reviewed (5 and 27) side
by side.

While categories are formulated to summarize the information for open-ended ques­
tions, it may be helpful to also quote some passages from the responses to the ques­
tions in the research report to give the reader a flavor of how the respondents answered
them.

Third, coding is a must in computer operations. The computer recognizes only num­
bers as symbols. Therefore, it is important to translate qualitative information into
numbers. With computers, coding is made even simpler. Coding can be immediately
done on the worksheet of the program being used. There is no need for a coding sheet
unless you would like to retain a hard copy prior to coding the information in the
computer.

In this manual, I am introducing the Microstatistics program since it requires only


one diskette. There are other programs available such as the Statistical Package in the
Social Sciences (SPSS). However, SPSS requires more than 10 diskettes and should be
copied to a hard disk. If you are using Microsoft software, MS Excel can be used for
making computations. However, Microstatistics and SPSS have statistical programs
that can be operated immediately.

When is coding implemented?

Coding may be implemented during data collection when responses are coded
and embodied in the instrument. Figure 17-3 provides codes for responses in a
questionnaire.

UP OPEN
UNIVERSITY
254 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Figure 17-3. Pre-codes in a questionnaire

A. Profile of respondent (Please check the appropriate answer in the space


provided.)

1. Current position:
(1) Technical
(2) Administrative
(3) __ Supervisory
2. Age:
3. Sex: (1) Male (2) Female

Pre-codes can easily be done for the fixed alternative type of questions. However, if
most of the questions are open-ended, coding can take place only after data collec­
tion, when the responses have been categorized already and prepared for coding.

Also, pre-coded or not, quantitative data is always coded.

What are the materials necessary to implement coding? These are a coding sheet and
a code book.

An example of a coding sheet was shown earlier. This is where the codes are put,
especially if you are processing data manually.

In the coding sheet:

Rows stand for respondents. The first set of columns represents the identifica­
tion number of the respondents. If the total number of samples is 300, assign
columns 1 to 3 for identification number. Thus, the first identification number
is 001 and is on the first row.

Columns stand for variables. If you have a value or score for a response that
has more than one digit, assign more columns to correspond with the number
of digits that a response could possibly have.

For example, the identification number of the respondent is assigned to Col­


umns 1-3 since it can accommodate a number with three digits. Columns 4-5
can be allocated for age, which is expected to be answered with at most two
digits as indicated in Figure 17-4.

UP OPEN
UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 255

Figure 17-4. Allocating columns for the identification number of respondents

identification age

I 2 3 4 5 6 7 8 9

I
first respondent 0 () I
I

Cells contain the scores or assigned values for categories of nominal variables
(i.e., sex, which may be coded as 1 for male or 2 for female). Remember that a
cell can contain only one score or value ranging from Oto 9.

However, if you are using the computer to processing information, you can do
this directly on the worksheet of the program you are using. Unlike the coding
sheet, every cell in a computer worksheet where a variable is located can record
the exact number that you will register. In a standard coding sheet, we are
restricted to single digits per cell. An example of a coded data set using a statis­
tical package is found in Annex A.

Also important in coding is the preparation of a code book. This contains the instruc­
tions for coding the answers of the respondents. The code book ensures that the
responses are dealt with in a standardized way. It also guarantees that the researcher
will not forget what was done with the answers of the respondents. For manual opera­
tions, the code book contains the following information:

Column number
Variable represented
Categories of the variable (especially for a nominal variable)
Code (for the categories of a nominal variable) or the actual values for ordinal,
interval and ratio-scaled variables

For direct processing of information with a computer, you need not assign column
numbers as one aspect of the code book. You replace column number with the number
of the variables referred to. (See Figure 17-6).

Take note that there are code numbers that can be used consistently for all the re­
sponses in the instrument. For example, a "Not applicable" response can be coded as
9; "No response" can be coded as 8. Using the same codes for the same categories will
avoid confusion in dealing with a response in processing. This can be incorporated in
the general instructions for the coder. Figure 17-6 is an example of a code book for
computer operations.

If you will be needing more than one coding sheet per respondent, add another col­ UP OPEN
umn and indicate the card number you are referring to for some variables. If you are UNIVERSITY
256 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

going to the second sheet, there should be a mark in the coding sheet indicating that
you are in sheet number 2. See how the instruction on this is incorporated in Figure
17-5.

Remember to store the coding sheet and code book carefully. In case you would like to
retrieve them, you should know where to locate them. Other researchers may also
want to reexamine your data and use them for a different purpose.

Figure 17-5. Sample code book for manual operations

General instructions:
For items without any response from the respondent, use code 8 for a variable that can
draw a single digit value and 88 for a double digit variable.

For variables that are not applicable to the respondent, code 9 for a single digit variable
and 99 for a double-digit variable.

Specific instructions:

Column Variable Item no. in Categories Code/Score


number the instrument

1-3 Identification Copy as is,


number starting
with 001

4 Card number

5 Gender #2 Male 1

2
I I
Female

6 I Educational #3 No schooling I 0
attainment

With some elemen-


I tary education 1

Elementary graduate 2

Some high school 3

I High school graduate


I
4

#4 I
I
7-8 Age Copy as is
UP OPEN
UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 257

Figure 17-6. Sample code book for computer operations

Colunm Variable Item no. in Categories Code


number instnunent

Identification Left margin, Copy as is;


number page 1 number
start with 1

2 Gender #2 Male

Female

3 Educational #3 No schooling 0
attainment

With some elementary 1


education
I i I
I I Elementary graduate 2

I
Some high school 3
I

4
I I
High school graduate

2.3 APPLICATION OF STATISTICAL TOOLS

In quantitative types of research problems, you apply statistical tools to summarize a


set of data. If you recall your statistics, you have many options for aggregating your
information for data analysis and interpretation. The choice depends on the type of
variable you have formulated and the type of research problem you hope to answer.

