Sie sind auf Seite 1von 17

This article was downloaded by: [UNESCO IIEP] On: 28 February 2012, At: 06:50 Publisher: Routledge Informa

Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Compare: A Journal of Comparative and International Education


Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ccom20

The relationship between teacher gender and student achievement: evidence from five Indian states
Amita Chudgar & Vyjayanthi Sankar
a a b

College of Education, Michigan State University, East Lansing, USA


b

Educational Initiatives, Ahmedabad, India

Available online: 10 Oct 2008

To cite this article: Amita Chudgar & Vyjayanthi Sankar (2008): The relationship between teacher gender and student achievement: evidence from five Indian states, Compare: A Journal of Comparative and International Education, 38:5, 627-642 To link to this article: http://dx.doi.org/10.1080/03057920802351465

PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-andconditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Compare Vol. 38, No. 5, October 2008, 627642

The relationship between teacher gender and student achievement: evidence from five Indian states
Amita Chudgara* and Vyjayanthi Sankarb
a College of Education, Michigan State University, East Lansing, USA; bEducational Initiatives, Ahmedabad, India

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

Little conclusive evidence is available on the relationship between teacher gender and student achievement. However, a great emphasis has been placed on hiring more female teachers, both internationally and in India. Given this context, this paper investigates the relationship between student learning outcomes and the presence of women teachers in Indian classrooms. Our analysis shows that male and female teachers differ in terms of their classroom management practices and their belief in students learning ability. In partial support of the policy of hiring more female teachers, it also shows that being in a female teachers classroom is advantageous for language learning but teacher gender has no effect on mathematics learning. Keywords: teacher gender; student achievement; India

A recent UNESCO advocacy brief (UNESCO 2006) argues in favour of hiring more female teachers in developing countries, and states that doing so will have two positive outcomes. Their presence will lead to an improvement in both girls enrolment and girls learning achievement. Other researchers have also suggested a beneficial relationship between teacher gender and enrolment and achievement, especially for girls (UNESCO 2000). However, as shown in the following sections, the evidence on the relationship between student learning and teacher gender is limited and contradictory. Given this context, this paper investigates the relationship between the student learning outcomes and the presence of women teachers in 4th and 6th grade classrooms in India. In addressing this relationship the paper also investigates and includes the male female differences in teaching practices which may mediate the relationship between teacher gender and student performance. The aim of the paper is to examine the following research questions: (1) Based on the data from five Indian states, how are male and female teachers similar or different in terms of their qualifications, teacher training, experience, and teaching practices? The differences in teacher background are descriptive in nature, but we use additional statistical analysis to ascertain further the differences in their teaching practices after controlling for their background. (2) Do Indian children perform differently in classrooms headed by male and female teachers? Do the data support the claim that girls tend to perform better in classrooms run by female teachers?
*Corresponding author. Email: amitac@msu.edu
ISSN 0305-7925 print/ISSN 1469-3623 online # 2008 British Association for International and Comparative Education DOI: 10.1080/03057920802351465 http://www.informaworld.com

628

A. Chudgar and V. Sankar

(3) If differences do exist in student performance in classrooms headed by Indian male and female teachers, can some of these differences be explained by male female differences in teaching practices? Conflicting evidence on the relationship between teacher gender and learning The literature on the relationship between teacher gender and student outcomes offers almost every possible conclusion. Thomas Dee (2006) investigated the effect of teachers gender using National Education Longitudinal Survey (NELS) data on 8th graders from the US and found that same-gender teachers had a positive effect, i.e. girls do better in school when taught by women and boys do better when taught by men. Dee also found that the effect of teacher gender varies depending on the subject; for girls the benefits of being assigned to a female teacher are concentrated in history. A study by Katharina Michaelowa using data from Francophone sub-Saharan African similarly finds support for the same-gender effect (Michaelowa 2001). Contradicting these studies, however, a large sample-based study in the US shows that regardless of student gender, students taught by women perform better than those taught by men (Krieg 2005). In accordance with Krieg, based on findings from the Southern and Eastern African Consortium for Monitoring Education Quality (SACMEQ), a recent UNESCO Education for All report notes that women teachers have a positive effect on both boys and girls achievement (UNESCO 2005). Similarly, a study from Pakistan (UNESCO 2000) finds that children in female teachers classrooms tend to perform better. But a large study in Pakistan presents findings that contradict the studies above. Warwick and Haroona (1994) studied 1000 teachers, 300 school supervisors, and 11,000 4th and 5th grade students in Pakistan. He found that rural students of male teachers scored significantly higher in math than did rural students of female teachers. However, the author notes that it is unclear whether the differences arose from teacher, student, school, or cultural factors. Further complicating this picture, other researchers have found no relationship between teachers gender and student outcomes. In the Netherlands, for instance, Geert Driessen (2007) found that teacher gender has no effect on student achievement, attitudes, or behaviour, regardless of student gender, ethnic background, or socioeconomic status. Thus the evidence that increasing the presence of female teachers will improve girls learning outcomes is at best limited. With respect to the positive relationship between the presence of female teachers and improved school participation for girls the argument is that the presence of a female teacher may help alleviate parental concerns about the safety and well-being of their daughters in traditional, gender-segregated societies and encourage them to send their daughters to school (Velkoff 1998; UNESCO 2000). Here the picture is less contradictory and the value of female teachers is more easily apparent. But on the question of student learning, what is the link between the presence of female teachers and differences in learning outcomes? One hypothesis could be that differences in the teaching practices of male and female teachers may help to explain the differential performance of their students. Differences between male and female teachers teaching practices Some researchers have found evidence that male and female teachers have different teaching styles. Research on gender differences in teaching practice has been primarily qualitative. While reviewing this literature it is also important to keep in mind that these

