This very interesting paper by Meghana Eswar and Bejoy K Thomas in Economic and Political Weekly reports on some important discrepancies in the census data on access of households to water for domestic use. There are three main findings.
First, though village amenities and house listing data sets (both of which are part of the census) provide data on rural drinking water sources, house listing data, which are collected from households, appear to be more reliable than village amenities data, which are reported by the village accountant/panchayat. Second, census data on access to taps for drinking purposes are not a good indicator of piped water coverage, since this data may be influenced by the water supply situation (insufficient supply and quality) at the time of census data collection. Third, the role of the enumerator and respondents/households is crucial in census data collection. The data on access to treated water may not be reliable, since households may not be in a position to tell if the water supplied to them is treated or not.
The full paper can be seen here.