(Well, a quotation from a larger work, not his entire assessment.)
Their validity is threatened in the sense that reported figures are not one-to-one reflections of events; the number of burglaries ‘known to the police’ in the official parlance do not equal the number of burglaries which have taken place. These statistics actually reflect the interaction between three sets of activities: thing which go on ‘out there’ in the environment (crime), the responses of those who are victims (reporting), and society’s effort to discover and record it (policing). The frequency of these activities vary across communities over time, and the mix reflected in any given set of crime statistics is problematic.
Skogan, Wesley G. (1974) “The Validity of Official Crime Statistics: An Empirical Investigation”, Social Science Quarterly, 55(1), 25-38.