Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
We present evidence that the use of police statistics as a source of crime data can seriously bias empirical tests of the model of deterrence. We use data for 21 areas in England and Wales in 2001-8. In addition to police-recorded crime data, we use victim-reported crime data from the British Crime Survey that are unaffected by changes in public reporting of crime and police recording of crime. We find that the estimated effect of the number of police on recorded and victim-reported crime is similar for property crime but different for violent crime. Our findings suggest that higher numbers of police not only reduce crime rates but also increase the share of crime, and in particular violent crime, that finds its way into police statistics. The resulting estimation bias is found to be large.