You are using an older browser version. Please use a supported version for the best MSN experience.

Why India’s official crime rate figures are unreliable

LiveMint logoLiveMint 07-08-2017 Dipti Jain

Three decades after it was set up to build a database on crime in the country, the National Crime Records Bureau (NCRB) has become an invaluable resource for policymakers, researchers and the media. Be it on farmer suicides or crimes against women, the national debate has often been enriched with NCRB data. However, a Mint analysis of NCRB’s methodology shows its crime rate estimates are unreliable, and even misleading for some categories of crime.

There are two key problems with the official estimates of crime rates, defined as crimes per million population. First, they rely on outdated population projections. Second, the methodology to compute crime rates is not consistent across years, which renders a simple comparison of crime rates across years meaningless.

Calculating the crime rate requires availability of population figures for each year. Post-2001, NCRB uses population projections made by Registrar General of India for these calculations. These projections are based on the 2001 census. The only exception is 2011, for which actual census population has been used. Comparing annual growth in population used by NCRB shows an intriguing trend. There is an abnormal spike in 2010-11 and a sharp fall in 2011-12. At the state level, the discrepancy between 2010-11 and 2011-12 growth is even bigger.

//datawrapper.dwcdn.net/oN1bC/2/414px

//datawrapper.dwcdn.net/Nl9i8/2/414px

Such drastic changes in population growth take place only in case of high-mortality events such as epidemics and wars. But given that nothing of that sort happened, the only plausible explanation for this volatility is that official projections between 2001 and 2011 were inaccurate. 

This problem is bound to manifest itself in crime rate statistics. If pre-2011 population projections were an underestimate, crime rate figures will be biased upwards. This is because the denominator used was smaller than what it should have been. Given the magnitude of problems with state-wise population projections, no meaningful comparisons can be made at the state level across years. 

To be fair, the NCRB does say that it is using projected and actual population figures in the respective years. However, the use of the outdated population projections has led to artificial changes in crime rates. 

The NCRB has no option but to rely on the population projections based on the 2001 census because the Registrar General Of India (RGI) has so far failed to release the official population projections based on the 2011 census, a person with direct knowledge of the matter said. This person declined to be identified, saying he is not authorized to speak to the media on this issue.

The NCRB may not be willing to use alternative population projections other than what the RGI provides, since this will lead to another round of re-calibration when RGI finally comes up with its new projections. But the use of the old projections, and the abrupt shift in 2011, has seriously compromised the reliability of crime rate statistics, which have been used widely by politicians, policymakers and the media -- including this newspaper.

There is an even bigger problem with crime rates for crimes against Schedule Castes (SCs)/Schedule Tribes (STs). Till 2011, the total population has been used to calculate crime rates in these categories. After 2011, only SC/ST population from 2011 census is being used. NCRB has been using same population for all years after 2011. This has led to a gross under-estimation of crime rates in the years leading up to 2011. Using the growth rate of SC/ST population between 2001 and 2011 to extrapolate SC/ST population till 2015, we find that the official estimates of crimes against SC/ST present a misleading picture of the trend in crimes against SC/STs.

//datawrapper.dwcdn.net/Vv8K2/1/414px

The above analysis suggests that using the NCRB crime rate figures, especially for dis-aggregated categories, is fraught with risks if one is analyzing trends over time. 

More From LiveMint

image beaconimage beaconimage beacon