Introducing the Student Bodies

Our final group of 449 respondents provides perhaps surprisingly wide variation in response to most of the questions posed even though it is an appropriate representation of the emerging middle class. The sample is divided almost evenly between the three universities (37 per cent at JNU, 31 per cent at Pune, and 32 per cent at Hyderabad) and the three major fields (35 per cent natural science, 33 per cent social science, and 32 per cent humanities). Females account for 49 per cent and males 51 per cent and 34 per cent of the respondents are between the ages of 18 and 21, while 40 per cent are between 22 and 25, and 19 per cent are between 26 and 30.

For most of the analysis to follow, we will focus on the 75 per cent of the respondents who identified themselves as Hindus (337 of the total 448). Another 11 per cent are Christians, 5 per cent Muslims, 4 per cent Buddhists, 2 per cent Jains, 1 per cent Sikhs, and 2 per cent did not provide a religious identification. Because caste is a somewhat more elusive marker, we asked about it in two ways. First, we asked each respondent to select the category which best represented their caste position; 39 per cent responded upper caste, 25 per cent middle caste, 9 per cent backward caste, 4 per cent scheduled caste, and 7 per cent did not answer. Then we asked them to write the name of their particular ‘jati’ and/ or ‘varna’ and had members of the Indian research team apportion them to the categories above, that they did with high consensus and reasonable correspondence to the respondents’ self-placement (a correlation of 0.77, significance level < 0.001). According to the “judges”, 40 per cent of our Hindu respondents were upper caste, 20 per cent middle caste, 7 per cent backward caste, and 11 per cent scheduled caste. We used both caste placements for different purposes, sometimes seeking subjective self-perceptions but more often heeding the objective assessments of unbiased judges, as was the case here.

Finally we utilised several different measures of what might be called class rank. First, we asked our respondents to place their father’s occupation within a set of categories provided. Here 25 per cent fell within our highest category of “professionals or executives,” 15 per cent were “family business owners or managers”, 17 per cent were “white-collar office workers”, 19 per cent were “farmers or agricultural workers”, and 6 per cent were “labourers”. Some 14 per cent placed themselves in an “other” category, and another 4 per cent did not respond. Next we asked of the language spoken at home, and 65 per cent reported a regional vernacular, 26 per cent said Hindi, and 4 per cent English. Finally, when asked what kind of secondary school they attended, 44 per cent reported a convent or private school, while 54 per cent said a government school.

Considering the substantial variance on all of these class factors, it is clear that our sample of university students reflects a new middle class that is at least as much a meritocracy as selfperpetuating elite. If education is a prime means of intergenerational status reproduction for half the sample, it is a prime source of mobility for the other half. But if this sample is blessed with ample variance concerning religion, caste, and class, what about other factors?

Caste Identification and Religious Involvement

We asked our respondents a number of questions concerning caste, but several suffice as an opening sub-set. All take the form of statements to which the respondent was asked to select among four possible answers: strongly agree, agree, disagree, and strongly disagree. For example, “My caste is an important factor in the way I think of myself”. Because this and the questions to follow would not be appropriate for many Muslims, the responses

Economic and Political Weekly September 2, 2006

summarised are for Hindus only. Of this group, 9 per cent strongly agreed that caste was important to their self-conception, 19 per cent strongly agreed, 40 per cent disagreed, and 32 per cent strongly disagreed. Thus, nearly one-third report having “caste off”, i e, with little sense of caste identification remaining, and another two-fifths seem poised to make the toss. Less than a 10th strongly identified with caste, and less than a third agreed to any extent that caste is important to their self-conception. Not surprisingly, the pattern is similar when we asked whether “others” caste is an important factor in the way I think of them”. Here 5 per cent strongly agreed and 25 per cent agreed, again leaving some 73 per cent to disagree (32 per cent strongly). Perhaps more surprisingly, the pattern held up when we asked their response to whether “My caste is an important factor in how others think of me”. Here only 16 per cent expressed any level of agreement, and fully 44 per cent disagreed strongly. One might have thought that caste would have been reckoned more important among the general category of “others” than to themselves. But it seems reasonable to suppose that the students were thinking more of others like themselves. In response to a later question of a different sort, fully 42 per cent felt that only one-quarter or less of “Indians today of (their) age and class” are “proud and observant caste members”.