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

Compare

629

studies are often based on university-level faculty members and a lot of these studies are conducted in the developed world. This research shows that male and female teachers do have different styles in terms of the time they spend on lecturing and classroom activities (Laird, Garver, and Niskode 2007). Female faculty members tend to use a more student oriented style of teaching that emphasises relating to the student (Grasha 1994). Compared with their male counterparts, women faculty also tend to have more liberal views about letting the students define and identify their own learning experiences and learning styles (Lacey, Saleh, and Gorman 1998). Women teachers are also more likely to involve their students in peer collaboration than in personalised activities in class (Chen 2000). Singer (1996) argued that female faculty members are more likely than males to utilise motivation or process paradigms. Moreover, women faculty members are more likely to spend time planning and designing and assessing learning activities (Singer 1996). More specifically in the developing country context, a UNESCO-sponsored study of four countries (Bangladesh, Nepal, India and Pakistan) on female teachers in rural primary schools found that female teachers were perceived as more effective in earlier grades by the administrators; children also saw them as more open and more comfortable to interact with than male teachers (UNESCO 2000). The data for this study were collected mainly by focus group discussions and interviews with teachers, principals, students, community members and administrators and policy makers. While reviewing these findings it is essential to keep in mind that gender dynamics of the classroom are difficult to assess using surveys, interviews and discussions as they are often unable to take into account the broader context within which the interactions between students and teachers take place. In a traditional rural setting for instance, where teaching itself accords a certain authority to the teacher regardless of their gender, a male and female teacher may look very similar inside the classroom in their teaching practices. In a detailed ethnographic study of rural schooling in India, Padma M. Sarangapani provides an excellent analysis of teaching and learning in the rural classrooms. Her analysis shows no differences between male and female teachers in their exercise of authority and their use of teaching practices that ignored, for instance, the students contributions completely. In fact, in her study the female teachers came from towns further away and they perceived themselves as more urbane and socially superior to the children compared with the male teachers who were local. She thus remarks that, contrary to the myth that children are happier with women teachers, the children were more free with the two male teachers from their village (Sarangapani 2003, 136). This research provides an important illustration of the dangers of viewing gender as a onedimensional variable devoid of the social contexts. The Indian context: teaching and learning achievement Until recently, teaching in India has been a male-dominated profession. In the mid-1970s women made up slightly more than 25% of the primary teacher labour force (Seetharamu 2002). Since 1987 the Operation Blackboard scheme has focused on the goal of hiring a minimum of 50% female teachers. This emphasis was also underscored in the 1992 programme of action accompanying the revision of the National Education Policy (Stacki 2002). By the late 1990s about 35% of teachers at the primary level were female (Seetharamu 2002). Indias most recent and ambitious education programme, Sarva Siksha Abhiyan (SSA) (loosely translated to imply education for all) continues to list the goal of 50% female teachers among its numerous targets. Recently this number had crept

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

630

A. Chudgar and V. Sankar

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

up to 43% but it is still not at the 50% mark (Department of Elementary Education and Literacy and Department of Secondary and Higher Education 2004). It is also important to note that there is an understanding in the literature that over time the proportion of female teachers in urban areas has generally been greater than the proportion of female teachers in the rural areas (Mehrotra 2006). Given the significant policy push to hire more female teachers, it is valuable to investigate the learning achievement of students in male and female teacher classrooms. In India, however, the issue of student achievement has been less explored than that of school access, which is not a surprise given the pressing need to increase school enrolment to meet the Education for All goals. An extensive review of learning achievement research in India by Sujata Reddy notes that, until the 1990s, the focus on learning achievement was limited both in the policy arena and in research (Reddy 2004).1 The seminal work by Stephen P. Heyneman and William Loxley (1983) showed that in low-income countries school resources and teacher quality matter more than family background variables a finding that has often been referred to as the H-L effect. More recently, David Backer et al. (Baker, Goesling, and Letendre 2002) questioned these findings, as they found evidence for a vanishing H-L effect. That is, they did not find that school variables had a particularly strong influence on student achievement in low-income countries. In this regard, for the Indian context, Reddy notes after an extensive review of achievement research in India that the home background and school factors exert equally important influences on childrens cognitive achievement (where school factors include teacher attributes as well) (Reddy 2004, 29). Specifically focusing on teachers, Reddys review notes that studies have found a positive relationship between teacher qualification and student achievement, but not between student learning and teacher experience. One study quoted in the report found a negative correlation between teacher experience and student learning. Reddy notes that research has also found that teaching practices such as giving more homework and providing more feedback have a positive relationship with student learning but found no consistent effect for class size. It is important to note that while the extensive literature review by Reddy (2004) may perhaps provide the most authoritative and comprehensive summation of the major findings in achievement research in India, it does not elaborate on the methods employed by most of the studies cited in the report. Also, from the perspective of the present study her literature review does not speak to the relationship between teacher gender and student achievement.

