EFFECT OF SCHOOL LEVEL VARIABLES ON ACHIEVEMENT GAP BETWEEN BOYS AND GIRLS
National Council of Educational Research and Training
New Delhi
ABSTRACT The central focus of the study is to explore the achievement gap between boys and girls on mathematics and language and school level factors that are associated with this gap. A sample of 1, 746 schools, comprising 4,879 teachers and 23,700 students was covered in the study. Hierarchical Linear Model and Meta-analysis were used to analyse the data. The findings revealed that girls score approximately 12 per cent and 11 per cent standard deviation lower than boys in mathematics and language respectively. The continuous stay of teachers for not more than five years in the same schools, proper qualification of teachers, appropriate number of teachers and higher percentage of female teachers are likely to improve the performance of girls and thus the primary education.
Introduction There is no discrimination for admission to educational institution, but socioeconomic factors often contribute to several segments of the population remaining socially and educationally backward. Generally, groups vulnerable.
STUDIES ON DPEP 27
to social discrimination also suffer from economic deprivation. Amongst the deprived section of the population, women are doubly disadvantaged- suffering from an unequal status from the family as well as-in the society. Facilities for education are equally available for both boys and girls, yet there is significantly lower level of utilisation by girls and women. Perhaps this leads to low level of achievement by the girls and finally dropout from school.
Against the background of demographic implication and the complex ground realities of Indian scene, the goals of Education for All, MHRD (1993) in the country constitute (i) Expansion of early childhood care and development activities for poor and disadvantaged group, (ii) Universalisation of elementary education (UEE), (iii) Drastic reduction in illiteracy, (iv) Provision of opportunities to maintain and upgrade education, (v) Creation of necessary structures which could empower women and make education an instrument of women's equality and (vi) Improving the content and process of education to relate it better to the environment.
In order to achieve Universalisation of Elementary Education, the Government of India launched District Primary Education Programme (DPEP) in 1993. In this regard, Baseline Achievement Study (BAS) was undertaken in 44 districts of eight states of Assam, Haryana, Karnataka, Kerala, Madhya Pradesh, Maharashtra, Orissa and Tamil Nadu to identify interventions for implementing the programme.
Review of Literature Three major studies undertaken in the National Council of Educational Research and Training (NCERT), involving several states, examined achievements of primary school education. Kulkarni(1970) conducted a study at three levels of education, i.e., at the end of primary, middle and secondary stage showed that with few exceptions boys achieved higher than girls. Dave.(1988) observed that students differ in their achievement in mathematics and language.
Shukla (1994) also conducted a study in 22 states and a union territory and inferred that boys scored higher than girls. There was a evidence for boys to achieve better in Arithmetic and Word Knowledge and teachers's knowledge of the subject was highly inadequate. Pupil's ability had the maximum influence on his learning of school related tasks and it was followed by home background variables, facilities for learning and educational environment at home.
The studies summarised in the five surveys of Research in Education published by the NCERT indicated that pupil achievement was found to be related with general ability and motivation of the pupil, availability of the teacher per class group, level of general and professional education of the teacher, facilities available in the school, institutional climate and leadership of headmaster.
28 IER: SPECIAL NUMBER 1995
The studies referred to above have shown that sex differences exist in the learning achievement of language and mathematics. However, no effort has been made in the direction that how the school level variables predict this gender gap in achievement. In fact cross-sectional or independent group(s) were used for data collection but the hierarchical data structure in educational system were not considered.
Research Questions The present paper attempts to study the following research questions
(a), To what extent boys and girls differ in their mathematics and language achievement? (b) To what extent the school level factors are associated with the gender gap?
The achievement scores were regressed on pupil and school level variables using "Hierarchical Linear Model" and were further synthesized for all the eight states using "Meta-Analysis".
Sample A sample of 1,746 schools from 43 districts of the eight states comprising 4,879 teachers and 23,700 students was drawn using three stage stratified random sampling procedure. In a school not more than 30 students were administered the test. Similarly, Teacher Schedule was given to not more than five teachers in a school. Data of district Dhar, Madhya Pradesh could not be included in the study due to non-availability of language scores on all the 84 items of the test.