Data like these need to be handled with caution. It is sometimes hard to know whether such responses reflect deeply ingrained attitudes and behaviour or a temporary ideological gloss. Moreover, scholars of religion and other traditional commitments across generations and through the life cycle know that it is at this very age that students are most likely to rebel against family characteristics. They also know that many of these identifications tend to reappear once students grow older and begin families of their own. In fact, we asked our respondents about this dynamic, seeking their agreement or disagreement with the statement: “Caste was a much more important factor for my parents’ generation in their assessment of self and others”. Here 23 per cent strongly agreed and another 50 per cent agreed, while only 21 per cent disagreed, and another 5 per cent strongly disagreed.

But what about caste’s relationship to religion, especially to Hinduism as perceived by these Hindu students? We asked their reaction to the statement, “It is impossible to be a good Hindu if one ignores caste”. Only 22 per cent agreed with the statement, of whom just 5 per cent agreed strongly. By contrast 30 per cent disagreed and another 35 per cent disagreed strongly. This offers a strong hint of a development that might surprise many Hindu ‘sants’ and classical experts. At least in the view of our somewhat marginal practitioners of the faith, it is quite possible to be a faithful Hindu even if caste is set aside. A strong majority of 66 per cent even disagreed with the Gandhian position that “the caste system would be acceptable without untouchability”, though only 37 per cent agreed to any extent that “Basically I support India’s “reservation” policy”.

But for many of these students, caste is a source of considerable tension because they see their own perceptions at odds with those of the nation at large. Of the 23 statements used in our agree-disagree battery, two claimed the greatest consensus. A full 75 per cent agreed or strongly agreed that “... the caste system is an embarrassing reflection of India’s backwardness as a country”, and another 81 per cent agreed or strongly agreed that “I wish that my grandchildren would have no sense of any caste position”. On the other hand, wishing doesn’t always make things so.

As observers of the society around them, these students described a tension between caste and class. When we asked their reaction to the statement, “In India, one’s caste is more important than the “economic success” one achieves”, almost as many expressed some level of agreement (45 per cent) as disagreement (53 per cent). However, a majority of 53 per cent also agreed that, “Without politics, caste would be virtually non-existent today”. This suggests that they see recent caste-based political appeals as important in fueling the continued salience of caste distinctions.

Despite the antipathy that many feel about caste, we asked about their sense of some traditional practices that many students at this age are confronting. Consider their agreement or disagreement with the following four statements. First, while some 54 per cent disagreed or strongly disagreed that “intra-caste marriages are more successful than inter-caste marriages,” some 39 per cent agreed or strongly agreed with the statement as an empirical proposition, while a higher than normal 7 per cent retreated from the issue by not responding. Not surprisingly, a similar pattern emerged with respect to arranged marriages. Some 45 per cent agreed (14 per cent strongly) that “I would prefer a marriage arranged by my family rather than choosing my spouse independently”, with 31 per cent disagreeing and another 20 per cent disagreeing strongly. But when asked if “it is good to observe caste even if only to confirm one’s family’s wishes”, 35 per cent agreed (7 per cent strongly), while 58 per cent disagreed (13 per cent strongly), and another 7 per cent gave no response. Clearly traditional practices continue to have some appeal, though loyalty to family would trump doubts about caste for only a third of our respondents. Finally, we asked about their sense of caste as a sacred social system by seeking their response to the statement: “Castes have become separate groups that do not work together to form a stable social and religious system”. Some 17 per cent agreed strongly and another 38 per cent agreed, while 30 per cent disagreed, and only 9 per cent disagreed strongly.

What about the students’ religious involvement? We asked them to consider four statements and select the one that best describes their religious behaviour: 27 per cent said they “both participate in rituals and often feel religious”; 13 per cent said they “participate in rituals more than (they) really feel religious”; 32 per cent said they “often feel religious but seldom participate in rituals”, and 24 per cent “rarely if ever participate or feel religious.” Recalling their responses to earlier questions concerning caste involvement, students seem more deeply involved in religion than in caste, though only a little more than one-fourth report both ritual participation and religious feelings – a percentage that is roughly equivalent to the 28 per cent who agreed to one extent or another that caste was important in the way they felt about themselves, and those variables are correlated at 0.15 with a 005 level of significance.

