EFFECT OF PUPIL AND SCHOOL LEVEL VARIABLES ON THE ACHIEVEMENT OF SCHEDULED CASTE/SCHEDULED TRIBE STUDENTS
National Council of Educational Research and Training
New Delhi
ABSTRACT The study focuses on the difference between SC/ST and non-SC/ST students on mathematics and language achievement and identifies the pupil and school level factors attributable to those differences. The sample comprised 5,292 SC/ST and 17,771 non-SC/ST students of IV/V grade. Hierarchical Linear Regression Model and Meta Analysis were used for the analysis of data. The results reveal that SC/ST students score lower than non-SC/ST students in both the subjects. Father's education contributes for better achievement of SC/ST students. In mathematics, the achievement gap does not vary significantly across the schools whereas in language, it varies in Karnataka and Kerala. Test and feed back provided by the teacher tends to reduce the gap between SC/ST and non-SC/ST students in language.
Context and Focus The Educational Policy of India as revised in 1992 has asserted the need for equality in education which has been reiterated in section 4.1 of the document.
STUDIES ON DPEP 53
It states that : "The new policy will lay special emphasis on the removal of disparities and to equalise educational opportunity by attending to the specific needs of those who have been denied equality so far." Since the Scheduled Caste (SC) and Scheduled Tribe (ST) groups are one of the recognized disadvantaged group of the society, special planning and efforts have been made for the last couple of years to achieve equality in education for them. Equal accessibility to education for this disadvantaged group has been substantially augmented by the sincere effort of government by providing schooling facilities to the habitations predominantly populated by SC/ST. According to Programme of Action (POA, 1992), the dropout rate during 1987-88 for SC (51.60) and for ST (65.20) at primary stage is quite alarming as compared to the general category (46.97). Efforts are still on to overcome the problems of enrolment and dropout of children belonging to this group. In addition to other corrective measures, attractive school atmosphere supplemented with successful learning experiences is thought to be a solution to avoid these problems.
Student's achievement as an area of research has been extensively studied in our educational system. Shukla, et. al. (1994) found that SC/ST students had low performance at primary level as compared to backward and other caste students in the entire country. However, the result of the same study varied across states. The low attainment level of SC and ST students also enunciated in the Baseline Assessment Studies on different States (Jangira & Ahuja, 1995: Jangira & Paranjpe, 1995; Jangira & Yadav, 1994; Yadav, 1994; Gupta, 1995: Varghese, 1994). These findings generally provoke an exploration of causes related to the low achievement of SC/ST children. With this as the framework an attempt has been made to address the following three pertinent questions, such as:
(a) How far SC/ST and non-SC/ST students differ in their mathematics and language achievement?
(b) What is the difference that exists within schools and/or between schools?
(c) Which are the school level factors that minimize such differences ?
Survey of Earlier Studies Studies regarding the variables affecting achievement of SC/ST students at primary stage are very scant. However, studies in the area of correlates of achievement at primary stage were reviewed. Buch and Buch (1983) synthesized more than 200 studies focusing on the determinants of learning outcomes at the level of primary education. The correlates of pupil's performance are categorized into three groups as family characteristics, school characteristics and individual characteristics. Parent's social class, education and occupation and family environment are found to be significant influencing variables among the family characteristics. Facilities and equipments in the school, institutional climate and
54 IER: SPECIAL NUMBER 1995
leadership behavior of the principal, teacher qualification and training, high morale of teacher and positive perception of the academic ability of learner constitute a powerful set of factors determining the learning levels of children. Dave (1988), in a national study, compared the achievement level of students in different grades (I to IV) on mathematics and language achievement and concluded that there was a declining trend in the performance of students with increase in grade. However, the comparison between project and non-project schools has indicated the possibility of raising the attainment level of children through better classroom transaction and improved curricular input.
Several studies on home background have revealed its positive impact on student's learning. Sarkar (1983) reported that there was a significant difference between the high and low achievers in terms of educational environment, income, social background, provision of facilities and parent-child relationship. Studies of Jagannadhan (1985) and Shukla, et. al. (1994) also stated the discernible positive influence of home background on student's learning. Shukla (1984) explored the relationship of socioeconomic status (SES) of primary school students with their achievement in arithmetic, general science and social studies. The study revealed that SES was positively and significantly related to the achievement in these subjects. Malik (1984) found that children of illiterate parents had significantly lower academic achievement after controlling for their socioeconomic status and intelligence.