Tools and Indicators Used A test of 40 items in Mathematics, based on grade III curriculum, was administered to the students of grade IV at the beginning of session in the states of Assam, Maharashtra, Kerala and Karnataka. Another test of 40 items, based on grade IV curriculum, was administered to students of class V in Haryana, Madhya Pradesh, Orissa and Tamil Nadu. Similarly, the test of Language, comprising 44 items for grade IV students and 84 items for grade V students, was administered in their regional language, which is also a mother-tongue in majority of cases. In addition to this three schedules namely, Pupil, Teacher and School were used to collect the required information. The description of indicators used in each of pupil's background (7), contextual (2), teacher's
STUDIES ON DPEP 29
quality (4), school resources (7) and academic climate of school (8) are given in Annexure I.
Analysis Procedures For an exploratory analysis, boys and girls have been compared on mathematics and language achievement using Student's statistics to know the sex differences in the groups. The achievement scores of boys and girls, whose SES was below 10th and above 90th percentile, were selected from the sample and graphs between SES and achievement scores were drawn to know their relationship.
In the present analysis, only two level Hierarchical Linear Model (1992) was formulated in order to obtain stable results. Level- 1 includes the pupil and level-2 includes the school level variables. The variation in mathematics and language achievement scores are partitioned into within (level-1) and between (level-2) units. Further, these variance have been adjusted at different levels to determine the significant effect of the variables. In order to see the unadjusted effect of gender gap on achievement, only GIRL was included in the model.
The within school variance of outcome variable was further adjusted for pupil level variables and termed as Type 'A' effect. This effect was further adjusted for the two school composition variable, i.e., mean of student's SES and per cent of SC/ST on the school means to minimize the bias due to these factors and termed as Type `B' effect. The three groups of variables, i.e., Teacher's Quality, School Resources and Academic Climate of School, were included in Type `B' effect to see the effect of these variables. Finally all the variables, which had the probability less than 20 per cent (p<.20) were included in the Model and the results based only on this reduced model have been discussed in this paper.
Since the present paper attempts to find the effect of school level variables on the gender achievement gap, only the effects of GIRL slope have been discussed here.
The scores of both the achievement tests have been standardised at mean zero and standard deviation one. The values of variables, which were in ordinal scale, were given weights using Logit distribution. Variables obtained from pupil and teacher questionnaire were aggregated and centered at school level for each state. The variables of school questionnaire were also centered. The mean of all the variables is given in Annexure II.
Results The percentage of girls (Annexure II) varied from 37.3 -in Orissa to 48.8 in Haryana. Kerala is the only state where all schools of the sample were co-
30 IER : SPECIAL NUMBER 1995
educational. Tamil Nadu had only one school where there were no girls. In Madhya Pradesh, 165 schools were exclusively for boys, whereas 100 schools were exclusively for girls.
TABLE 1
Means and Standard Deviation of Achievement Scores
State Subject Sex Mean SD N `t'
Assam Math Boys 20.17 7.17 880
Girls 19.54 7.51 826 1.77
Lang Boys 21.58 6.80 880
Girls 20.85 7.38 826 2.14*
Haryana Math Boys 15.52 5.44 1274
Girls 14.86 5.35 1215 3.04**
Lang Boys 36.79 11.30 1274 *
Girls 37.43 11.67 1215 1.40
Karnataka Math Boys 15.75 7.82 1541
Girls 15.31 7.82 1027 1.39
Lang Boys 16.50 7.20 1541
Girls 16.10 7.27 1027 1.38
Kerala Math Boys 14.97 5.78 1601
Girls 14.80 5.56 1488 0.80
Lang Boys 20.46 7.932 1601
Girls 20.92 8.309 1488 1.57
Madhya Math Boys 11.50 5.30 4600
Pradesh Girls 10.86 5.31 3418 5.33**
Lang Boys 29.17 14.48 4600
Girls 28.02 14.96 3418 3.45
Maharashtra Math Boys 11.99 6.27 1207
Girls 11.25 6.75 936 2.63**
Lang Boys 17.30 8.21 1207
Girls 15.70 8.51 936 4.40**
Orissa Math Boys 14.71 6.76 855
Girls 14.45 6.33 509 0.69
Lang Boys 34.86 13.13 855
Girls 34.82 12.54 509 0.00
Tamil Nadu Math Boys 11.19 4.35 1260
Girls 11.36 4.97 1063 0.89
Lang Boys 30.93 10.23 1260
Girls 30.05 11.59 1063 1.94
Note : Level of significance * p<.05; ** p<.01
STUDIES-ON DPEP 31
In order to answer the first research question, two approaches, i.e. sex differences on achievement scores by Student's 't' statistics and achievement gap obtained from HLM analysis have been discussed in the next section.