In a descriptive research problem, you have the following options:

1. For nominal variables, univariate tables and charts may be useful in simplifying
your data.

You may also use proportions, percentages, and ratios to aggregate the informa­
tion for every category of the variable you are presenting.

2. For continuous variables (or variables with assigned scores or values),

central tendency or general patterns can be shown by the mean, median, or


mode.
dispersion or variations can be shown by standard deviation and range. UP OPEN
UNIVERSITY
258 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

In an explanatory problem where two variables are being related, the following infer­
ential statistics may be applied:

1. Two nominal variables

For two nominal or discrete variables, the appropriate statistical tool is a chi-square
test. This is applied based on a bivariate or two-dimensional table. For example, the
variable "gender" (categorized into male and female) may be related with "participa­
tion in community activities" (with answers grouped into those who got "involved"
and those who were "uninvolved").

Remember that you are locating the categories of the independent variable along the
columns (gender is the likely independent variable here) and the dependent variable
along the rows.

A chi-square test can read only the significance of the relationship of the two variables
being examined. It will not read the pattern of relationship. You will have to apply
percentage or proportions analysis to show the pattern of relationship.

Hypothetical data demonstrating the relationship between gender and participation


in community activities are shown in Table 17-1.

Table 17-1. Hypothetical data relating gender


and participation in community activities

Name of participation Male Female

I
F* % F %
Got involved 80 80 120 60
Uninvolved 20 80 80 40
TOTAL 100 100 200 100

*F stands for frequency.

Relying on raw data alone can be misleading since the frequency count for males who
are involved total 80. On the other hand, the females who got involved number 120.
Thus, I have chosen to apply percentage statistics to standardize the figures.

Proportions may be used if the total number of cases is 30 or less.

With the percentage analysis, 120 female respondents constitute only 60% of the total
females as against 80% of the males.

In applying proportions/percentage statistics, compute along the direction of each cat­


egory of the independent variable (i.e., male and female categories for gender).
UP OPEN
UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 259

In analyzing data, however, read the data horizontally. In other words, you are con­
trasting the percentage of males and females who got involved in community activi­
ties. I would analyze the data presented here in this manner:

Eighty percent of the males got involved in community activities as against only 60%
of the females. The data indicate to us that males, more than females, participate in
community activities.

Oftentimes, students read or contrast information along the column. This would high­
light only a description of a category of the independent variable such as the following:

Eighty percent of the males got involved and 20% of them were uninvolved. On the
other hand, among the females, 60% were involved as against 40% who were unin­
volved.

In the latter mode of analysis we become too absorbed with a discussion of males, then
the females, and forget about how we can contrast them to highlight the difference.

Box 17-1. How to analyze a table

Compute % downward.
Read data sideways.

Take note that I started with the factual presentation before I gave my conclusion or
generalization. This is inductive reasoning. A generalization is made after facts are
presented.

The presentation could also be reversed. That is, you may start with a generalization
before presenting factual data. This is deductive reasoning. For the information given
above I may present my analysis thus:

The data indicate to us that males, more than females, participate in community ac­
tivities. This is borne out by the fact that 80% of the males got involved in community
activities as against only 60% of the females.

However, do not forget to cite the chi-square test results to indicate the significance of
the relationship that is established. You may incorporate this in the presentation of the
data. This may be stated thus:

The relationship is statistically borne out with a chi-square test (computed value of
11.1) with a significance level or a of .0008.

Also, in the discussion of findings, do not forget to give your interpretation. This
means that you bring in the theoretical foundation of the hypothesis embodied in
your theoretical framework. In other words, you provide an explanation for a pattern UP OPEN
that is observed. This may bear out your theoretical argument. If not, then you have to UNIVERSITY
260 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

provide another explanation, which again goes back to a theoretical argument. For
example, in the preceding observation, the explanation may be given as follows:

More males than females may be involved in community activities since males are
expected to assume more external responsibilities. Women, on the other hand, may be
tied down to household chores and therefore find difficulty in engaging with additional
responsibilities outside of the home.

You see, this argument could be the subject of further investigation. But for the data
collected in this particular study, it is an explanation offered to show why males and
females differ in their behavior.

2. Nominal and continuous variables

In relating a nominal or discrete variable and a continuous variable (a variable that is


scored through interval or fixed ratio scale), two general options are available:

Option 1: Reduce your continuous variable to a nominal or discrete variable.


Then summarize your data into a table. Then apply a chi-square test.

Option 2: Retain the character of both variables but obtain the mean of the
continuous variable for each category of the nominal variable. Then apply a
difference of means test or one-way analysis of variance CANOVA).


M F


UP

l)OWN
I/

To demonstrate Option 1, let us compare gender (classified into male and female) and
level of involvement in community affairs (which may be measured by the total number
of activities each is engaged in ranging from 1 to 9). These two variables can be tallied
UP OPEN to formulate a bivariate table where participation is simply categorized into High (5-9
UNIVERSITY activities) and Low (1-5 activities). Thus, we may obtain the following data:
DATA ANALYSIS/ Processing and Analysis 261

Table 17-2. Hypothetical data relating sex and level of involvement


in community activities

Level ofInvolvement Male Female

F* % F %
High 160 61.5 130 52
Low 100 38.5 120 48
TOTAL 260 100 250 100

Note: Computed chi-square is 4.30; a= .04.

You may then follow the same procedure you did in an earlier example showing how to
analyze the results in a bivariate table, and test for significance using a chi-square test.

For Option 2, you are to obtain the mean of the continuous variable per category of the
independent variable, applying either a difference of means test (Z test for large sam­
ple or t test for small samples) or an analysis of variance (AN OVA). The advantage of
AN OVA is that you may have more than two categories for the nominal variable and
still be able to obtain the mean for the continuous variable.

