Schooling, early marriage and cooking fuel
Each dot is a rank correlation inside one state, between each district's rate and one thing about the district measured by the National Family Health Survey in 2019-21. A value above zero means the districts with more of that thing lost more names. A value near zero means no relationship. The note under the chart gives how large a correlation has to be in each state before chance becomes an unlikely explanation.
Rank correlation between the rate and three district circumstances, state by state
Registered deaths
The survey asked households about deaths in the three years before it and whether each was registered. A district where most deaths are registered has a paper trail an officer can check a name against.
Rank correlation between the rate and the share of deaths registered
Towns and crowding
The same correlation with the share of people living in towns and the number of people per square kilometre, both from the 2011 Census, the last count by district. These are read on the districts of 2011, as on the urban page, and cover the ten states that page covers.
Rank correlation between the rate and urban share and density, 2011
Income per head
The same correlation with income per head, from each state's own table of district income. The tables are for different years and prices, but only the order of districts inside one state is used, and that does not depend on either. Income is read on the districts of each table, so Mumbai City and Mumbai Suburban count as one, and Thane with Palghar, as Maharashtra's table prints them.
Rank correlation between the rate and income per head
Why these cannot be pulled apart
Every correlation, state by state
How the figures were put together
The district circumstances are from the National Family Health Survey 2019-21, the fifth round, which published a fact sheet for every district. Every value here was read from the district pages of each state's Compendium of Fact Sheets, published by the International Institute for Population Sciences and the Ministry of Health, in copies held by the OpenCity data portal. All twelve states with district figures on this site are covered.
The survey used the districts of 2019 and 2020. A district made after that is counted under the district it was made from, and the districts sharing one survey district are added together before the rate is worked out. Bengaluru's four districts are counted as the survey's Bangalore, Kolkata's two as Kolkata, Vijayanagara under Bellary, and Mulugu and Narayanpet under the districts they came from. Telangana's Warangal and Hanumakonda are matched to the survey's Warangal Rural and Warangal Urban, which they replaced with changed boundaries in 2021. Rajasthan's districts made in 2023 are counted under the district each was mostly made from, so Beawar is counted with Ajmer and Kotputali-Behror with Jaipur. Tamil Nadu's six districts made since 2019 are counted under theirs, Chengalpattu with Kancheepuram for one. Delhi is counted on the eleven districts of its 2025 roll, which are the survey's.
Each state is read in its own measure, as on the urban page. In Gujarat, Karnataka, Maharashtra and Telangana it is the names left off the draft per 100 on the roll before, and Telangana's district figures are the state's as reported in the press. Tamil Nadu, Uttar Pradesh, Delhi and Uttarakhand are measured the same way. In Bihar it is both phases per 100 on the 2025 roll. West Bengal and Kerala published no removals below the state, so there it is the fall in the roll, which takes in ordinary deaths and moves too. Rajasthan published its removals for the state only, so there it is the fall in the roll from the 2023 election to the final roll of 2026. A rank correlation is the ordinary correlation worked out on the districts' ranks, so one very large district cannot drive it. The size it must reach before chance becomes an unlikely explanation is 1.96 divided by the square root of one less than the number of districts. Kerala has 14 districts, so there a correlation must pass 0.54.
Income per head is from Bihar's Economic Survey 2024-25 (2022-23, constant prices), Karnataka's Economic Survey 2025-26 (2024-25, current prices), Maharashtra's Economic Survey 2025-26 (2024-25, current prices), Tamil Nadu's district income handbook (2023-24, constant prices) and Telangana's Socio-Economic Outlook 2024 (2022-23, current prices). Bihar's table was read from the Survey itself. The other four were read from copies transcribed by the OpenCity data portal, and the district totals in each were checked against the state total the same table prints. No district table was found for the other states.
Every figure on this page is computed from the published tables when the page loads. The district table downloads as CSV.