A short note about crime surveys
The most commonly recognised sources of crime data are:
i) the official Home Office Criminal Statistics, which takes its recordings from figures of arrests, charges, convictions and prison sentences/populations, and
ii) the British Crime Survey, which obtains data from crime victims and reportings.
Both versions have severe disadvantages.
In respect of the first, criminologists and politicians have argued that the methods of obtaining the data are routinely manipulated in order to achieve desirable findings and meet targets; for example, by concentrating police activity on particular offences, areas and groups. This will inevitably mean that successful prosecutions and prison populations are representative of this activity. There are also very convincing arguments that politicians can manipulate figures by literally redefining crime; what was previously considered a serious violent crime is redefined as an anti-social or disorder offence, which decreases the likelihood of such offenders reaching the prison population, or indeed receiving concentrated police attention in the first place.
The BCS, although widely respected, also has a major limitation - that victim data is not truly representative; for example, unreported crimes are not included, which can obviously have a severely impairing effect on the figures. Arguably the most under-reported crimes (rape, spousal assault, bullying) are often among the most violent, and the most heavily reported are those involving the loss of property. Furthermore, reporting trends tend to vary culturally and geographically. It is always necessary to question the context of any population sample taken for survey, and the author's desired hypothesis prior to the survey.