Instead of collapsing your continuous variable as you did in Option 1, in this option
you retain the actual value of the continuous variable. Given the example in Option 1
where we relate gender and level of involvement in community affairs, we now obtain
the following profile of males and females:

Mean level of involvement

3.5 4.8

Females Male

A difference of means test may yield, for instance, a value of3.6, which is significant at UP OPEN
alpha ( a) of .0002. UNIVERSITY
262 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

With the data, you may proceed with analysis and interpretation. Given the hypotheti­
cal data presented above, you may argue that:

Males and females differ significantly in their level of participation. Males demon­
strate a mean level of involvement of 4.8 vis-a-vis females with a mean of 3.5. This is
statistically borne out by a difference of means test with a Z of 3.5 (a of .0002). Thus,
it can be argued here that gender is related to the level of community involvement, with
males manifesting a higher level than females.

In an AN OVA, the value of the F test is reviewed and the actual significance level ( a)
of the F value. In computer operations, the exact talue of the alpha is provided.

(How did I present my arguments here? Would you say this is inductive or deductive reasoning?
See the answer in Box 17-2 on page 264.)

c. Two or more continuous variables

If you are dealing with two or more continuous variables, the tools that can be applied
include:

Correlation or Linear Regression (for two continuous variables)


Multiple Regression (for more than two continuous variables)

In a correlation test, the pattern of relationship is indicated by the positive or negative


valence in the test result (r value). Let us remember that the correlation result ranges
from a scale of:

-1 ................. 0 ............... +l

Hence, a computed value of r ( + .35) or correlation of two variables relating educa­


tional attainment (measured by the total number of years of schooling) and level of
participation in community activities (measured by the actual number of activities
each respondent is engaged in) with a critical value of +/-.3065 with an alpha of .05,
indicates that the correlation is positive. The computed value of .35, which is higher
than the critical value, shows that it is significant. This means that an increase in level
of educational attainment leads to a corresponding increase in level of community
participation. Contrariwise, the lower the educational attainment, the lower the level
of community participation. This finding, for instance, may be explained by the fact
that those with higher educational attainment appreciate the value of citizen partici­
pation. They may also be more articulate in expressing their views and therefore have
more confidence about getting involved in community activities.

A negative valence in the correlation test signifies an inverse or indirect relationship


(r= -.35), not the absence of a relationship. Given the same set of variables, the higher
the educational attainment, the lower the level of participation. On the other hand, the
lower the educational attainment, the higher the level of community participation.
UP OPEN
UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 263

No relationship is indicated by obtaining a value that is close to 0. Nevertheless, the


lack of significance will be indicated by the computation of the actual critical value
either converting this to a Z score or obtaining a computer-generated significance
level for an acceptable cut-off value.

Doing a regression is another option in relating continuous variables. The advantage


of a regression analysis are:

R squared (R2) or coefficient of determination indicates the percentage of time


the independent variable influences y (dependent variable).
Regression coefficient can predict the value of y if you vary x.:
Regression coefficient can indicate the pattern of relationship (if positive or
inverse).
Many independent variables can be included in the test, which a correlation
will not be able to do.
A linear regression (for a study of two variables) or a multiple regression (for
more than two variables) can be undertaken.

The data below show the relationship between number of police cars and the speed of
the motorists. The argument is that there is an inverse relationship between the two
factors since more policemen will be able to monitor the movement of the vehicles.
Suppose we have the following sample data:

Cars(x) Speed (y) in km/hr.

3 64
1 71
4 61
5 58
7 56

Through visual inspection, we can see that there is a decline in the speed of the car
with the increase in the number of police cars. With 1 car, the speed is 71 km/hr with
5 cars, it is 58 km/hr. However, we can make an estimate of how much this decline
occurs by computing the regression coefficient and determining how significant it is.
This is why regression statistics are very helpful.

Table 17-3 shows the result of the regression test.

The data tell us that:

T he relationship is significant (based on the F ratio of 49.075 with a signifi­


cance level of .005 (since we read the probability with E-03 as three decimal
places to the left of 5).
94% of the time, x will influence y (based on R squared of .94).
The influence of x on y is inverse (viewing the regression coefficient values: if
the number of police cars is -2.55, the speed will be 72.20). UP OPEN
UNIVERSITY
264 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Table 17-3. Sample regression test

Variable Regression Standard error t test Prob.


Coefficient

Polcars -2.55 .36 -7.005 .006


Constant 72.20
R squared= .93
R = -.97
Analysis of Variance Table :

Source Sum of D.F. Mean F Ratio Prob.


Squares Square
Regression 130.05 1 130.05 49.075 5.973E-03
Residual 7.95 3 2.65
Total 138.00

We may estimate the value of y if there are two cars by using the regression coefficient
data:
y A+ Bx

Where: y is the value of the dependent variable (speed) given x (the number
of police cars)
x is the value of the independent variable that is given
A is the constant, or the value of the dependent variable when the
value of the independent variable is nil (zero)
B is the value of the independent variable which does not yield a sig­
nificant effect

Thus, if there are two cars:

y = 72.2 + (-2.55) (2)


= 72.2 - 5.1
= 67.1

A multiple regression test can demonstrate the significance of various independent


variables in relation to a dependent variable. For instance, in a set of hypothetical data
for 30 respondents in a barangay, three factors (age, educational attainment, and in­
come) were related with involvement in community activities. The regression test is
shown in Table 17-4.

UP OPEN
UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 265

Table 17-4. Multiple regression of factors related to community involvement


of sampled household respondents in a barangay*

Variable Regression Standard t test Prob. Partial R 2


Coefficient error
age -2.21E-02 3.662E-02 -.603 .55142 .0138
edat 1.37 .24 5.76 .00000 .5607
income 5.085E-04 l.l10E-o3 .458 .65076 .0080
constant .86
Standard error of the Estimate = 1.72
Adjusted R2 = .62
R 2 = .79

*F Ratio = 13.953, probability = l.287E -05

The preceding data tell to us that:

The model fits. At least one independent variable among the three influences
the dependent variable. This is indicated by the fact that the F test of 13.953
yields a significance level of .00001 (based on probability of l.287E - 05).