ACHIEVEMENT GAP BETWEEN GIRLS AND Boys Table 1 shows, on an average, the performance of girls is lower than boys in mathematics. Significant sex differences are found in Haryana, Madhya Pradesh and Maharashtra, where the girls scored marginally higher (.17) than boys.
In language, the performance of girls is lower than boys and are significantly different in the states of Assam, Madhya Pradesh and Maharashtra. However, the girls scored marginally higher, not statistically significant, in Haryana and Kerala.
FIG. 1 Effect of SES on gender gap in language achievement of extreme groups (Maharashtra)
On the basis of sex differences, two states, Madhya Pradesh and Orissa were selected to see the relation of SES on mathematics achievement. Similarly, Maharashtra and Orissa were selected for language achievement.
In Maharashtra (Fig. 1), the gap in language achievement exists distinctively in low and high SES groups. The girls performed lower than boys.
32 IER: SPECIAL NUMBER 1995
FIG. 2 Effect of socioeconomic status on language scores of extreme groups (Orissa)
However, the gap marginally narrows down as SES of students increases.
In Orissa (Fig. 2), within low SES group, the girls performed better than boys in language. The situation is reversed as the SES of students increases from their group mean. The gap widens when it moves towards extreme points.
TABLE 2
Percentile Scores of Socio-Economic Status for Boys and Girls
State BOYS Girls
10th 90th 10th 90th
Assam - 844 1.044 -.844 1.192
Haryana -.833 .961 -.556 1.033
Karnataka -.881 .980 -.655 1.186
Kerala -.994 .823 -.705 .873
Madhya Pradesh -.739 .824 -.739 1.281
Maharashtra -.982 .978 -.780 1.032
Orissa -.855 .989 -.855 1.103
Tamil Nadu -.816 .842 -.816 1.101
STUDIES ON DPEP 33
Fir,. 3 Effect of SES on mathematics achievement of extreme groups (Madhya Pradesh)
In mathematics (Fig. 3), the girls performed lower than boys in both SES groups in Madhya Pradesh. However, the gender gap has narrowed down as SES of students has increased. Similar to Madhya Pradesh, girls performed lower than boys in Orissa (Fig. 4). The low SES students do not differ on the achievement. However, the gap increased as SES of students increases.
TABLE 3
Unadjusted Within and Between School Variance in Mathematics and
Language Scores
Variance Assam Haryana Karna- Kerala Madhya Mahara- Orissa Tamil
taka Pradesh shtra Nadu
Mathematics
Between School
48.3 31.3 60.6 19.6 45.6 39.3 44.1 36.0
Within School
51.7 68.7 39.4 80.4 54.4 60.7 55.9 64.0
Language
Between School
35.0 35.6 45.2 13.7 45.4 37.7 29.3 32.8
Within School
65.0 64.4 54.8 86.3 54.6 62.3 70.7 67.2
34 IER : SPECIAL NUMBER 1995
FIG. 4 Effect of SES on mathematics achievement of extreme groups (Orissa)
One may observe from Table 2 that at 90th percentile, girls had the high SES than boys in all states. Whereas, at the 10th percentile, boys of Haryana, Karnataka, Kerala and Maharashtra had the high SES than girls. This shows that girls of high SES and boys of average SES are studying in these schools.