When we asked the students to report as observers of the Indian social scene, and had them choose the most religious among several different categories, 64 per cent selected women over men, and 66 per cent selected older people as opposed to middleaged and younger. However, when we asked about caste, many more selected upper castes as the most religious (45 per cent) than middle castes/OBC (21 per cent), scheduled castes (7 per

Economic and Political Weekly September 2, 2006 cent), or no difference (10 per cent) – possibly because Hinduism confers more honour upon the upper castes. Lastly, while 33 per cent selected high economic class as the most religious and 32 per cent selected low economic class, only 7 per cent selected the middle class and 10 per cent no difference – an interesting clue concerning the emerging middle class at the core of this paper’s concern. If the upper class sees itself as the rewarded custodian of religion, and the lower class sees religion as offering a theodicy of escape in the next life, it is the middle class that is left to develop a more secular “middle way” – to use a Buddhist phrase. Our sample of university students may be sensitive to this perspective since, even at this nascent stage of their careers, they are mostly middle class bound, if not middle class born.

Indexes of PIC and OCS

As a start in seeking out the relationships among these various issues, we need to distil some of the key variables into a more parsimonious set of scales or indexes, each of which combines several items in a single measure of their overall meaning. To this end, we used a simple factor analysis among the 23 items in our agree-disagree battery. The first two factors that emerged in the rotated component matrix reflect two quite different aspects of caste that we presented earlier. The first and most tightly clustered factor involved three items we introduced in the previous section and reproduce in Table 1 with their factor loadings as the components of an overall index of PIC.

Table 2 presents another set of items as part of a second factor reflecting a different facet of caste. Here the shared meaning involves opposition to caste as a characteristic of society as measured by an index of OCS. Three items clustered sufficiently tightly to be included in an equally weighted overall scale, and they are described below with their factor loadings.

Once again, the three items were introduced in the previous section with their separate frequency distributions.

At this point, we have two indices concerning caste – one dealing with the respondent’s PIC and other dealing with the respondent’s more abstract judgment about caste in society and its future. Not surprisingly, the two measures bear little relation to each other with a correlation of -0.079 that is significant only at the 0.167 level. Clearly there is a good deal of slippage between the two indices, as shown in Table 3. Note that we have again rounded each percentage to the nearest whole number, because providing tenths of a per cent would convey a sense of spurious precision with such small numbers. Note too that, although it is rare to take the time and present cross-tabulations for what is basically a zero relationship, some of the cells deserve emphasis. Thus, the single largest cell in the table involves the 79 respondents who score low in their PIC and are high in their societal caste opposition – a combination that certainly seems “rational”. One might expect the next largest cell to be those who score low on both indices, but no. It is instead the upper left hand corner with some 56 students who score high on both PIC and OCS

– and there are almost as many who score medium on PIC but high on OCS. These combinations would seem decidedly inconsistent and non-rational, but together they account for more than a third of our respondents.

In general, it is far more common for respondents to be inconsistent in this way than in its opposite. That is, a substantially larger minority of our members are caught up in caste while rejecting it ideologically as opposed to those who accept it ideologically while not being involved personally. Overall, this is not the stuff of religious or social rebellion.

Relations between PIC, OCS, and Caste, Class and Religion

So far we have provided a descriptive overview of the students’ reactions to a series of issues concerning caste, class and religion separately and examined two indices of caste behaviour and attitudes. But now we want to explore how these indices are related to actual caste position as well as to religious involvement and social class. What are the relationships among these domains, if any? Does class hold the key to the fundamental attitudes and values represented here, as Marxists might suppose? Or are religion and caste independent of class? How do caste and religion relate to each other, or do they? In exploring these issues, we have used responses to three different questions mentioned earlier to measure caste status, religious involvement, and social class respectively. Caste: Earlier we noted that there are several different questions concerning caste available. On the one hand, we asked each respondent to place their jati or varna into one of the general categories of upper caste, middle caste, backward caste, and scheduled caste. On the other hand, having ascertained their jati or varna, we received a more disinterested placement into the same categories from a small panel of social scientists. While there is little difference between the two distributions, we have used the latter here.