Educational attainment influences the dependent variable. This is indicated


by the t test result (5.760) which has a significance level of .0000. Two other
factors (age and income) have obtained low significance levels for their regres­
sion coefficients (.5512 for the first and .65076 for the second).

The relationship between educational attainment and activeness in commu­


nity affairs is positive. This is borne out by the regression coefficient which
shows that an average grade improvement of 1.37 can yield a mean involve­
ment of .86. This may be because educational attainment could have provided
the individual the facility in expressing his/her views and the confidence to be
involved in community activities.

However, it is encouraging to note that age and income do not matter in moti­
vating a person to get involved. Whether one is young or old, the inclination to
participate is demonstrated. Furthermore, the opportunity to participate is not
the prerogative of a particular class.

The net influence of educational attainment is 56 %. This can be gleaned from the
partial R2 of .5607. The two other results for partial R2 are insignificant with
only 1.3% of the dependent variable being attributed to age and 0.8% for income.

On the whole, the combination of the three factors yield an influence of 57%
judging from the R2 of .57.
UP OPEN
UNIVERSITY
266 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Computer programs are available to facilitate the computation of voluminous data. In


Annex 17-A, you are given instructions on how to implement Microstatistics. Sample
data are provided so you can practice applying it.

See how much you understood by doing the SAQ below. Good luck!

SAO 17-1

I. Read, analyze, and interpret the following ta�le:

A government insurance system commissioned a survey team to investigate the


attitude of its clients to the office. A week after completing a visit to the office,
clients were asked whether they thought that they had to complete too many
forms and whether they felt the agency was run efficiently or inefficiently. Given
the accompanying cross-tabulations of their responses, what can the office do to
improve its image with clients?

Too many forms

Opinion of the office No Yes


Inefficient 150 140
Efficient 170 60

II. Identify the appropriate statistical tool.

Problem set: The following are some of the variables investigated by a research
team to determine factors related to national legislators' commitment to social
development. The key variables were operationally defined as follows:

1. Advocacy for social development-ascertained by the total number of bills


passed by legislators with a social development commitment
2. Legislator's position-distinguished as Senator or Congressmen
3. Educational attainment of legislators-distinguishing those with a social
science orientation vis-a-vis those without
4. Participation in seminars/workshops with social development orientation­
the total number of seminars/workshops legislators who have participated
in for the last five years in preparation for social development
5. Gender-being male or female
6. Satisfaction with performance in advocacy-whether or not the legislator is
satisfied with the number of bills advocated for social development

Direc6on: A list of variables to be related is provided in the next page. Recommend


a particular statistical tool to determine the relationship being established.
UP OPEN
UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 267

If you are recommending the application of:

Ch;-square: Formulate a dummy table and indicate the location of the independent
and dependent variables.

A difference of means test: Identify the groups that would be compared and the
corresponding factor that would be computed for the mean per category of the
nominal variable.

Regress;on test or correlatl'on: Indicate the independent variable and the depend­
ent variable.

1. Legislators' position and satisfaction with their level of advocacy for social
development
2. Gender and advocacy
3. Educational attainment and advocacy
4. Participation in seminars on social development and advocacy
5. Sex and educational attainment

III. Read, analyze and interpret the following statistical results:

1. A researcher asessed the concern for equity of administrators who had the
benefit of exposure from a training program (Group 1) undertaken by the
Civil Service Commission to make them concerned with this value. They were
compared with another group of adminstrators who did not benefit from this
program (Group 2). Thirty participants per group were given the test. The
highest possible score that can be obtained from the test is 20 to indicate
concern for equity. Results of the test show the following:

Group 1 with a mean of 18 t test= 6.42


Group 2 with a mean of 12.9 a= .000

On the basis of the test, would you recommend the reinstatement or the
scrapping of the program?

2. An exploratory test was undertaken relating the administrator's performance


from Group 1 on a test mentioned in number 1 and if this is related with
the number of years of training in degree programs in public administration
(bachelor's, masters, doctorate). This yielded the following result in a
correlation test:

r= .20432 a= .0001

Given this finding, would you still maintain your conclusion in number 1?

UP OPEN
UNIVERSITY
268 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

ASAQ 17-1

I.
The data indicate that clients' perception of the office is affected by the number of
forms they had to fill up. Seventy percent (70%) of the clients who said there were
too many forms to accomplish perceived the office to be inefficient as against
only 47% of the clients who remarked that they did not have to fill up too many
forms.

The data suggest that improvements can be mad'e in operations of the organiza­
tion to improve its image to the public. While th"e total of respondents who per­
ceived the office to have made them fill up too many forms is smaller than those
who said they filled them up (200 as against 320), it is still worrisome to have
70% of them say this.

However, other reasons for the perceived inefficiency should be determined among
47% of the respondents who said there were not too many forms to fill up.

See Table 1 to bear out the arguments made in this report.

Table 1. Perceptions of clients regarding the forms filled up in the office


and their opinion of the insurance agency

Too Many Forms

Opinion of the Office No % Yes %


Inefficient 150 47 140 70
Efficient 170 53 60 30
Total 320 100 200 100

II.
1. A chi-square test is applicable since these are two nominal or discrete vari­
ables. The independent variable here is the legislator's position and how each
could be distinguished in their personal satisfaction of their performance in
passing social development bills. Satisfaction with bills is not likely to influ­
ence the position of legislator. My dummy table is as follows:

Satisfaction with their advocacy of bills Legislator's position


Senator Congressman

Satisfied
Not Satisfied

UP OPEN
UNIVERSITY
DATA ANALYSIS/ Processing and Analysis 269

2. A difference of means test or a one-way ANOVA is applicable since one


variable is nominal (sex) and the other is continuous (advocacy).

Male -----IJII, x (or mean advocacy)


Female -----• x (or mean advocacy)

3. A difference of means test or a one-way ANOVA is applicable. Educational


attainment is nominal and advocacy is continuous.

