A paper[pdf file] by Sonalde Desai, Cecily Darden Adams & Amaresh Dubey
India Human Development Survey, Working Paper No. 6, 2008.
The results reported in this paper are based primarily on India Human Development Survey, 2005. This survey was jointly organized by researchers at University of Maryland and the National Council of Applied Economic Research…Part of the sample represents a resurvey of households initially surveyed by NCAER in 1993-94.
…
Indian society has long been stratified along the axes of caste, ethnicity and religion. A large number of studies report inequalities in various outcomes along the caste, ethnicity and religion. Not surprisingly, this inequality is reflected in educational attainment too. However, the precise mechanisms through which inequality in educational attainments manifests itself remains open to debate with a variety of hypotheses being advanced such as poverty, child labor, lack of access to schools, teacher discrimination and lack of parental interest in education.
Unfortunately, there is little empirical research examining these hypotheses. Nor are the processes through which social disadvantages manifest themselves, clearly articulated. This paper utilizes a newly collected nationally representative survey data from over 41,550 households to examine social inequality in children’s educational outcomes. The focus is on 8 to11 year old children’s reading and mathematical skills.
…
While a variety of affirmative action programs are in place to bridge educational, occupational and income disparities between the dalits (Scheduled Caste), adivasis (Scheduled Tribe) and general populations, substantial educational disparities persist. Table 1, based on our past research (Desai and Kulkarni, forthcoming), shows that the dalits and adivasis as well as Muslims tend to lag behind Hindus and other religious groups. We have also found that a great deal of this inequality emerges in primary school with children from the marginalized groups dropping out before completing primary school. In fact, if these children manage to complete primary school, their likelihood of completing middle school is much closer to that of the other groups (Desai and Kulkarni, forthcoming). This suggests that primary school is an important site for the creation of educational inequality.
…
Tables 5 and 6 show the basic distribution of these skills for urban and rural children and children of various social groups separately. Not surprisingly, reading and mathematical skills are higher for urban than for rural children. Social group differences are also clearly evident in these descriptive statistics. Even among children at the same grade level, children from upper castes and religious groups like Christian, Sikh and Jains do far better in their educational attainment than the four other groups, OBC or the middle castes, dalits, adivasis and Muslims.
…
Females have lower reading levels than males – a finding that contrasts with most of the U.S. literature where girls have slightly higher reading scores than boys. The impact of social stratification on reading level is very large for this model. Other backward castes are about half as likely to attain any given reading level as upper castes, dalits are slightly more than one-third as likely (0.36 times as likely) and adivasis are only .32 times as likely.
Model 2 controls for current enrollment and completed education. As can be expected, the differences between different social groups diminish suggesting that at least some of the achievement differences are mediated through school enrollment and grade promotion between various groups. But surprisingly, this dampens inter-group differences at only a modest level. Muslims are 0.39 times as likely as upper caste Hindus to attain a given reading level in Model 1; after controlling for current enrollment and grade completed Muslim children are only about 0.47 times as likely to attain a reading level as upper caste Hindu children.
Models 3 and 4 add two basic socio-economic factors, urban residence and household economic status measured by the household ownership of consumer durables and housing assets. These two factors, particularly the household assests variable, dampen the relationship between social group and reading achievement substantially. But even so, dalits are only about 0.58 times as likely to achieve a given reading level as upper caste Hindus. Similar differences persist for other social groups.
The two variables controlling for adult education in Model 5 further reduce this relationship, although surprisingly this reduction is not very large. The number of years of completed education for the most educated adult in the household, and a dummy variable for literate adult in the household shows that higher level of household education helps diminish the negative impact of caste, ethnicity and religion on children’s reading achievements. However, even after all these controls are added, other backward caste children are 0.87 times as likely as upper caste children to attain higher reading scores and comparable proportions for dalits, adivasi and Muslims are 0.63, 0.79 and 0.64, respectively. It is important to note that even with these controls the negative effect of caste, ethnicity and religion persists.
We note that many of the variables that are included in our final model, Model 5, are themselves affected by caste, ethnicity and religion. Educational attainment in parental generation is also a function of social stratification. Additionally, the same school factors that result lower skill attainment for children may also affect their progression from one grade to another. So controlling for these factors, underestimates the impact of caste, ethnicity and religion on children’s skill attainment. But even so, substantial differences between children from different social backgrounds are obvious in the result we present.
…
This suggests that the differences in educational attainment between people of different social strata are not simply due to difference in enrollment rates nor are they solely due to parental lack of education and resources. Even when children from disadvantaged groups attend school, they fail to learn as much as their peers. Qualitative research and anecdotal evidence provides a variety of explanations for these findings. Teachers typically come from higher castes and have very low expectations for children from marginalized groups. They are also more predisposed to seeing the behavior of these children as being problematic than that of higher caste children. In our survey, we also asked children if the teacher treats them nicely. We found that children were extremely reluctant to say that the teacher did not treat them nicely but even so, while 76 percent of the upper caste children responded that their teacher treated them nicely, only 66 percent of the dalit and 65 percent of the Muslim children felt that way.
***
Read the full paper[pdf file] here.
