The survey data.
The anonymised responses of 401 informal workers, the codebook that explains every column, and the log of every change made while cleaning the data.
Three files, all CSV.
Every column
The question, the type of answer and the allowed values for each column.
Licence: the dataset, codebook and cleaning log are released under Creative Commons Attribution 4.0 (CC BY 4.0). You may copy, share and adapt them, including commercially, if you credit Act Now India and PinPoint Ventures. You may not try to identify any respondent.
Cite as: Act Now India and PinPoint Ventures (2026). Four hundred and one workers: Katihar survey report 2026. The report is at the report page, and an interactive view of the same data is in the explorer.
How respondents are protected.
Names, fathers' or husbands' names, addresses, ward numbers, enumerator names, paper form numbers and interview dates are not in the file. Religion and education were removed because, combined with the other columns, they point to individuals in villages this small.
Age is grouped into three bands (15 to 35, 36 to 45, 46 and over) and social category into three groups, with Other Backward Classes and General combined. Every combination of panchayat, gender, age band and social category holds at least five respondents. Where a combination was smaller, the age band and social category were replaced by "Suppressed" (for 9 respondents the panchayat and gender were also replaced). Answer categories chosen by fewer than five respondents in a column were replaced by "Suppressed". Rows were shuffled and given a random number.
This lowers the risk of identifying a person. It does not remove it, because a combination of many answers can still be distinctive to someone who already knows a household well. Please do not try to identify respondents, and do not link this file to other data about the same households.
What the file cannot do.
The sample is purposive, not random. The field team chose respondents in the wards where Act Now India works, so the figures describe these 401 people and not the district.
The cleaning log lists each cleaning rule. Because education and religion are not in this file, the regression models in the report cannot be reproduced from it. Researchers who need those columns can write to actnowindia01@gmail.com.
The coding rules for the two open-ended questions (Q37 and Q38) are not published with this file. The theme columns show the result of that coding.