The study The study is based on secondary data collected by Educational Initiatives (EI) from January to March 2007 as a part of their benchmarking study to assess student learning.2 EI is a research-based assessment organisation in India that works extensively with government agencies, national and international non-government organisations and academic institutes to understand student learning through assessment products and services developed in English as well as the other vernacular languages. In order to collect these data for the benchmarking study, 300 schools were sampled from the five most populous towns in five states (Andhra Pradesh, Gujarat, Uttarakhand, Rajasthan, Chhattisgarh).3 Both public and private schools were surveyed with a primary emphasis on public schools. The survey focused on students and teachers from Grades 2, 4 and 6 (age 7 to 11 years). In this paper we limited the sample to public school children in Grades 4 and 6. For each school surveyed, information was collected on the school itself, on the

Compare

631

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

principal and on school teachers. Students were given an achievement test and were also asked to fill in a small background questionnaire. For the achievement tests, EI developed test papers in three languages Telugu, Hindi and Gujarati, which were the regional languages of the states that participated in the study. The test blueprint and the competency framework were developed after doing a detailed analysis of textbooks in all the states and with inputs from subject experts. Substantially more test items (three times more) were developed and pre-tested in all the three languages for ensuring that an adequate number of items were available for selection for the main tests, While the paper for mathematics was developed in English and appropriately translated, for language an outline of the test papers was developed initially in English. Based on this outline, language teams developed the papers in Hindi, Gujarati and Telugu. Subsequently the Hindi version was treated as the master and iterative changes were made in all papers to make them as comparable as possible. A number of critical issues related both to translation validity and appropriateness of certain questions came up at the time of translation and hence the development of papers in multiple regional languages was an iterative process. This collaborative effort ensured that the original meaning of the question, the reading level of the text, the difficulty level of the item did not change and the likelihood of another possible correct answer for the test item did not arise. This process also ensured that cultural adaptations were documented and the test items were acceptable to all of the dialects of the language in which the test was administered. There were many skills that could not be purely tested through a written test. Three forms of test administration were therefore used written, group oral and individual oral. In a written test, students had to read the question paper and write their answers. In a group oral test, the questions were read out to the whole group by trained administrators, and the answers had to be written individually by students in the written test. In an individual oral test, each child was tested individually. The achievement data used in the data sample here are based on the student scores in the written and the group oral forms of the test.

Variables At the teacher level, the data provided information on teacher gender, age, and a host of teacher background variables including teacher experience, teacher qualification, teacher training, and in-service training (Appendix 1). In addition, the survey also asked the teachers their opinion on questions that revealed their classroom practices. This study focuses on four questions. These questions address the beliefs about need for strict discipline, importance of fear in maintaining discipline, the importance of maintaining a serious (non-humorous) classroom environment and the belief about childrens learning ability (see Appendix 1 for more details on the variables, including the coding scheme). The first three questions can be identified as questions more related to classroom management practices and the last question refers to the teachers view on the students learning ability. In order to accord greater weight to classroom management practice in the final analysis, we also constructed an index of classroom management practice with a summation of the teachers attitudes on the three classroom management variables. Teachers could receive scores ranging from 0 to 3.

632

A. Chudgar and V. Sankar

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

At the student level, separate analyses were conducted for student scores in mathematics and language. Information was also available on student gender, age, grade in school (4th or 6th), and a measure of socio-economic status (SES). This measure of SES was created by summing the students responses on three questions about the availability of certain facilities at home. The questions asked students about the type of vehicle at home (cycle, scooter, none, or both), cooking method used at home (firewood, kerosene, or cooking gas), and ownership of radio and television (both or neither). A student who had access to none of the facilities listed would receive the minimum score of zero; one who had access to all of them would receive the maximum score of nine. Each student was linked to a teacher based on the combination of school identification and subject and grade taught by the teacher. Thus this analysis used a sub-sample of the teacher data for which student test scores and background data were also available. Since all the analysis was conducted separately for mathematics and language tests, two separate datasets were created for language and mathematics. In schools where two teachers taught the same subject, grade combination made it impossible to link students to teacher; therefore those data were excluded from the analysis. Thus the final sample analysed here reflects a situation where for a given grade and a given subject each school had no more than one classroom in the analysis. Data analysis The study employs primarily two regression techniques: binomial logit regression and multiple linear regression. In both the cases to account for clustering of students within a school a cluster correction command was used in STATA. Using cluster correction ensures that the standard errors for the coefficients estimated are adjusted for the homogeneity or clustering of students population within any given schools. Failing to use cluster correction may lead to finding significant relationships when actually there may not be a significant association between the variables. Given the dichotomous nature of teaching practice variables, in order to examine the relationship between teacher gender and teaching practices the binomial logit regression method was used. For each of the four teaching practice variables listed above, four separate models were estimated.4 After estimating the regression models in all of the above cases we used the STATA command prvalue developed by Xu and Scott Long (2005) to calculate the predicted value of a given teaching practice by teacher gender. Similarly predicted values for male and female teachers were also calculated by specific subgroups of teachers based on teacher experience, training and so on. The confidence interval estimation offered by the prvalue makes it possible to comment on the statistical differences between male and female teachers in terms of these teaching practices in the overall data and by teacher subgroups. For the student test scores on mathematics and language, two separate linear regression models were estimated controlling for student and teacher level variable. An additional interaction term was added to identify when the student and the teacher were the same gender by interacting the student gender variable with the teacher gender variable.5 In the first step, the regression equations did not include any of the teaching practice variables; these variables were introduced one at a time in the second stage of analysis. Finally, these regression equations were also analysed separately for various subgroups of teachers to verify certain findings by teacher subgroups.