Consider the index of PIC. If one were analysing the general population, there are plausible reasons for expecting both a positive and a negative relationship from the perspectives of both high and low castes. From a rational-choice perspective, those

Table 1: Index of Personal Involvement with Caste

My caste is an important factor in the how others think of me 0.794 My caste is an important factor in the way I think of myself 0.753 Others’ caste is an important factor in how I think of them 0.740

Table 2: Index of Opposition to Caste in Society

I wish that my grandchildren would have no sense

of any caste position 0.731 I feel that the caste system is an embarrassing reflection

of India’s Backwardness 0.628 Castes have become separate groups that do not work

together to form a stable social and religious system 0.625

Table 3: Personal Involvement with Caste by Opposition to Caste in Society

Economic and Political Weekly September 2, 2006

of high caste are sufficiently rewarded by their caste standing that they have an incentive to remain involved with it, yet they also have greater access to education and alternative cultures that lean against the caste system. At the same time, those of low caste have ample reasons to jettison the caste system that punishes them, but they lack the cultural exposure to alternative theodicies.

In fact, one’s actual caste status shows virtually no relationship to our subjective index of PIC among Hindu students. The non-relationship is captured by a 0.003 Gamma and 0.002 Spearman’s rho. One might suppose that these zerolevel coefficients mask a curvilinear relationship by which, say, both upper and scheduled castes have high identification at the extremes while those in the middle rank low. But this is not the case either.

What about actual caste’s relationship to attitudes about caste in the society at large and our index of OCS? Here too there is virtually no relationship, with a gamma of 0.027 and a rho of 0.020 – nor is their evidence of curvilinearity. It is true that the signs are reversed for actual caste status’ relation to personal identification and societal opposition, and this suggests a tendency for those of higher caste standing to be more opposed than those of lower. But at this level of statistical significance, such inferences are akin to the proverbial effort to make a silk purse out of a sow’s ear.

Perhaps the caveat with which this section began is governing. It is crucial to remember that our respondents do not constitute a sample of India’s general population. And while these students do come from somewhat disparate social backgrounds, they share a common educational experience whose levelling influence may trump the effects of such background factors as their jati or varna. Meanwhile, what about class and religion? Class: Again as noted earlier, our questionnaire included several factors that may be seen as indicators of class. These include language spoken at home – whether English, Hindi, or a regional vernacular – type of secondary school attended – whether government versus private or convent – and father’s occupation. Because the first two factors are somewhat ambiguous and regionally varied in their class implications, we used only father’s occupation here. For purposes of this analysis, we combined its response options into three categories: professionals and executives, white collar, and farmer or labourer.

Starting again with the respondents’ PIC, we find at least small negative correlations that are both statistically and plausibly significant. Here the gamma is –246, while the rho is –0.194. That is, the lower one’s father’s occupational status hence family class standing, the higher one’s level of caste identification. Perhaps the most plausible explanation for this relationship is that low caste standing tends to produce low class standing in the main, and negative caste effects are most likely to translate into stronger – though not necessarily more “positive” – caste identification. Of course, even correlations of this relatively low magnitude constitute a fragile skeleton upon which to heap much interpretive flesh.

This precariousness is underlined by the relationships with our index of OCS. If class has at least weak effects for caste identification, here the effects drop away to near nothingness. With a gamma of – 0.067 and a rho of – 0.052, it appears that class has little impact on opposition to caste. If there is any significance to these very small coefficients, they suggest that under-caste resentment is more likely to produce caste opposition that does higher caste liberalism. But again, the operative phrase should be caveat emptor. Religion: Finally, what about one’s form of religious involvement, if any? Recall that we asked all our students to place themselves in one of four categories concerning whatever religious tradition they identified with: (i) “I both participate in ritual and often feel religious; (ii) I participate in rituals more than I really feel religious; (iii) I often feel religious but seldom participate in rituals; and (iv) I rarely if ever participate or feel religious”. With this as our basic measure of religiosity or religious involvement, we converted the four categories into three by combining the middle two.

Perhaps not surprisingly, our Hindu students produced a positive relationship between religious involvement and caste identification, albeit a slight one. A gamma of 0.137 and a rho of 0.110 indicate at least a faint tendency for those who are most religiously responsive to also be more engaged in caste-oriented thinking and behaviour. To briefly summarise the cross-tabulation between the two variables at the table’s extreme corners, 26 per cent of those who were both ritually active and felt religious scored high on caste identification, while only 11 per cent scored low. On the other hand, 22 per cent of those who neither felt religious nor participated in religious ritual scored low on caste identification compared to only 6 per cent who scored high. Still it is clear that religiosity alone accounts for little caste identification – and vice versa.