With social science orientation - ---IJII, x (or mean :advocacy)


with this orientation -----... x (or mean advocacy)

4. The two variables here are continuous. A correlation test or linear regression
may be applied.

5. A Chi square test is applicable. Both variables are nominal. Gender is the likely
independent variable since males and females may distinguished for their edu­
cational attainment. Educational attainment is not likely to produce or affect
sex. See my dummy table.
------------.----------------
Educational Attainment Gender
-------------j---
Male----+-- -Female
- ---
With social science
Without social science

III.
1. The data indicate that there is a significant difference between the group
exposed to the program and the other which did not have the benefit of this
exposure. A higher performance was noted among those which had the benefit
of exposure to the training program. The edge of the participants indicates
that the program merits continuity.

2. There is a positive correlation between performance of administrators from


Group 1 and their length of exposure to the field of public administration.
Exposure to public administration could have imparted to the administrators
the value of equity. As a policy maker viewing the results of this study, I would
have misgivings about the impact of the training program. Participants could
have performed better because of their exposure to the field of public admin­
istration. However, if Group 2 has the same average exposure in the field of
public administration but did not perform as well as Group 1 participants,
then the program has made an impact on the participants. It is, therefore,
important to assess the background of Group 2 as well.

How d;d you fare? You should have been able to correctly answer more than
half of the ;tems. Rev;ew the sec6on where you d;d not fare well. UP OPEN
UNIVERSITY
270 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

ACTIVITY 17-2

You should have your proposal ready for implementation by this time. Incorporate
your plan for data analysis.

SUMMARY

This module stressed the responsibilities of the researcher as the research proposal is
implemented in the field. Directions on how to relate with the respondents or subjects
of the study and substantive requirements in dealing with methodological issues were
given.

Data gathered from the field must be processed in preparation for data analysis and
interpretation. The approaches to processing data discussed here include categoriza­
tion, coding, and application of appropriate statistical tools.

Some guidelines in data analysis and interpretation were discussed as a review of the
different statistical tools especially of the inferential type.

REFERENCES

Keller, Gerald, Brian Warrack and Henry Bartel


1988 Statistics for Management and Economics: A Systematic Approach. California: Wadsworth
Publishing Company.
Meir, Kenneth and Jeffrey Brudner
1997 Applied Statistics for Public Administration. 4th ed. New York: Harcourt Brace.
Parel, Cristina
1979 Data Analysis and Inte1pretation. Quezon City: Philippine Social Science Council.
Williams, Frederick
1986 Reasoning with Statistics. New York: Holt, Rinehart and Winston.

UP OPEN
UNIVERSITY
MODULE18

WRITING THE RESEARCH REPORT

INTRODUCTION

Writing the research report is a crucial phase in the research process. As stressed in
an earlier module, a research undertaking does not contribute to the fund of knowl­
edge if it is not shared with the public. Thus, it is important to pay attention to
how the research report is prepared and packaged so that our target audience will
comprehend the message we are imparting in a research undertaking. The grand­
est research design and the most exciting results are meaningless unless we are
able to impart them effectively.

In the preparation of the report, three important issues should be addressed by the
researcher: to whom is the research report directed ( the audience of the report);
what is the content of the report; and what is the style of the presentation of the
report. These will be discussed in this final module.

OBJECTIVES

At the end of this module, you will be able to:

Characterize the different types of audience of a research report;


Identify the standard components of a technical report; and
Apply the rules in the presentation of the technical report.

1.0 KNOWING THE AUDIENCE OF THE RESEARCH UNDERTAKING

One of the issues in the preparation of a research report is knowing who will serve
as its audience. This is important because the content and style of the report will
be affected by the audience we are targeting.
272 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Four categories of audience may be the target of a research report.

One group is the academic/research community, made up of teachers, researchers,


and students. They are interested in both the conceptual and practical implications of
the empirical bases of your study.

Normally, members of the academic/research community expect a comprehensive ac­


count of the processes implemented, including the problems encountered in the field
and how your methodology was affected by these problems; a detailed discussion of
findings and interpretations; and the theoretical/practical implications. These may be
presented in various forms such as a monograph, � book, or an article in a journal.

Journals have specifications about the number of pages they can accommodate so that
only the highlights of the methodology and the findings of the study have to be pre­
sented by the researcher. There could also be some specifications regarding the format
and bibliographic entries.

Students preparing their thesis or dissertation are also expected to give a detailed ac­
count of the processes implemented and the findings/analysis/interpretation of the
results of the study. Some schools indicate their own formats in the presentation of
thesis and dissertations and the students can be guided by these expectations.

A second type of audience of research reports are sponsors of research undertakings.


They normally indicate their expectations of the report to be submitted to them.Usu­
ally, the objectives of the sponsoring institution are spelled out and the format is fitted
to its Terms of Reference. They may range from one that includes only the highlights
of the findings to be presented and a sketchy discussion of the methodology of the
study, to a very detailed report.

Some sponsoring institutions may not even expect a review of literature to be included,
which is normally expected of an academic research paper.

Sponsoring institutions often expect an Executive Summary to be presented to make


it easy for top executives (who have limited time to read the entire report) to read the
results of the study. Nevertheless, as a researcher, you must prepare a comprehensive
technical report as a basis for writing a summary report. Sponsoring institutions usu­
ally expect that an oral presentation be made. This means preparing summary state­
ments that can be presented through an overhead projector, flipchart, or computer.

Research reports that are geared for policy makers require an even more sketchy pres­
entation of the methodology and findings of the study. The barest essentials are needed
since the main concern of the policy maker is to determine the stand of the researcher
and the basic data on which this argument is based. The details of the research process
could be provicfed as an annex to a policy argument. The policy studies program, a
sub-field of specialization in public administration, provides additional orientation
on how to present research outputs for effective advocacy. The production of advocacy
UP OPEN materials in the form of flyers or brochures is taught in colleges teaching mass com­
UNIVERSITY munication.
DATA ANALYSIS/ Writing the Research Report 273

The public, especially the beneficiaries of programs/projects/services, may also be the


audience of research results. In order to convince the users of the importance of some
methodologies or services being propagated, beneficiaries may be informed of the find­
ings of the study. There are also special skills and techniques for reaching various
beneficiaries with different levels of understanding and knowledge. Information edu­
cation communication (IEC) materials can be prepared in the form of brochures, fly­
ers, handouts, flipcharts, video presentation, etc. to convince the target beneficiaries.