Compare Results Comparing male and female teachers

633

The first section of Table 1 lists the sample size and the proportion of male and female teachers for each of the five states. The second and third sections of Table 1 list the proportion of male and female teachers for each teacher background variables and for each of the attitudes to teaching practice.6 It is important to note that the majority of the sample were female teachers (72%), and that the mean age of male and female teachers was 42.3 years and 42.7 years respectively. While this does not reflect the overall composition of the teacher labour force in India, the data reflect findings from elsewhere that, in urban areas and larger towns, the teacher labour force has a greater proportion of women than overall in the country (Mehrotra 2006). Compared with male teachers, relatively more female teachers have lower educational qualifications (38% of female teachers with less than Bachelors education compared with

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

Table 1. Teacher background variables and malefemale teacher comparisons (numbers in percentages). Variable Name 1. State Andhra Pradesh (N5416) Gujarat (N5301) Uttarakhand (N5174) Rajasthan (N5216) Chhattisgarh (N5201) 2. Teacher background (Complete Data) Teacher qualification Less than Bachelors education Bachelors education Masters education Teacher training No teacher training Certificate/diploma Degree in Education In-service training Never attended More than one year ago Within last 12 months Teaching experience Less than 10 years More than 10 years 3. Teaching practices Do you think strict discipline is necessary for proper teaching? Do you think there can be no discipline without fear of the teacher in students? Do you think the teacher loses respect if she/ he plays and cracks jokes with the students? Do you think all children are capable of learning? Total number Female Teacher % 62.5 82.0 75.2 78.7 66.2 37.9 30.5 31.5 4.6 57.8 37.5 4.2 9.1 86.6 37.1 62.9 62.5 43.8 15.6 55.9 949 Male Teacher % 37.5 17.9 24.7 21.3 33.8 28.1 35.1 36.7 5.9 48.9 45.1 4.8 12.7 82.4 46.3 53.7 75.1 55.1 23.8 45.1 370 Total % 100 100 100 100 100 35.1 31.8 32.9 5.0 55.3 39.6 4.4 10.1 85.4 39.7 60.3 66.0 47.0 17.9 52.9 1,319

Note: Differences between male and female teachers are statistically significant for all variables except teacher age.

634

A. Chudgar and V. Sankar

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

28% of male teachers at that level), but overall a slightly larger proportion of women have received teacher training (4.6% of female teachers are without teacher training compared with 5.9% of male teachers). Of all those who have received teacher training, relatively fewer women have obtained a degree in education (37.5% of female teachers compared with 45.1% of male teachers). The data also reveal that relatively more female teachers have received in-service training in the recent past (86.6% of female teachers compared with 82.4% of male teachers), and a larger proportion of women teachers are more experienced than male teachers (62.9% of female teachers have more than 10 years of experience compared with 53.7% of male teachers). Section 3 of Table 1 lists the overall proportion of male and female teachers who exhibit various views on classroom management practices and student learning ability. On the whole, the numbers here reveal that male teachers seem to focus more on maintaining classroom authority. Statistically, a larger proportion of men than women emphasise the need for strict discipline (75.1% compared with 62.5%), the importance of fear in maintaining classroom discipline (55.1% compared with 43.8%), and the importance of the teacher maintaining a serious tone in the classroom (that is, the inappropriateness of joking with students) (23.8% compared with 15.6%). Compared with male teachers, a relatively larger proportion of female teachers agree with the view that all children are equally capable of learning (55.9% compared with 45.1%).7 To address the concern about social construction of gender relations, we conducted the same analysis separately by state. The sample size issues lead to fewer statistically significant findings when looking at one state a time. However, as noted in Table 2.1 in Appendix 2, in the instances when male female differences are significant, the patterns observed in the overall data continue to hold. Especially interesting is the significant differences in malefemale attitudes towards the importance of fear in maintaining classroom discipline across three out of five states.

Table 2. Results of binomial logit regressions for teaching practices (presented as odds ratio), N51,319. Variable Name Female teachers Teacher age Teacher qualification2 Bachelors education Masters education Teacher training3 No teacher training Certificate/diploma In-service training4 Never attended More than one year ago Less than 10 years teaching experience5
Note:
***

Model 1
1

Model 2 .70 1.00


**

Model 3 .64 1.01+


**

Model 4 1.39* 1.00 .81 1.45+ 1.01 2.2*** 2.13+ .73 .39***

.62 1.01+

**

3.07*** 2.15*** 1.05 .42*** .26*** .71 2.97***

1.86*** 1.46+ .77 .69* .97 1.34 1.50***

2.58*** 2.38*** .62 1.26 1.38 .89 1.59*

5p#.001, **5p#.005, *5p#.01, +5p#.05 15male teachers; 25less than bachelors education; 35Degree in Education; 45Attended in-service training within last 12 months; 55more than 10 years of teaching experience Model 1: Do you think strict discipline is necessary for proper teaching Model 2: Do you think there can be no discipline without fear of the teacher in students Model 3: Do you think the teacher looses respects if she/he plays and cracks jokes with the students Model 4: Do you think all children are capable of learning

