Criminology Research Council grant ; (3/78)
This research project concerned the feasibility of estimating true criminal offence rates (the sum of recorded and unrecorded rates) via an indirect statistical (econometric) method. The method involved specification of a model of the generation and control of crime, which incorporates a model of the crime recording process. In the complete model, it is postulated that the actual number of offences (of a given type) committed per period is largely determined by the level of enforcement (affected by the size of the police force), by the types and sizes of sentences imposed on those convicted, and by various socioeconomic factors (level of unemployment, population density, etc.). It is also postulated that the proportion of actual offences recorded is largely determined by the resources available to and workload of the police: the less the workload of the police, the more likely are they to record petty offences, and to investigate reports (to ascertain Whether or not the reports of offences are likely to be valid); also, a lighter workload makes the police more 'accessible' to the public and is likely to encourage victims to report offences.
It is possible to summarise these notions in a statistical model, then to manipulate the model so that its parameters can be estimated by applying standard statistical (econometric) procedures to published data. Most importantly, by following this procedure, it is possible in principle to infer the true offence rates for particular crimes, or to infer the ratio of the change in the true rate to measured changes in the recorded rate.
In practice, the procedure itself throws up a number of acute statistical problems which seem impossible to resolve. The problems are complex, but revolve around the point that for the procedure to be capable of yielding the type of estimates required, a very limited range of options for mathematically specifving the complete model are available, and each of the options themselves throw up other statistical problems. A test of some of the most appealing options led to disappointing results which could not be sensibly regarded in any way as 'reasonable' (using extraneous information-from victimisation surveys for example-to assess 'reasonableness').
The main conclusion of the study then is that the econometric technique is not likely to offer a cheap and practical alternative or complement to other methods for estimating the amount of unrecorded crime. However, some further work in this area will proceed, using a more disaggregated approach, in the hope that something useful can be salvaged.