In this module, we will be more concerned with the preparation of an academic report
since other fields of specialization in public administration and related fields can im­
part to you specialized requirements and techniques for their preferred stfle of presen­
tation. Also, whatever style of presenting reports you adopt later on, much depends on
the source material, the basis of the presentation. Thus, the technical report is an
important input to these different modes of packaging information.

2.0 CONTENT OF THE TECHNICAL REPORT

The technical report has the following parts:

1. T itle Page

This indicates the major variable of the study and the context of the study. For in­
stance, a doctoral dissertation prepared by Marilyn C. N oval-Gorra was entitled "Meas­
uring and Assessing the Quality, Equity and Efficiency of Public Hospitals in the Phil­
ippines." You will notice that the key variables in this research and the institutions
that serve as the focus of the study are spelled out.

2. Foreword

This may contain a statement regarding the research problem and objectives of the
research. This may also acknowledge the sponsoring institution, if there is one. This
may be prepared by the researcher who conducted the study and/or the institution
who sponsored the undertaking. Sometimes, the head of the sponsoring agency may
also give an endorsement of the study and may situate this in the context of the goals
and purposes of the research organization.

Some researchers also indicate their acknowledgments of the people who assisted them
in the conduct of the study, both substantively and administratively. This is also an
opportunity to express gratitude to the population/groups covered by the study and
the sponsoring institution who financed the research. There could be a separate head­
ing for acknowledgment if the foreword is already extensive.

3. Table of Contents

The Table of Contents provides a brief overview for the reader of the major topics
covered by the research. This contains the major headings, sub-headings, and page UP OPEN
numbers. UNIVERSITY
274 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

4. Introduction

The introduction discusses the general context under which a research problem is to
be studied. For instance, before saying why a program is being evaluated, indicate
what the program is all about and why there is a need to conduct an evaluation. Some­
times, the introduction may also cover a characterization of the organization imple­
menting the program, which enables the reader to situate what agency implements it
and what specific components ofthe program will be highlighted in the evaluation, its
objectives, processes, and methodology.

In a thesis or dissertation, a program and its org:anizational context may require a


separate section or chapter for discussion. This depends on the complexity of the cov­
erage of the study. An introductory statement can be made on why it is important to
focus on the program or project being evaluated or assessed in a particular research.

A study funded by a sponsoring institution may begin with the expectations and needs
of the funding institution.

5. Research Problem

This part focuses on the problem of the study, the specific variables, the need to focus
on the problem or the rationale and objectives of the study (both practical and theo­
retical), the statement of hypothesis, the theoretical or conceptual foundation of the
study showing the linkages and meaning of the relationship of the different variables
from the perspective of the researcher, a review of related literature (citing both em­
pirical and theoretical arguments made by other researchers), and operational defini­
tions or indicators of the variables focused on.

In a thesis or dissertation, some ofthese parts can appear as a separate section or chap­
ter, depending on the length or extensiveness of the discussion required to cover the
materials available.

6. Methodology

This covers the four major components ofresearch design, unit ofanalysis, techniques
of data collection, and statistical tools for processing information for quantitative re­
search.

The section on methodology in the research report discusses the processes actually
applied by the researcher and not what he/she had hoped to implement. In other words,
this can indicate if there are deviations made in the implementation of the study and
in what ways. For instance, you could have planned that stratified sampling will be
made of beneficiaries of an agricultural program, differentiating between small and
medium sized farmers. However, the absence of a sampling frame or list has made you
resort to random sampling ofhouseholds where these types of farmer are located (with
the aid of a spot map).
UP OPEN
UNIVERSITY
DATA ANALYSIS/ Writing the Research Report 275

The discussion of each component of the methodological part can indicate the prob­
lems in executing the plan.

This part should be able to give the reader an opportunity to determine the validity of
the arguments to be made in the findings of the study. The discussion of methodology
can also direct other readers to the sections they may wish to replicate in their own
investigations.

7. Findings

This section of the report discusses the data gathered from the data sources. It should
be able to respond to the questions raised at the beginning of the researcli.. Remember
that data should not only be described. Your role as a researcher is to give your analysis
and interpretation of the data at hand. Giving facts and figures from the data is not
enough. You have to indicate the conclusions or patterns you discern (analysis). Is
there a positive or an inverse relationship? Is the relationship borne out statistcially?
If the research is descriptive in character, what patterns did you observe?

You also have to indicate the theoretical or practical significance of the data you have
gathered. This is the interpretative part of your research. You can give a reason as to
why a positive or inverse relationship is noted in your data.

8. Summary, Conclusions and Recommendations

The summary of the study may cover the highlights of the research or the major points
raised. It provides a quick tour of the findings of the study. It may follow closely the
organization of the report.

The conclusions provide the interconnections of the findings of the study in a di­
gested form. This part actually refers back to the theoretical framework, indicating
whether the framework is workable. You may even show that the framework is not
borne out by your data. In this case, you contribute to theory building by providing
additional explanations for the failure of your initial arguments.

If the study is qualitative in nature and is implemented mainly to describe a given


event or process, the conclusive part of the study is made through the conceptual or
theoretical formulations derived. For instance, in a study by Rebullida (1991) on Catho­
lic and Protestant Churches' perspective of development, she concluded that:

The churches stressed the centrality of man as the agent and the beneficiary of
development, particularly the poor. They accepted the United Nations' notion
of quality oflife and basic services but added the element of social justice from
the very nature of social structures. They checked modernization and eco­
nomic models without social development, that is, sharing of goods reflected
in improved conditions of life. Though the churches used Marxist class and
UP OPEN
UNIVERSITY
276 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

social analysis, this was for the purpose of understanding objective social reali­
ties, but did not advocate violence in revolutionary change....