Compare

635

However, all these proportions were generated without controlling for relevant teacher background characteristics. Columns 1 to 4 in Table 2 present the results of a binomial logit analysis for the four teacher practice variables listed in Table 1. Teacher gender and the teacher background variables listed in Table 1 are used as controls to understand the relationship between teacher gender and teaching practices. To ease interpretation, the coefficients for each of the four binomial logit models are reported as odds ratios. An odds ratio greater than 1 indicates a positive relationship with the outcome variable or in this case the relevant teacher practice variable. An odds ratio of less than 1 indicates a negative relationship with the outcome variable. The primary variable of interest here is teacher gender. First, we note that the coefficient on the female teacher variable is significant in all four models. This significance implies that malefemale differences continue to persist after accounting for teacher background variables. More specifically, the coefficients of less than 1 for models 1 to 3 indicate that after controlling for teacher background variables, we still find that female teachers are less likely to emphasise the need for strict discipline in classroom management, to indicate that fear is important to maintain classroom discipline, and to say that joking with students in class is inappropriate. In Model 4, however, we see an odds ratio greater than 1. This indicates that, controlling for teacher background, female teachers are more likely than male teachers to say that all children are capable of learning. In summary, the results displayed in Table 2 show that, controlling for teacher background, the malefemale differences in these Indian classrooms persist. In other words, the background of the male and female teachers listed here cannot, by itself, explain these differences in classroom management practice and differential beliefs in student learning ability. To determine the extent to which male and female teachers differed on these four variables, we also generated the predicted probability of a given response for a male and for a female teacher.8 The results of the predicted probability test showed that the widest differences between male and female teachers are in their views on the need for strict discipline and the importance of fear. Compared with their male counterparts, female teachers are respectively 10 and 9% less likely to hold these views. They differ by 7% in their views on the appropriateness of joking in class. And female teachers are 8% more likely than male teachers to state that all children are equally capable of learning. Finally, given these systematic gender differences in teaching practices, it is pertinent to ask how these malefemale differences play out if we look at specific subgroups of teachers, such as those with a graduate degree, or those with a Bachelors or Masters degree in education. In other words, are there any interactions between teacher training and teacher gender that may in turn be related to the teachers opinions on the teaching practice questions from the survey? Once again, we find that for any subgroup of teachers (based on experience, pre-service and in-service training, and qualifications) the male female differences continue to persist. (Results available in Appendix 2, Table 2.2. Results are expressed as probability difference between female and male teachers.) To conclude, the findings from this section show that the Indian male and female teachers in the present sample differ systematically in terms of both their background and the specific teaching practice variables on which survey data are available. Relationship between teacher gender and learning outcome In this section, we focus on the relationship between teacher gender and student performance, which is often presented as an argument in favour of hiring more female teachers to promote greater achievement among girls. Specifically, in this section we focus on the following questions. Is teacher gender significantly related to achievement scores in

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

636

A. Chudgar and V. Sankar

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

language and mathematics? Do girl students tend to benefit especially from being in classrooms with female teachers? If students in male-led and female-led classrooms perform differently, to what extent can these differences be explained by explicitly including the teaching practice variables investigated in the section above? As noted earlier, the analysis in this section uses a sub-sample of the teacher data for which student test scores on language and mathematics and background data were also available. In terms of descriptive analysis, these teachers are similar to the larger sample of teachers analysed for the previous question. Table 3 below lists the student descriptive data for the sample used for mathematics and language, including the student sample sizes from each state. Table 4, column 1, presents the results of a multiple linear regression on student achievement in language, controlling for student and teacher backgrounds. Column 2 presents similar results for mathematics achievement.9 To focus first on column 1, the results in the Table show that controlling for student background and teacher characteristics, being in a female-teacher classroom is associated positively and significantly with test scores on language achievement. The Table also shows that, as expected, student SES is positively associated with language score, i.e. students from higher SES families tend to perform better in these tests. Additionally, students whose teachers have less than 10 years of experience are likely to have a higher language achievement score compared with those with teachers who have more than 10 years of experience, a finding that accords with earlier research cited in Reddy (2004).10 Finally, students whose teachers have had in-service training in the previous 12 months tend to perform significantly better than students whose teachers have had no such training. Focusing on the same-gender benefits of being in a female-headed classroom, first, Table 4 shows no statistically significant differences between the overall language scores of boys and girls as such. Second, the interaction term that identifies a female student in a female teacher classroom is not statistically significant. In other words, while children in female-headed classrooms performed better in language overall, girls did not show a significant benefit when compared with boys. The results for mathematics are different from those for language on most counts, except for SES and same-gender relationship. First, SES is positively related to student learning in mathematics. For mathematics achievement, the coefficient for a female teacher is positive but it is not significant. However, the negative coefficient on female students shows that in terms of mathematics achievement, girls are generally at a disadvantage compared with boys.
Table 3. Student descriptive statistics for language and mathematics, means and (standard deviations). Variable name Mean achievement score Female students Average age SES (range 09) State (student sample size) Andhra Pradesh Gujarat Uttarakhand Rajasthan Chhattisgarh Total number Language 50.3 (21.8) .54 10.8 (1.68) 4.7 (2.4) 1,236 1,066 519 352 868 4,041 Mathematics 48.0 (19.18) .55 10.7 (1.64) 4.7 (2.4) 1,322 1,132 583 454 817 4,308

Compare

637

Table 4. Results of multiple linear regression on language and mathematics with student and teacher level controls. Variable Name 1. Student variables Female student Student age Student grade SES Fem student* Fem teacher 2. Teacher variables Female teachers1 Teacher age Teacher qualification2 Bachelors education Masters education Teacher training3 No teacher training Certificate/diploma In-service training4 Never attended More than one year ago Less than 10 years teaching experience5 Total numbers
Note:
***

Language .11 .46 1.84 1.43*** 1.07 5.31** .03 2.45 22.8 2.0 1.6 28.0+ 1.19 5.21+ 4,041

Math 23.47** .65 24.54*** .63*** .78 3.15 2.01 1.88 3.05 3.12 2.93 .62 21.36 .81 4,308

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

5p#.001, **5 p#.005, *5 p#.01, +5 p#.05. 15male teachers; 25less than Bachelors education; 35Degree in Education; 45attended in-service training within last 12 months; 55more than 10 years of teaching experience.