As for our measure of OCS, the correlations are at roughly the same level but move in the opposite directions. Here the negative gamma is -0.181 and the negative rho is -0.138. Thus, those students who are low in religious involvement tend to be high in caste opposition, and conversely those who are high in religious involvement tend to be low in caste opposition. Again, the numbers are small and the relationship between the importance of caste and religion is slight though plausible.

Conclusions

It is worth bearing in mind that the aforementioned findings do not reflect India as a whole, and for three reasons. First, this is a beginning study using a relatively small sample, and a larger replication is needed. Second, as important as a group of young adult university students may be, they are often more the exception than the rule and hardly constitute a reliable representation of the population as a whole. As they mature, they will mellow and “regress towards the mean” of the country overall even if they never quite reach it. Third, this is a sample that we selected precisely because they were atypical as part of a newly emerging middle class that may signal new directions for India. In sum, while it would be exciting to extrapolate such relationships to the Indian population at large, part of the excitement would stem from the irresponsibility involved.

Cultural change – especially seismic change – is not quickly produced or readily absorbed. At least this is the lesson here for anyone predicting a radical “caste-ing off” of the social and religious lines of caste and Hinduism. This research could scarcely have focused upon a more likely group of rebels. Our respondents were largely young adults in brief suspension between adolescence and parenthood, students in three of

Economic and Political Weekly September 2, 2006 India’s most prestigious universities, and students marked as middle-class by their backgrounds, their aspirations, or both. Their age, education, and social class would all seem to predict the kind of cultural upheaval that so many other societies have experienced through the generations. Even so, the results are far less foreshadowing.

It is true that a number of our respondents seem to combine lower scores on what we have called index of PIC with higher scores on our index of OCS. But lacking comparable data for other Indian sub-groups, it is impossible to know if the tendency among these students is actually greater than among other categories of the population, or as strong as studies of generational change would predict. What we do know is that the number of our respondents who bear these traits is actually less than the number manifesting the opposite combination. Indeed, our single largest category – though not by much – involves those who tend toward the high end of the distribution in their PIC and tend toward the low end in their OCS. Finally, it is also striking that only a slightly smaller number scores low both indexes, and there is even a small number who score high on both – i e, who behave in caste terms and oppose caste in principle.

In the words of the 19th century American transcendentalist, Ralph Waldo Emerson, “a foolish consistency is the hobgoblin of little minds,” and one might add, of little cultures. The greatness of Indian culture is perhaps reflected in our finding that some of the very aspects of it that people oppose in the abstract, they cannot escape in their everyday lives. It is worth recalling that, when we asked our young respondents what they wanted and what they predicted for their grandchildren, there was near consensus on a disappearance of caste. Other patterns emerging from these data suggest that this is highly improbable.

A number of questions await a more extensive survey plus important qualitative interviews to flesh out our quantitative findings. First, insofar as caste is becoming detached from Hinduism and seems increasingly able to stand on its own as a principle of social hierarchy, what are the implications? Second, insofar as there are increasing secular spaces in Indian society that are removed from both caste and religion, to what extent is this restricted to the urban middle class instead of radiating through the country as a whole, and what are its ramifications for the culture and structure of India overall? Third, how reliable is the suggestion here that caste is retaining its importance in intimate matters of family and kinship while it is losing its influence in the economic realm, i e, retaining its cultural capital while losing its social capital? But fourth, insofar as caste is separated from even some of its traditional cultural meanings, it might take on new meanings for new generations and different meanings for different caste groups? Fifth, isn’t it possible that the same caste will have different meanings and different significance in different regions as, for example, jats may be dominant in Haryana or Punjab but sub-dominant to the rajputs in Rajasthan. Finally, insofar as class is beginning to exceed the importance of caste in some areas, what are the consequences for the casteification of “other backward classes”?

All of these are pressing issues for the transformations of India that have already begun. Fortunately, just as India will not be alone in experiencing these or similar changes, we will be far from alone in addressing them. Caste is at the critical core of the Indian case, but there are comparable issues at stake in other changing societies, and comparable research being undertaken concerning them.

EPW

Email: demerath@soc.umass.edu jodhka@mail.jnu.ac.in

[We are grateful to Dipankar Gupta, Sujata Patel, Aparna Rayaprol, Murli Dhar, G Nagaraju and Sajja Srinivas for their help at various stages of this work. Usual disclaimers apply.]