In a recent study Shukla (1994) investigated the relationship of schools factors with language and mathematics achievement of primary school students. She found that facilities for teacher, adequate classroom facilities, teaching experience of headmaster, time devoted to leaching arithmetic, educational facilities available in the school and functional parent-teacher association had a salutary effect on the student's achievement.
The reviewed studies provide a very significant information for improving the quality of primary education in general and enhancing student's achievement in particular. But these factors have not been investigated precisely for SC/ST students at the primary stage. These students being the weaker section of the society confront with economic deprivation, lack of enriching and stimulating environment at home which may irreparably limit their learning skills. Further educational and occupational status of parents of these children may restrict their active initiation to glamorize education as a valued experience. Additionally, educational deficits that SC/ST children brought with them to the classroom may be multiplied with various factors operated at the school level. On the other hand, in the earlier studies it is found that the influencing variables were not treated at pupil and school levels separately through robust statistical measure. Keeping in view these two aspects in mind a systematic attempt has been made in the present study.
STUDIES.ON DPEP 55
The exploration of these factors are not only prescribed for research but also assumed to fulfil the current objectives of District Primary Education Programme (DPEP). It will be immensely helpful for reducing the achievement gap between SC/ST and non-SC/ST children and improving their level of attainment at primary stage.
Data and the Sample Structure The data collected under Baseline Assessment Studies (BAS) were used in this study. The BAS covered 44 districts of Assam, Haryana, Karnataka, Kerala, Maharashtra, Madhya Pradesh, Orissa and Tamil Nadu. Detailed information about teacher, school and pupil's characteristics were collected through different schedules meant for teacher, school and pupil respectively. To estimate achievement of students on language and mathematics, standardized achievement tests (Shukla, et. al. 1994) were used. The sample comprised of 5,292 SC/ST and 17,771 nonSC/ST students of IV/V grade.
TABLE 1
Percentage of Schools at Different Levels of Concentration of SC/ST
Students
Level of Concentration
Mean
States Total 0% 0.25% 25.50% 50-75% 75-100% %
Schools of SC/ST
in the
Sample
Assam 161 18.45 12.42 6.21 5.59 27.33 33.19
Haryana 144 14.58 47.22 22.22 10.42 5.55 25.55
Karnataka 176 22.73 34.66 23.86 10.23 6.52 25.30
Kerala 113 30.97 54.87 10.62 3.54 0 11.41
Maharashtra
135 12.59 31.11 28.15 8.89 19.26 37.53
Madhya
Pradesh 733 21.28 29.74 24.97 11.46 12.55 30.67
Orissa 165 14.54 16.97 20.61 16.36 31.51 48.23
Tamil
Nadu 119 26.05 36.97 16.81 6.72 13.44 26.78
Total 1746 23.02 31.10 21.25 10.14 14.49
The scrutiny of SC/ST sample (Table 1) reveals that the highest percentage (48.23) of SC/ST students are found in the state of Orissa. Also concentration of SC/ST students in schools is the highest in Orissa where SC/ST students constitutes 75 to 100 per cent of the total children in 31.51 per cent schools.
56 IER: SPECIAL NUMBER 1995
Taking total sample under consideration, 23.02 per cent of schools have no SC/ ST students and 31.10 per cent schools have up to 25 per cent of SC/ST students.
Data Analysis and Procedure In order to answer each of the three questions mentioned above, the- analysis can broadly be classified into three stages. The first stage consists of exploring the raw scores with respect to mean achievement of SC/ST and non-SC/ST. The achievement scores are dependent on students' background and schools policies and practices. Both of them are not independent but are nested in each other. The second stage comprises analysis of data using hierarchical data structure in multilevel modeling (Bryk and Raudenbush, 1992). In this process of HLM analysis, various models relevant to monitor school performance (Willms, 1992) have been used. In order to explore the behavior of achievement of the SC/ST students further, a step-wise regression analysis is undertaken to estimate the contribution of some of the important student level variables, i.e., related to home background and whether the student has repeated a class or not. The aforesaid analysis is carried out separately for mathematics and language achievement in two states (Kerala and Tamil Nadu). The results of adjustment for student background are expected to be at variance because of variations among states. The results are synthesized by using meta analysis after adjustment of achievement scores for students' background. Finally, the adjustment for contribution of school practices and policies have been attempted in those states where slope of SC/ST vary significantly across the schools at students' level. This analysis aims to identify the school level factors responsible for explaining the gap between SC/ ST and non- SC/ST students.
