Finding truth in Karnataka's caste census beyond the challenges

The Karnataka Social and Educational Survey, or the caste census as it is popularly known, has come up against two challenges. There is an internal problem of the training of enumerators and an external one of the rebellion of the privileged. These challenges will affect the quality of the data the survey generates but an appropriate interpretation of the data could still ensure that the exercise serves its purpose.

The inadequate training of the enumerators is evident in the way the questionnaire was canvassed. Without adequate training the enumerators did little to control their bias. They typically rushed to assume answers rather than wait for the respondent’s reply. They often assumed that the caste of the wife would be the same as that of the husband. While this may have been true in most of the cases, the assumption ensured the survey will not capture the extent of inter-caste marriages. This data would have been useful as an indicator of a decline in caste practices.

The second, and in some senses more serious, challenge has come from sections of the most privileged in the state’s society acting in ways designed to distort the results. Prominent families have refused to participate on the grounds that they are not backward. It is interesting that some of those who made their fortunes through the sale of knowledge are not aware that an accurate picture of backwardness can only emerge from a comparison with the forward sections of society. The poor in the United States are defined in contrast to what the middle class and the rich in that society have. This is what leads to their being considered poor even if they have assets that the middle class in South Sudan or Afghanistan cannot dream of owning. By opting out of the survey the privileged are ensuring that any internal comparison between the richest and the poorest would be distorted. 

The knee-jerk response to these deficiencies would be to question the validity of the survey and dismiss its results. This would, however, be a case of the best being made an enemy of the good. The value of the survey is best seen in relation to what exists. As of now there is very little scientific data on caste. The last census of caste was nearly a century ago. And yet caste plays a significant role in Karnataka’s public policy. This is formally recognised in the case of reservations, but it can also play a role in informal narratives around public policy. The absence of data allows for doubts in the public mind about actions that may, in fact, be perfectly legitimate. Is, for instance, the allotment of large tracts of land for the knowledge industry biased in favour of castes that have traditionally had a monopoly on knowledge?

There may be a temptation to fill in some of the blanks left in the data by the privileged abandoning the survey. It would be tempting to assign castes to households that have refused to participate but are headed by persons who have attached their castes to their names or have names associated with specific castes. Such exercises are, however, far from perfect and could seriously distort the data.

It would be much more scientific to let the data emerge as it will but interpret only those results that can be expected to have a higher degree of accuracy.

Such an interpretation would be sceptical about the results that could be distorted by the inadequate training of enumerators. The credibility of data on questions where the enumerators tended to transfer the features of the head of a household to all its members would be limited. This was not true of all intrahousehold data. The enumerators have not treated the education of individual members of a household as being identical to that of the head of the household. The data can then be used to capture gender differences in education levels within the household. There is thus a need for discretion in deciding which intrahousehold data to use, rather than a complete acceptance, or total rejection, of all data about conditions within a house. 

Beyond the difficulty with inadequately trained enumerators, not all the questions that were asked were designed to receive accurate responses. Even with the best of intentions, it is unlikely that respondents would have provided an accurate estimate of household income. Calculating the entire income of a person from multiple sources, from a job to bank deposits, is a time-consuming exercise which neither the enumerator nor the respondent typically had time for. There is also a tendency to underreport income in surveys where respondents believe they would benefit from reporting lower incomes. This risk is compounded in surveys like the present one where communities would like to present themselves as more backward and there has been an organised effort to tell their members how they should respond. There is thus a case to ignore income data generated by the survey.

Economic status

It is fortunately possible to get a better sense of the economic status of a household through the questions on assets. There is a fairly detailed list of both movable and immovable assets in the questionnaire. Attributing accurate economic values to these assets will be a challenge. The pucca house that is reported could be a one-room structure in a poorer part of the state or a palatial building in an expensive part of Bengaluru. What can be done, though, is to define levels of households. A household that owns a car is likely to be better off than one that owns only a two-wheeler which, in turn, is likely to be better off than one that owns only a cycle. It is then possible to develop an indicator of the distance of the household from absolute deprivation. A well-defined indicator of this distance would be more useful in determining levels of backwardness than going by the response to the question on income.

The rebellion of the privileged could also be countered by focusing on the information that is provided by the refusal to participate in the survey. The refusal of a sitting Rajya Sabha member to participate on the grounds that her family is not backward is itself a statement of economic status. The spatial pattern of those refusing to participate strengthens the view that the refusal is predominantly by those who consider themselves to be the elite. A large concentration of refusals to participate is in Bengaluru, and Bengaluru urban accounts for nearly 40% of the state’s Gross State Domestic Product (GSDP). The pattern of refusal is evidence of the urgent need to examine regional inequalities in the state not just in terms of backward regions but also in terms of the excessive dependence of Karnataka’s economy on a single district. 

The information provided in the survey, including the refusal to participate, can be useful in understanding the dynamics of backwardness in Karnataka, as long as we interpret the results in a way that allows us to hear all that the data is telling us, not more and not less.

(Narendar Pani is JRD Tata Chair Visiting Professor at National Institute of Advanced Studies, Bengaluru)