The HLM analysis reveals that the between school variance (Table 3) varies from 19.6 per cent in Kerala to 61 per cent in Karnataka for mathematics. In language, it varies from 13.7 per cent in Kerala and 45.2 per cent in Karnataka. The variation is significant in all the states.
The variation of achievement gap after adjusting the pupil and school level variables has been discussed in the following section.
One can observe from Table 4 that the achievement gap has increased after controlling for the pupil level variables in both the subjects in almost all states. Kerala is the exception, where the achievement gap in mathematics has decreased from 0.048 to 0.007. The meta-analysis shows that the girl slope is negatively associated with the mathematics (-.122, p<.01) and language (-113. p<.01 ) achievement for Type-A effect in most of the states. However, in Kerala. the slope (.007) is marginally positively associated with the language
It indicates-on an average, girls score lower than boys, approximately 12.2
STUDIES ON DPEP 35
36 IER: SPECIAL NUMBER 1995
BETWEEN AND WITHIN SCHOOL VARIATION In mathematics, the between school variation (Table 5) of gender gap varies from 6.8 per cent in Madhya Pradesh to 21.9 per cent in Maharashtra. The within school variation of the achievement scores varies from 32.5 per cent in Karnataka to 69 per cent in Kerala. In language achievement, the between school gender gap varies from 6.1 per cent in Kerala to 20.5 per cent in Maharashtra. The variation is significant across the schools in all the states.
TABLE 5
Residual Variance as Percentage of Total Variance for Type `A'
Effect
Asm Har. Kar. Ker. M. P. Maha. Ori. T.N.
MATHEMATICS
Gender Gap
11.49 10.82 13.88 11.38 6.84 21.93 7.14 12.05
Pupil's Score
43.77 62.03 32.54 69.00 50.52 42.50 52.47 56.83
LANGUAGE
Gender Gap
13.63 8.09 13.01 6.07 8.80 20.48 8.72 13.52
Pupil's Score
55.77 59.45 39.81 78.46 48.80 47.06 65.61 56.12
It also shows that, for both the subjects, learning achievement of students varies across the states. Hence, there is a need to further explore the school level factors affecting the learning achievement of children.
In order to answer the second research question the effects of school level variables from HLM analysis are presented in the following sections.
Effect of Teacher's Quality Meta-analysis (Table 6) shows that, on an average, teacher's qualification, i.e., years of schooling. is positively associated (.080, p<.05) with the gender gap on language achievement in Assam. Maharashtra, Orissa and Tamil Nadu. It implies, on an average. one unit (year) increase of teacher's schooling may increase approximately 8 per cent of language achievement of students from mean.
Orissa and Tamil Nadu have been considered to know the predicted achievement slope of boys and girls for three cases of period of teacher's schooling as (i) the state average, (ii) 10 years and (iii) 14 years of education. For the state average, girl-TCHQUAL slope indicates that girls tend to score
STUDIES ON DPEP 37
TABLE 6
Effects of Teacher's Quality on Gender Gap
Variables Asm. Har. Kar. Ker. M. P. Maha. Ori. T.N. A V
Coef. Coef. Coef. Coef. Coef. Coef. Coef. Coef. Coef.
(SE) (SE) (SE) (SE) (SE) (SE) (SE) (SE) (SE)
MATHEMATICS
Teacher Qual. 158** .077 -.040
(.055) (.051) (.117)
Period of Stay. 018 -010* .017 .006
(.013) (.005) (009) (.010)
LANGUAGE
Teacher Qual.
.075 .058 .071 .115** .080*
(.042) (.049) (.041) (.046) (.022)
Inservice .245 .089 .117*
Training (.143) (.066) (.063)
Period of Stay -.018 .027* -.042** .017 -.002
(.015) (.013) (.016) (.009) (.016)