Variables Covered The scores obtained by SC/ST and non-SC/ST students on language and mathematics are treated as outcome variables. The maximum score in the mathematics test is 40 whereas the score in language is 44 and 84 in class IV and V respectively. Mathematics and language scores are standardized at the individual level using the state mean and standard deviation. The average scores obtained from HLM analysis are expressed as effect size, that is, as fraction of a standard deviation.
Pupil Level Variables A series of pupil background variables are constructed from pupil schedule (present). The variables sex, caste, repeated the class and preschool experience
STUDIES ON DPEP 57
are treated as dummy variable. Ordinal measures are constructed for father's education, mother's education and father's occupation. Socio-economic status (SES) is a composite variable of the above mentioned three variables related to parental status. Family size, head of the family and preschool experience are excluded from treatment examining their negligible influence. Pupil background variables are centered around at the pupil level on the basis of their state mean. So the estimated value is considered as the score of a child having the average state characteristics of that variable.
School Level Variables School level variables are selected from the teacher and school schedules and some are aggregated from pupil schedule. The first two are contextual variables. i.e., Mean SES and percentage of SC/ST children which are constructed by aggregating data from pupil level to school level. Rest of the school level variables are clustered into four categories - (a) Teacher Quality, (b) School resources. (c) School climate, and (d) Intervention launched in schools. Teacher qualification. teaching experience, in-service education and stability of teacher in terms of duration of service in the same school are the components of teacher quality. In school resources seven variables - teaching material for teachers, instructional material available in school, physical facility, number of teachers, pupil-teacher ratio, school with only primary section and percentage of female teachers--are included. Academic pressure in terms of specific attention to language and mathematics subjects, test and homework given by teacher, teacher commitment. parent involvement and leadership of head teacher (all together eight variables) are clustered in to school climate. Seven variables are constituted on operation blackboard and different incentive schemes for intervention category. Among these variables, operation blackboard and school with only primary grade are treated as dummy variables. Rest are operated as continuous variables. Composite variables are formulated by giving proper weightage through logit distribution. All variables are centred at the school level on the basis of their state mean.
Results In the first step of analysis significant difference between raw achievement scores in mathematics and language of SC/ST and non-SC/ST was tested through 't' test. The results presented in Table 2 shows that in mathematics achievement SC/ST students performs less than non- SC/ST students in all the states except in Maharashtra and Tamil Nadu. Whereas, in language, all the states indicated the same results except in Maharashtra. Further, the gap in achievement in both the subjects is the highest in the state of Madhya Pradesh. These findings are not of
58 IER: SPECIAL NUMBER 1995
much use because it is confounded with the interaction effects of student background and school level factors where students are nested with the school. The following discussions are based on the estimates after controlling, for these variables.
TABLE 2
Comparison of SC/ST and Non-SC/ST Students on Mathematics and Language Achievement
(Unadjusted Scores)
Achievement Scores
States Group N Mathematics Language
Mean SD t Mean SD t
SC/ST 411 18.91 7.76 20.36 7.40
Assam 3.03 2.84**
Non-SC/ST 1295 20.17 7.18 21.50 6.97
SC/ST 603 14.14 5.06 35.63 11.03
Haryana 5.57** 3.62**
Non-SC/ST 1886 15.53 5.47 37.57 11.59
SC/ST 594 13.87 7.72 14.96 7.16
Karnataka 6.09** 5.34**
Non-SC/ST 1974 16.09 7.78 16.75 7.20
SC/ST 351 14.11 5.37 18.30 7.42
Kerala 2.72** 5.88**
Non-SC/ST 2738 14.98 5.70 20.99 8.15
SC/ST 745 11.40 6.68 16.56 8.54
Maharashtra 1.39 .14
Non-SC/ST 1398 11.81 6.38 16.61 8.29
SC/ST 2081 10.61 5.18 26.72 14.21
Madhya 6.10 * * 7.09**
Pradesh Non-SC/ST 5937 11.44 5.34 29.37 14.77
SC/ST 517 13.47 6.78 32.30 12.37
Orissa 5.02 * * 5.75**
Non-SC/ST 847 15.31 6.40 36.40 12.98
SC/ST 627 10.98 4.64 29.29 10.51
Tamil Nadu 1.82 3.34**
Non-SC/ST 1696 11.38 4.64 30.99 10.98
** Significant at .01 level
STUDIES ON DPEP 59
PUPIL BACKGROUND VARIABLES AND ACHIEVEMENT OF SC/ST STUDENTS The achievement scores contain effects of students' background and school policies and practices. In order to partial out the effects of these factors, the variation in standard scores is split into two components, i.e., within schools and between schools for each state. The within school variance accounts for pupil's background, i.e., students level whereas between school variance contains effects of school policies and practices. A large amount of variance pertaining to pupil level ranges from 39.42 (Karnataka) to 80.44 (Kerala) per cent in mathematics and 54.62 (Madhya Pradesh) to 70.69 (Orissa) per cent in language. The between school variance ranges from 19.55 (Kerala) to 48.89 (Assam) per cent in mathematics and 13.81 (Kerala) to 48.81 (Karnataka) per cent in language. Each of these variances are highly significant. It indicates that in addition to the within school difference there is a significant difference across the schools on mathematics and language scores.