The churches' organizational and administrative approaches for implementa­


tion provide an alternative to the public administrative system of the govern­
ment. Traditional values of efficiency, effectiveness and economy were retained.
However, the churches emphasized values of responsiveness, representation,
participation, integration, and decentralization by regionalization (Rebullida
1991:92).

The sub-section on recommendations can indicate the practical contributions of your


study. If you have focused on applied research in public management (such as evalua­
tion, operations research, and policy research), you can provide specific directions to
policy makers, program managers, and implementors regarding the measures or ap­
proaches that can be adopted or improved to correct problems encountered in your
study.

At the theoretical level, you can also focus on the possible contributions of the study to
the field of public administration. What concepts or theories were borne out by your
data? What new explanations can be made to elaborate on a concept or theory?

Suggestions for Further Research is another topic that can be included. There could
be areas that have not been tackled in the research. These provide directions for fur­
ther investigation.

9. Appendices

These may include materials that are too cumbersome to read in the body of the paper
but are useful references for some readers. For instance, the instrument used for data
collection can be included. Significant policy statements that may be tedious to read
in the body of the paper can be incorporated here as a reference.

10. Bibliography

This covers the different materials (e.g., books, journals, monographs, public docu­
ments, etc.) used as references in the study. You may use a specific style you are com­
fortable with. Make sure, however, that this is the style is used consistently throughout
the report.

The aforementioned headings are some of the common topics which can be embodied
in a research report. Other topics can be included to suit the preferences or require­
ments of the researcher.

Some researchers may separate the text by chapters instead of having one continuous
UP OPEN document that is separated by headings or sections.
UNIVERSITY
DATA ANALYSIS/ Writing the Research Report 2 77

3.0 STYLE IN REPORTING

In this part, let me discuss with you some of the basic guidelines in the preparation of
a report.

1. Prepare an outline of the report to make sure that you are arranging information in
a logical order. In explanatory types of studies, you normally start with a descrip­
tive part, before going on to the inferential argument.

2. Write clearly and use simple sentences as much as possible.

3. You are not obliged to make a literary piece out of a technical report. A major
responsibility is conveying accurate information. You need not use pompous lan­
guage. Give adequate attention to the correct use of grammar and spelling.

4. Be consistent with your tense. You may use the present tense to discuss your find­
ings. However, if there are many past experiences that have to be referred to, then
shift to the past tense.

5. Be consistent with the style of your headings and sub-headings and the font you
are using. The style or font of headings signal shifts to different parts in the report.
Make your reader "see the rhythm" of your report.

6. Tables and charts must be numbered. They are referred to after the narrative state­
ment and not before. You have to introduce the table before letting your reader
view it. The tables and charts supplement your discussion.

7. You need not discuss the statistical process implemented to obtain a statistical
result in the body of your paper. You can present the statistical results in the fol­
lowing manner:

Indicate a chi-square test result below the table, including the significance
level. Make sure to apply percentage or proportions analysis for a bivariate
table. The chi-square test may be presented like the one shown in Table 18-1.
An example of an analysis of the data is given below.

An important finding in this research is the prominent role of the father in


decision making pertinent to health matters. Based on the data reflected in
the table, mothers are not the sole decision makers on health matters. Moth­
ers as sole decision makers in each of the four decision areas occur only in
34.34 percent or less of the total number of respondents (1,992). Further­
more, there is a progressive decrease in the unitary role of the mother in
decision making when more complex decision making is made. For exam­
ple, mothers emerge as sole decision makers in 34.34 percent of the total
respondents who rendered emergency treatment at home to a sick family
member. However, the role of the mothers as sole decision makers is di- UP OPEN
UNIVERSITY
278 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

minished in some areas of responsibility. These areas include who to con­


sult when a family member is ill and can no longer be managed at home
(28.26%); how much money to spend on health matters (24.4%) and when
to hospitalize a sick family member (21.59%).

Joint decision making figures prominently in who to consult when a fam­


ily member is ill (32.93%) and when to hospitalize a member of the family
who is ill (39.26%).

The husband figures as the most prominent person in making a decision


regarding how much to spend for health matters (37.7%) perhaps because
he is the major breadwinner in the family.

There is a statistical difference between the persons involved in each deci­


sion making area with a chi-square test (149.84, d.f. = 12, alpha = <.01).
The findings here coincide with an earlier saying that a "thinking task"
like determining who to consult when a family member is sick and when
to hospitalize tends to be shared, more than the "doing task" such as giv­
ing an emergency treatment at home (Contado 1991: 159).

Take note that the chi-square test is reflected below the table.

Table 18-1. Decision areas of respondent mothers and their spouses


on health matters

Decision maker(s) Decision Areas

Emergency Who to Amount When to


treatment Consult to spend Hospitalize

Respondent 34.34 28.26 24.40 21.59


Spouse 29.82 29.92 37.70 30.77
Both 29.57 32.93 31.78 39.26
Both and Others 2.61 3.93 2.61 4.22
Others Only 3.61 3.21 3.51 4.07
No Answer .05 .10

Total 100.00 100.00 100.00 100.00


n 1992 1992 1992 1992

Chi square test= 149.84


d.f. = 12 Alpha < .01

Source: Bautista 1992


UP OPEN
UNIVERSITY
DATA ANALYSIS/ Writing the Research Report 2 79

For a correlation test, show the:

variables being related


correlation coefficient
significance level

See Table 18-2 for a sample of how correlation tests may be presented. The analysis of
the results in this table is given below.

