Punjab conducted a door-to-door census of farmer and farm labourer suicides through its agricultural universities, covering the period from 2000 onward, village by village, which very few Indian states have attempted.
It recorded several thousand deaths over the survey period, concentrated in the Malwa districts of Sangrur, Mansa, Bathinda, Barnala and Moga, with debt named as the principal factor in the large majority of cases.
Farm labourers, mostly Dalit and landless, were a substantial share of the deaths and are frequently left out when the figures are quoted.
The national crime statistics record a much lower figure for Punjab, because of how cases are classified and reported by the police, and the discrepancy between the two counts is itself the finding.
The state announced compensation for families of the recorded dead. Disbursement has been partial.
A survey that walked into every village and asked is a better instrument than a police return, and Punjab has one and its results are not what the national figures say.
The work was divided between Punjabi University Patiala, Punjab Agricultural University Ludhiana and Guru Nanak Dev University Amritsar, each taking a set of districts, with investigators going into every village in the assigned area and taking statements from households and from the panchayat.
Its weakness is that it depends on what a household will say to a stranger holding a form, and that varies from village to village. Its strength is that somebody went to the village at all, which no other instrument in this country does.
It has not been repeated at that scale. The survey covers a fixed window, and for the years since it closed there is no comparable count, so what has happened after it is known only through the returns the survey itself showed to be wrong.