Table 3A represents the achievement gap between SC/ST and non- SC/ST students separately for each state before and after adjusting for all the pupil's background variables. SC/ST variable was constructed by assigning 1 if students belong to SC/ST group and 0 to non-SC/ST. So the estimated effect size is the difference of achievement between SC/ST and non-SC/ST in the respective subjects. It may be observed that gap in achievement for both the subjects between SC/ST and non-SC/ST has decreased uniformly in all the states after adjustment of pupil's background. The reduction is quite noticeable in the state of Assam, Maharashtra and Orissa where the gap in mathematics achievement became not significantly different from zero. Such change of status is observed for only Orissa in the case of language achievement. However, this gap varies across states in both the subjects. Kerala state has recorded the highest gap in mathematics and language, i.e., 23.4 per cent and 38.4 per cent of their respective standard deviations. It may be noted that in Kerala, the within school variances of both the subjects are the highest among all the states. Among the states depicting the statistically significant gap, it is the lowest in the case of Madhya Pradesh in both the subjects (6.1 per cent in mathematics and 7.5 per cent in language of their respective per cent standard deviations).
The variation in the estimated size of the gap in achievement of mathematics and language across states may have a component of sampling variance. In an attempt to adjust for the sampling errors in the estimates of the gap, the meta analysis with known variance is applied separately for mathematics and language (Table 3B). The results indicate that average achievement gap is statistically significant for both in language and mathematics at three per cent and one per cent respectively. The analysis -further points out that the variation in the achievement gap across states in mathematics is not statistically significant. On
60
STUDIES ON DPEP 61
the other hand, estimates of gap in language achievement vary significantly across states at one per cent level. These findings indicate that SC/ST students across states score lower than non-SC/ST students by 8.6 per cent of the standard deviation in mathematics. Whereas in language SC/ST students score lower than non-SC/ST students by 10.1 per cent of the standard deviation.
SCHOOL VARIABLES AND ACHIEVEMENT OF SC/ST STUDENTS Further, the achievement gap between SC/ST and non-SC/ST in mathematics does not vary across schools within a state. Whereas, the gap in language achievement
TABLE 4
School Variables Affecting the Achievement Gap of SC/ST and Non-SC/ST Students in the
States of Karnataka and Kerala
Karnataka Kerala
Effects Coeff. SE Effects Coeff. SE
Av. Ach.Gap (Lang) .008 .054 Av. Ach. Gap (Lang) -.325** .063
Teaching Exp. -.011 .006 Teaching Exp. -.027** .012
Physical Facility .014 .022 Teacher Gives Test .364** .111
Pupil Teacher Ratio -.002 .003 Parent Involvement -.286* 135
Primary Grade only -.113 .115 Operation Blackboard -.356** .120
Per. of Female Teacher-.004 .003 Free Uniform -.054** .021
Parental Involvement .095 .056 Free Text Book -.007 .004
Mid-day Meal -.003 .002
Variance Across the Karnataka Kerala
School -
Variance df Chi-Sq. P-value Variance df Chi-Sq. P-value
Average Ach. Gap
(After Adjusting
Pupil Background
Variables) 0.115** 106 166.19 .001 0.129* 77 126.36 .001
Average Ach. Gap
(After Adjusting
Contextual Variables)0.106** 106 164.54 .001 0.141** 77 122.48 .001
Average Ach. Gap
(After Adjusting
School Level Variables)0.086* 98 138.33 .005 0.035 71 79.81 .22
*Significant at .01 level.
*Significant at .05 level.