The table summarizes the findings on tests of factors related to availment of


services for children. Three factors, except family income, are retated with
overall utilization per household. Positive correlations are noted between uti­
lization of services for children per household and such characteristics as the
educational attainment of mothers and fathers and the household size. The
fact that income is not a significant factor for availment augurs well for the
program because this indicates that the economically deprived are not dis­
criminated by program implementors.

Table 18-2. Summary of correlation tests for selected sociodemographic


characteristics and utilization of health services for children

Factors related with Correlation Alpha level Result:


availment of services Level Significant (S)
for children or Not Significant (NS)

Mother's education .18 .0001 s


Father's Education .15 .0001 s
Household size .05 .01 s
Family income .01 .73 NS

Source: Bautista 1992

For a difference of means test, indicate the:

groups in a variable being compared for their mean


mean for a variable in each group
zit test result
significance level

An example of a difference of means test is shown in Table 18-3. The table focuses on
mean level of schooling among those who participated/did not participate in commu­
nity activities in the Area-Based Child Survival and Development Program, compar­
ing experimental (ExB) and control barangays (CB). The analysis of mean level of
schooling is given before the table.

UP OPEN
UNIVERSITY
280 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

Mothers who are participative have higher educational attainment (8.3 years
of schooling) and have husbands with a higher educational attainment (8.1
years of schooling) than those who are nonparticipative (5.6 years of schooling
for the respondents and 5.8 for the fathers).

Between the experimental and the control barangays, the participative moth­
ers from the former have a lower educational attainment than the latter (7.05
v.s. 9.48 years of schooling).

The influence of education on participation rciiy be attributed to the fact that


those with higher educational attainment appreciate better the value of com­
munity participation. Furthermore, those with higher educational attainment
have the facility to articulate their views and demands. The father's educa­
tional attainment is related to the mother's involvement since the fathers may
have encouraged the mother's decision on whether or not to get involved in
community activities.

Table 18-3. Mean level of schooling among those who participated/


did not participate in community activities, by ExB and CB areas

Areas n Mean T test Alpha SINS

Mean years of schooling of respondents

With participation
Experimental Bgys. 81 7.05 -4.07 .0001 s
Control Barangays 110 9.48 w/ unequal var.

Without participation
Experimental Bgys. 972 5.14 -4.6
Control Barangays 828 6.17 w/ unequal var.

Source: Bautista 1992

For a regression test, indicate:

regression coefficient
t test for each independent variable
significance level of the t test for each variable
F test of the model and the significance level of the F test.

An example of how the results of a study are presented is shown in Table 18-4. The
table reflects the different factors related to the reduction in illness from the onset of
the program (1993) called Partnership for Community Health Development (PCHD)
UP OPEN to the period when the assessment was made (1996). The analysis of the findings are
UNIVERSITY presented here.
DATA ANALYSIS/ Writing the Research Report 281

One of the arguments in this research is the influence of PCHD on the reduc­
tion of illnesses in the locality. PCHD intervention is indicated by the extent
in level of community participation. This pattern is posited because of the
possible influence of the community to initiate activities that will redound to
health and related concerns. Thus, the net effect is the improvement in the
health condition of the population.
However, apart from community participation, other variables were also ex­
amined.
The table shows that the basic argument in this research is borne out. A posi­
tive relationship is established between level of participation and -improve­
ment of the health condition of the population. The latter is measured based
on the number of illnesses reduced from pre- to post-PCHD. This is highliy
significant at alpha of .004.
Furthermore, a positive relationship also exists between observable improve­
ments in the performance of service providers and the reduction of illnesses.
Thus, both the performance of the service providers and the initiative of the
community to manage their own health needs augur well in improving the
health condition of the community This pattern is significant since participa­
tion alone may not be sufficient as the community members do not have enough
knowledge to respond to their health needs. On the other hand, participation
enhances the performance of service providers since there are activities that
can be facilitated through their support, such as labor and financial resources.

Table 18-4. Factors related to reduction in illness from pre- to post-PCHD


in a regression test

Independent variables Regression t value Significance level


coefficient

Constant .39
Performance of Service .49 3.1 .002 s
Providers
Increase in health facilities .08 .46 .65 NS
from pre- to post-PCHD
Total time for travel .001 1.56 .12 NS
Extent of participation .04 3.03 .004 s
Number of orgts mobilized .01 .175 .86 NS

Source: Bautista 1996.

UP OPEN
UNIVERSITY
282 PM 299.2 RESEARCH AND PUBLIC MANAGEMENT

SUMMARY

On the whole, this module provides tips to researchers regarding the preparation of
the research report. Three basic aspects of preparing the research report were dis­
cussed.

One pertains to knowing the audience of the investigation. The audience targeted may
be the academic/research community, sponsors of the research, policy makers, and
beneficiaries. The basic rule is to suit the language of the report to the target audience
of the investigator. A common denominator of various materials for groups with dif­
fering levels of knowledge and understanding is a technical report, which is normally
expected by an academic/research community. This details the processes applied in
gathering data and the analysis and interpretation of the information gathered from
the field.

The second aspect has to do with the content of the report. The minimum require­
ments for a report are: a title page, foreword, table of contents, introduction, statement
of the research problem, methodology, findings, and summary/conclusions/recommen­
dations, appendices and bibliography.

The third sub-topic in this module was on the style in presenting the report. Some
basic guidelines were suggested, with focus on how data can be presented in a report.

ACTIVITY 18-1

You should implement your proposal and be ready to prepare your research report.

This is where we end this module. Good luck!

REFERENCES

Bautista, Victoria.
1996 Partnership for Community Health Development Appraisal Report. Quezon City: College
of Public Administration, University of the Philippines.
1992 Towards the Institutionalization of ABCSDP: Highlights of the Feasibility Assessment.
Philippine Journal of Public Administration XXXVI, 3 (July).
Rebullida, Ma. Lourdes Genato.
1991 Church Development Perspective: Policy Formulation and Implementation. Philippine
Journal of Public Administration XXXV 1 (January).
UP OPEN
UNIVERSITY

Das könnte Ihnen auch gefallen