No other teacher-level variables were found to be significant for mathematics. In mathematics, like language, no same-gender relationships were found. This implies that while girls did poorly in mathematics overall compared with boys, that performance was not further enhanced or worsened by the gender of the teacher. In contrast, the findings from Table 4 show that for language achievement, teacher gender is related positively with language achievement. Further exploring these findings for language, while the results of the regression equation show a positive association between language learning and being taught by a woman, it is possible that some of these differences may stem from the differences between males and females teaching practices. If that is the case then explicitly accounting for teaching practices should undermine the significance of the coefficient on female teacher or in other words should make the coefficient for female teacher less significant or even insignificant. Four regression equations were analysed, each incorporating a different teaching practice variable in the equation. The variable for female teacher continued to remain significant and positive.11 (The magnitude of the coefficients ranged from 5.4 to 5.7.) Additionally, none of the teaching practice variables were statistically significant12 (Tables available upon request). In order to accord greater weight to classroom management practices, we also used the classroom management practice index in the regression which was generated by a summation of the three classroom management variables. Once again, even with this aggregate variable, the teacher gender variable continued to remain significant in the language equation and even the aggregate classroom management practice variable failed to attain significance in the model13 (Table available upon request). These findings imply that while male and female teachers are statistically significantly different from one another on the teaching practice variables measured here, these

638

A. Chudgar and V. Sankar

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

teaching practice variables in themselves do not matter for language achievement. More importantly, explicitly accounting for any one of these variables does not reduce the significance of teacher gender, in other words, at least these specific variables in themselves may not be capturing what about male and female teachers may lead to differential learning outcomes of their students. Finally, we used one additional approach to explore the positive association between female teachers and student achievement, once again focusing on subgroups of teachers as defined by their backgrounds. This analysis yielded an especially interesting outcome. Female-headed classrooms generally continued to outperform male-headed ones, but the female advantage seemed to disappear when we compared male and female teachers by their years of experience. Comparing male and female teachers with less than 10 years of experience showed that female teachers retained an edge in terms of their students language achievement. However, when we compared the male and female teachers who had more than 10 years of experience, the female teacher advantage was not significant. In conjunction with the finding in Table 4 that greater experience is associated with relatively lower language learning outcomes, this finding may imply that, for teachers with less experience, female teachers have an edge over males. This female teacher advantage also disappeared when we focused only on teachers who have not received inservice training in the recent past. Again, for this group, students do not do remarkably better in female-headed classrooms compared with male-headed ones14 (Tables available upon request). Discussion Overall the research on determinants of student achievement in developing countries is limited. One of the key challenges in taking this research forward has been limitation of data, an issue that this study is not entirely free from itself. This study of Indian teachers from large towns in five states and their effect on student learning would have benefited from stronger student background controls given the value of student background in understanding their learning achievement. In order to address this situation partially, we selected the sample of students who had different mathematics and language teachers and then used their test score in one subject as an additional control for the regression model for achievement in the other subject. This was an attempt to control for student ability which would inherently include the background influences that we were unable to measure explicitly. While the overall fit of the model improved both for language and mathematics, the relative importance of female teachers for language remained unchanged. Interestingly, with this additional student control, the female teacher variable which was previously insignificant for mathematics also became significant and remained positive. Sample size limitations disallowed us from taking this line of investigation forward, but the findings are reassuring given the overall conclusion of the study. It is also important to note that the data presented in this study have been collected from large towns in five Indian states and a fair amount of caution should be exercised in extending these findings to the country as a whole or to the broad developing country context. In order to draw more nationally representative findings, the sample ought to include rural data as well. For instance, research elsewhere has shown that schools located in remote areas or schools with poor infrastructure tend to have higher teacher absenteeism. Research studies have also found greater absence among male teachers and a relationship between teacher absence and student performance. Clearly, considerations like this may change the findings from rural and remote areas significantly compared with urban areas (Kremer et al. 2005).

