The Pulse

What Attempts to Measure India’s Caste System Get Wrong

Recent Features

The Pulse | Society | South Asia

What Attempts to Measure India’s Caste System Get Wrong

Modern surveys on caste can feel like inadvertent replications of a colonial mistake.

What Attempts to Measure India’s Caste System Get Wrong
Credit: Depositphotos

The Washington, D.C.-based Pew Research Center recently released a landmark survey on religion and society in India. Administered to 29,999 respondents in 17 languages between November 2019 and March 2020, the survey is certainly a formidable endeavor. The final report provides fascinating insights into the coexistence of religiosity, pluralism, and secularism in modern India.

When it comes to analysis of caste, however, this survey disappoints. Pew argues that “India’s caste system, an ancient social hierarchy with origins in Hindu writings, continues to fracture society.” But caste has been defined narrowly and interpreted broadly. The assumptions are far removed from both lived experience and academic understandings of how caste really works in India. This is unfortunate – history teaches us that how we define caste, and design surveys to ask people their caste, can inadvertently create feedback loops that intensify social divisions.

At first glance, Pew’s statement is well-supported by the specific numbers. Nearly all Indians (98 percent) identify with a caste, regardless of whether they are Hindu, Muslim, Christian, Sikh, Buddhist, or Jain. About one-quarter (24 percent) say all their close friends belong to their caste, and 46 percent say most of their friends are from their caste. Surprisingly, 82 percent of Indians say they have not personally faced caste-based discrimination, and less than a quarter see evidence of widespread discrimination against disadvantaged communities. But as Pratap Bhanu Mehta reminds us, these numbers could mean that India has “not have even progressed from exclusion to discrimination.”

While most Indians would agree that caste is divisive, the survey seems to have been set up to get that result. All respondents were asked whether they are “from a General Category, Scheduled Caste, Scheduled Tribe or Other Backward Class.” Self-identified responses to this question were carefully coded. The methodology appears to have been adapted from Pew’s surveys of racial divisions in the United States. These specific categories are official labels that come from the Indian state; they do not correspond to identity in daily life. Choices of neighbors, friends or marriages are not based on these identities.

What question should the survey team have asked instead? Well, that’s complicated. The definition of caste is deeply contested in modern India. There is actually no word for caste in any Indian language. The word “caste” is actually a derivative of the Portugese word “casta.” It was applied to India by Portuguese travelers in the 16th century.

Caste in real life is complicated. Most Indians viscerally understand caste identity as their “jati” of birth. This is a Sanskrit word with equivalents in many local languages that refers to the 3,000-plus closed endogamous communities with distinct occupations, property ownership patterns, diets, and lifestyles. Historically, each jati was a part in an interdependent and non-competitive system of labor. While jati identity is inherited, the status of a jati varies spatially. In practical terms, this means that in a typical village, there are several jatis that fit into a hierarchical structure where some groups have faced severe discrimination for generations. But there is no pan-Indian system of ranking them. Empirical research confirms that there is variation within and variation across jatis.

When diverse people place themselves into just a few discrete categories that are presented in a clear hierarchy, it is difficult to infer anything about the subtle processes of exclusion and discrimination that operate in real life. Drawing further inference about the lingering impact of archaic religious practices is even harder. Pew’s statement about current caste divisions being an “ancient hierarchy” with roots in “Hindu writings” is thus quite puzzling. The authors seem to be referring to the ancient varna scheme – the idealized social order described in the Rig Veda, a Hindu text.  But most scholars agree that there is much more to caste than an ancient religion.

Pew’s definition and methodological approach seem to actually have far more roots in India’s colonial era than antiquity. Scholars like M.N. Srinivas, Nicholas Dirks, Susan Bayly, and others have argued that in the 19th century, the British ruled India by subsuming the plethora of castes all over the country into a unified framework with the explicit goal of controlling the country. The legal system, for example, featured separate laws for different religious groups. For Muslims, Shariah or Islamic law was adapted. For “Hindoos,” however, the enormous diversity of prevailing beliefs and ways of living presented a challenge. So administrators commissioned the translation of Sanskrit texts (including the Rig Veda), elevated them to canonical status, and used them to derive laws. Although the British did not create caste, a rigid caste hierarchy – one that bore an uncanny resemblance to feudal England – emerged under their watch.

Ever since colonial times, large-scale surveys have had the power to create and reinforce social identity. Surveys have changed over time. In the early years of British rule, surveys were used to measure the racial, anthropometric, and socioeconomic characteristics of India’s population. By the time of the first census of 1881, complex jati identities were simplified and placed in a strict hierarchy. By 1935, broad groups such as Scheduled Caste and Scheduled Tribe created new forms of political representation in the waning years of empire.

Against this backdrop, modern surveys on caste can feel like inadvertent replications of a colonial mistake. The last British surveys were more than 80 years ago, but Indians are still sensitive about being asked their caste. Recent experiments by economists  demonstrate that the very act of asking people (even kids) about their caste prior to completing a cognitive task quickly changes behavior, with individuals from disadvantaged groups displaying a discernible drop in performance. This suggests that asking questions about caste creates powerful feedback loops that ultimately intensify divisions.

To be fair, Pew’s caste question and associated coding methodology may have been deliberately kept somewhat similar to the Indian Census and other government surveys to minimize fallout. But the Indian government does not seek to study whether Indian society is divided along caste lines – it already knows that it is. Since the Indian Constitution bans discrimination on the basis of caste, it cannot easily ask people that question. Earlier this year, Union Minister of State for Home Affairs Nityanand Rai said in Parliament, “The Government of India has decided as a matter of policy not to enumerate caste-wise population other than SCs and STs in Census.”

Some recent surveys, such as the India Human Development Survey, have balanced the use of discrete government categories with a rich set of additional questions about marriage, work, and social lives of respondents. These have yielded excellent insights into the workings of caste in modern India. These studies however, rarely made the kinds of headlines that Pew is getting in the international press.

All in all, it should not surprise anyone that a survey question that reduces identity to simple discrete boxes indeed finds evidence of societal divisions. Pew’s surveys in the United States routinely reach similar conclusions. A 2015 survey for example, codes race into discrete categories and reports that more than 70 percent of Americans choose their close friends from their own race.

What we need right now is real insights into specific processes of discrimination, prejudice, and exclusion that operate in divided societies like the United States and India. Getting at those mechanisms requires survey questions that are derived from lived experience as well as academic knowledge of history, religion, and society.