Caste in the United States

What This Report Is

Equality Labs published Caste in the United States in 2018. It claims to document caste discrimination among South Asian Americans. The survey collected responses from about 1,500 people through the organization’s own website and social media.

What We Looked At

How a study is done determines whether you can trust what it found. The Citation Integrity Dashboard (CID) evaluates methodology — how the research was conducted — not conclusions. We do not judge whether the report’s findings are right or wrong. We judge whether the methods can support the claims being made. We classified this report as a Survey (a study that asks questions to a defined group of people). That classification means we evaluate it on sampling design, question quality, and data access.

What We Found

The biggest problem is how respondents were recruited. Dimension 3 measures case capture and sampling — whether the people surveyed actually represent the population the report claims to describe. This report scored 2 out of 10. Equality Labs distributed the survey through its own advocacy website (castesurvey.org) and its social media accounts. People who have experienced caste discrimination are far more likely to follow an anti-caste advocacy organization. They are far more likely to click through and fill out the survey. The result is a sample that systematically overrepresents the very experience it claims to measure. The report’s headline finding — that 67 percent of respondents reported caste discrimination — describes those self-selected respondents. It does not describe the South Asian American population. Carnegie Endowment’s footnote 29 in its own 2020 survey is the clearest published critique of this problem.

The second major problem is data access. Dimension 6 measures verification standards — whether someone outside the organization can check the data. This report scored 3 out of 10. The raw survey data has not been released. No formal process exists for researchers to request access. No one can independently verify the results. Compare this to Pew Research Center, which maintains an archive where outside researchers can request survey data.

The report also lacks any engagement with criticism. Dimension 8 measures counter-evidence — whether the authors acknowledge limits or respond to outside critiques. This report scored 2 out of 10. There is no limitations section anywhere in the report. When outside researchers raised sampling concerns, published responses from the organization characterized those critiques as motivated by caste privilege. They did not engage with the methodology questions.

Question design created an additional problem. Dimension 1 measures definitional precision — whether the survey’s key terms are defined clearly enough that two researchers would interpret them the same way. This report scored 5 out of 10. The survey items do not distinguish caste discrimination from ethnic or racial discrimination. That conflation makes the 67 percent figure hard to interpret even within the self-selected sample.

The Bottom Line

This report scored 3.8 out of 10. That places it in the Advocacy-Grade band (2.0 to 3.9), which means it functions as advocacy material rather than independent research. A non-compensatory cap was applied — a scoring rule that prevents any report with a sampling score below 3 from exceeding 5.9, no matter how well it performs elsewhere. The raw weighted score before the cap was 5.2. This is the highest-impact report in our calibration set and the weakest methodologically. It was cited in California legislation and Fortune 500 corporate policies. The score reflects methodology only. The report’s conclusions about caste discrimination may be correct — but this survey cannot establish that.