Compare

639

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

With these caveats in mind, the study offers a valuable insight into the potential benefits of feminisation of the teacher labour force. Indian male and female teachers are different in their classroom management preferences and their views on students. However, the variables measured in this study on classroom practices or views on students are not sufficient to account for different learning outcomes of children in female- and male-headed classrooms in themselves. A potential line of investigation may focus on these gender differences in teaching practices to identify perhaps some variables that may indeed affect student achievement too. Additionally, these findings also reemphasise the limitations of focusing on teacher gender as a one-dimensional variable that only takes into account the sex of the teacher and pays no additional attention to the broader context within which the teacherstudent interactions are situated. On the broader issue of the impact of Indian female teachers on learning achievement, the results presented here offer only a partial support for hiring more female teachers in India if the goal is to improve girl students learning achievement. Female teacher classrooms tend to perform better in language but there are no such benefits for mathematics. Furthermore, these data do not show any evidence of specific beneficial effect on language or mathematics achievement for girl children of the presence of a female teacher. Equally importantly, the findings seem to indicate that the benefits of femaleheaded classrooms over male-headed classrooms may be limited to specific sub-groups of teachers such as the early-career teachers with fewer years of experience. Overall these findings based on data from large towns in five Indian states offer a cautious support for the policies to hire more female teachers to improve learning outcomes, as these relationships between teacher gender and student achievement may be mediated by the subject taught and teachers own background characteristics. Acknowledgements
The authors would like to thank Michelle Mwalimu, Michigan State University for excellent research assistance. The authors would also like to thank the editors and the anonymous reviewers for their helpful feedback on earlier drafts.

Notes
1. In recent years, a few notable exceptions in the Indian context have helped focus greater attention on achievement issues. A key effort in this direction has been Prathams Annual Status of Education Report (ASER), a rapid assessment of learning outcomes. This multi-year project focuses on creating a snapshot of status of schooling and status of learning. That is, it aims to record whether children within a certain age group can read a simple text, write a simple dictated sentence and do basic arithmetic operations. The data collection exercise has focused on obtaining a wide coverage of education in India through these simple tests (http://www.pratham.org). The second author on this paper from EI was involved with the data collection process. EI owns the data used in this paper. The first author obtained necessary permission from EI to use the data for the limited purpose of this article. However, all rights on the data continue to remain with EI. Funding available for the study limited the number of states that could be a part of the study. Further selection of states was based on the decision to have three Hindi speaking and two non-Hindi speaking states and the willingness of the state to participate. So for instance for teacher belief on the importance of strict discipline where the teacher thinks that strict discipline is necessary (1) or not (0), the following regression model was used: ; odds DISCIPLINE~1~exp a FEMTEACHzX b 1 In equation (1) a is the coefficient on female teacher (or teacher gender variable) and b is a vector of coefficients for the remaining teacher background variables expressed by the matrix X.

2.

3. 4.

640
5.

A. Chudgar and V. Sankar


So the multiple linear regression equation for language test-score for instance was: LANG~cfemstud FEMSTUDzcfemstud FEMTEACH ; ; zcsamegender SAMEGENDERza STUDzb TEACH 2

6. 7. 8. 9. 10.

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

11. 12. 13. 14.

In equation (2) cfemstud is the coefficient on female student which if significant would indicate that there are significant differences between girls and boys in their language scores. cfemteach is the coefficient on female teacher which if significant would indicate that students in male and female teacher classrooms do perform differently, and the significance of csamegender would indicate that girls in female teacher classrooms may have an additional (positive or negative depending on the sign of the coefficient) effect on their test score compared with boys in female teacher classroom. a and b are respectively the coefficient vectors for the remaining student and teacher variables expressed by matrices STUD and TEACH. Gender differences between male and female teachers were statically significant for all variables except teacher age. Once again, gender differences between male and female teachers were statically significant. This analysis was done using the prvalue command described earlier. The cluster correction command in STATA was used to account for the clustering of students within classrooms. Additional analysis of teachers with different experience revealed that of the less experienced teacher only 3% had received no teacher training compared with 13% in the more experienced sample. Similarly, the incidence of in-service training in less experienced teachers was higher at 88% compared with 73% in more experienced teachers. In fact, 12% of more experienced teachers reported no in-service training. Statistically significance at p#0.005. Statistically significance at p#0.05. Statistically significance at p#0.005. All the significance measured at p#0.05 unless otherwise indicated.

References
Baker, D.P., B. Goesling, and G.K. Letendre. 2002. Socioeconomic status, school quality, and national economic development: A cross-national analysis of the Heyneman-Loxley Effect on mathematics and science achievement. Comparative Education Review 46, no. 3: 291312. Chen, Y.M. 2000. Feminization in writing pedagogy: A study of teachers gender at EFL university composition classrooms. Taiwan: National Chung Chen University. Dee, T.S. 2006. The why chromosome: How a teachers gender affects boys and girls. Education Next 4: 6975. Department of Elementary Education and Literacy and Department of Secondary and Higher Education. 2004. Annual Report 20042005. New Delhi: Ministry of Human Resource Development, Government of India. Driessen, G. 2007. The feminization of primary education: Effects of teachers sex on pupil achievement, attitudes and behaviour. International Review of Education 53, no. 2: 183203. Grasha, A. 1994. Teaching with style: The integration of teaching and learning styles in the classroom. Essays on Teaching Excellence: Toward the Best in the Academy. Colorado State University: Professional and Organizational Development Network in Higher Education. Heyneman, S., and W. Loxley. 1983. The effect of primary-school quality on academic achievement across twenty-nine high-and low-income countries. The American Journal of Sociology 88, no. 6: 116294. Kremer, M., N. Chaudhury, F.H. Rogers, K. Muralidharan, and J. Hammer. 2005. Teacher absence in India: A snapshot. Journal of the European Economic Association 3, nos. 23: 65867. Krieg, J.M. 2005. Student gender and teacher gender: What is the impact on high stakes test scores? Current Issues in Education [Online], 8, no. 9. http://cie.ed.asu.edu/volume8/number9/. Lacey, C.H., A. Saleh, and R. Gorman. 1998. Teaching nine to five: A study of the teaching styles of male and female professors. Paper presented at the Annual Women in Educational Leadership Conference, 1112 October, in Lincoln, Nebraska. Laird, T., A.K. Garver, and A.S. Niskode. 2007. Gender gaps: Understanding teaching style differences between men and women. Paper presented at the Annual Meeting of the Association for Institutional Research, 26 June, in Kansas City, MO.

