Reaching the marginalized: EFA global monitoring report, 2010; 2010

100CHAPTER 32Education for All Global Monitoring Report8. In three test rounds,scores for low-castechildren fell by 14%,25% and 39%.Figure 3.24: In India, scheduled castes and tribes remaindisadvantaged at all levels in educationAttendance rates by age group in India, by community, rural/urban residenceand gender, 2004/2005School attendance rate (%)10080604020National averageScheduled casteScheduled caste girls, ruralScheduled tribe6–11 12–14 15–18Age groupScheduled tribe girls, ruralNotes: The attendance rate for an age range is the proportion of children of that agerange who report attending school at the time of the survey. The age ranges correspondapproximately to primary education, upper primary (or ‘middle’) education and secondaryeducation, respectively, in the Indian school system.Source: Bhalotra (2009) based on National Sample Survey data (61st round).To what extent do these differences stem fromdistinctive caste and tribe disadvantages ratherthan wider social and economic factors? That is akey question for policy-makers seeking to equalizeopportunity. Research for this Report helps providea partial answer (Bhalotra, 2009). Using householdsurvey data, and controlling for household andindividual characteristics, the study found thatabout 60% of the attendance gap for scheduledcastechildren aged 6 to 14 could be attributed tohousehold characteristics, mainly poverty and lowerparental education. For scheduled-tribe childrenin the same age group, household characteristicsweighed less heavily, accounting for about 40% ofthe attendance gap. One conclusion to be drawn formembers of both scheduled groups is that povertymatters a great deal in perpetuating educationaldisadvantage. However, the non-poverty componentis larger for scheduled tribes partly because of theweight of social and cultural discrimination.Public attitudes have consequences that go beyondschool attendance. Institutionalized stigmatizationcan erode self-confidence and levels of expectation,undermining children’s potential for learning.One particularly striking illustration comes from anexperimental investigation into the impact of casteperceptions on test scores (Hoff and Pandey, 2004).Children aged 11 and 12 were chosen at randomfrom a low caste and three high castes, and givena series of puzzles to solve. When caste was notannounced to the participants, it had no bearingon the initial score or on the improvement in scoreregistered in subsequent test rounds. But whencaste was announced before the test, the scores forlow-caste children fell dramatically (Figure 3.25). 8These findings underline the degree to whichsocial identities that are a product of history,culture and personal experience can createpronounced education disadvantages throughtheir effects on individual expectations.The critical role of languageLanguage and ethnicity are deeply intertwined.Having a distinctive language is often a crucialelement of personal identity and group attachment.Just as a local language may be a point ofassociation for members of an ethnic group, itcan also be an element in their marginalization.People who cannot speak a country’s dominantlanguage may have less access to written andspoken sources, restricting their opportunitiesfor employment and social mobility (Smits andGündüz-Hosgör, 2003; Smits et al., 2008). Parentswho do not speak the official language in whichtheir children are being educated may have lessopportunity to engage with teachers, educationauthorities and homework. And their childrenmay not grasp what is being taught if teachersdo not speak their home language. The resultinginequalities in opportunity are a major factorFigure 3.25: Social stigma can undermine test performanceExperimental impact of the announcement of caste on solvingpuzzles in IndiaAverage number of puzzles solved43210Caste not announcedCaste announcedHigh casteLow casteNote: Children aged 11 and 12 were given a packet of fifteen maze puzzlesand asked to solve as many as they could in fifteen minutes.Source: Hoff and Pandey (2004).172