Compare

641

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

Mehrotra, S., ed. 2006. The economics of elementary education in India. New Delhi: Sage Publications India Private Ltd. Michaelowa, K. 2001. Primary education quality in Francophone sub-Saharan Africa: Determinants of learning achievement and efficiency considerations. World Development 29, no. 10: 1699716. Reddy, S. 2004. Status of learning achievements in India: A review of empirical research. Bangalore: Azim Premji Foundation. Sarangapani, P.M. 2003. Constructing school knowledge: An ethnography of learning in an Indian Village. New Delhi: Sage Publications India Private Ltd. Seetharamu, A.S. 2002. Status of elementary teachers in India: A review. In India education report, ed. R. Govinda, 190201. New Delhi: Oxford University Press. Singer, E.R. 1996. Espoused teaching paradigms of college faculty. Research in Higher Education 37, no. 6: 65979. Stacki, S. 2002. Women teachers empowered in India: Teacher training through a gender lens. New York: UNICEF. UNESCO. 2000. Increasing the number of women teachers in rural schools: A synthesis of country case studies: South Asia. Bangkok: UNESCO Principal Regional Office for Asia and the Pacific. . 2005. EFA global monitoring report 2005: The quality imperative. Paris: UNESCO. . 2006. Advocacy brief: The impact of women teachers on girls education. Bangkok: UNESCO. Velkoff, V.A. 1998. Women of the world: Womens education in India. Washington, DC: US Dept of Commerce, Bureau of the Census. Warwick, D.P., and J. Haroona. 1994. Teacher gender and student achievement in Pakistan. Comparative Education Review 38, no. 3: 37799. Xu, J., and J. Scott Long. 2005. Confidence interval for predicted outcomes in regression models for categorical outcomes. The Stata Journal 5, no. 4: 53759.

Appendix 1. List of variables with coding


Variable name Teacher gender Teacher experience Teacher qualification Teacher training In-service training Variable coding binary variable where female teacher51 and male teacher50 binary variable that equaled 1 if teacher had less than 10 years of experience and equaled 0 if the teacher had more than 10 years of experience categorical variable with values 15less than degree level education, 25Bachelors degree and 35Masters degree or more categorical variable with values 15no teacher training, 25certificate or diploma level training, 35a Bachelors or a Masters in education categorical variable where 15no in-service training, 25in-service training received more than 12 months ago, 35in-service training received within last 12 months binary variable where female student51 and male student50 binary variable where Yes51, No50 binary variable where Yes51, No50 binary variable where Yes51, No50 binary variable where Yes51, No50

Student gender Do you think strict discipline is necessary for proper teaching? Do you think there can be no discipline without fear of the teacher in students? Do you think the teacher loses respect if she/he plays and cracks jokes with the students? Do you think all children are capable of learning?

642

A. Chudgar and V. Sankar

Appendix 2: Additional tables


Table 2.1. Comparing teacher teaching practices of male and female teachers by state (numbers in %). Do you think strict Do you think there Do you think the teacher Do you think all children are discipline is necessary can be no discipline loses respect if she/he capable of for proper teaching? without fear of the plays and cracks jokes learning? with the students? teacher in students? States Andhra Pradesh Gujarat Uttarakhand Rajasthan Chhattisgarh Female Teacher 99.3 31.5 45.0 61.1 69.9 Male Teacher 99.6 48.1* 34.8 65.2 76.4 Female Teacher 59.6 27.1 23.6 55.2 51.8 Male Teacher 69.2* 22.2 41.8* 69.5+ 50 Female Teacher 24.2 6.4 9.9 18.8 18.05 Male Teacher 30.7 7.4 44.1* 17.3 13.2 Female Male Teacher Teacher 14.2 65.9 77.8 77.6 68.4 13.4 62.9 74.4 82.6 57.3

Downloaded by [UNESCO IIEP] at 06:50 28 February 2012

Note: *5p#.05, +5 p#.01.

Table 2.2. Difference in predicted probability between female and male teacher for a given response (numbers in %). Teacher attribute Do you think the Do you think all Do you think Do you think children are teacher loses strict discipline is there can be no capable of discipline without respect if she/he necessary for learning? proper teaching? fear of the teacher plays and cracks jokes with the in students? students? 27 211 211 24 211 29 211 27 29 29 29 28 29 29 28 29 28 26 24 211 27 26.5 25 26 8 8 8 8 8 8 8 8

More than 10 years of experience Less than 10 years of experience Less than Bachelors education Graduate degree No in-service training or in-service training more than 12 months ago In-service training in the last 12 months No or limited pre-service training Bachelors or Masters degree in education

Note: These are all differences in predicted probability between female and male teachers. So for instance, a female teacher with more than 10 years of experience is 7% less likely to say that discipline is necessary compared with a male teacher with more than 10 years of experience. Similarly, she is 8% more likely to say that all children are equally capable of learning compared to her male counterpart with more than 10 years of experience.

Das könnte Ihnen auch gefallen