Digging into the data on crime and punishment

Siddhartha Bandyopadhyay
Friday, 15 January 2016

This is the first of four pieces we are publishing, the product of a partnership with the University of Birmingham, discussing correlations between police recorded crime, criminal justice activity and social factors.

Crime policy is never far from the forefront in public debate but the conversation often seems to occur in parallel worlds.

Searches on whether prison works for example reveal diametrically opposing views, from prison really works to prisons never work. And each side conveniently finds figures to back up their argument. It is perhaps no wonder that there is such a distrust in the mind of the general population about statistics, (lies, damned lies and statistics).

Yet, careful analysis of the data is one of the best ways to understand the debate. And that means not just looking at raw figures and making up a story that suits one’s ideology but careful statistical modelling that considers major factors that can affect crime.

That necessarily includes understanding the effectiveness of prison but also alternatives to prison. It must also include other factors that can affect criminal behaviour, such as policing activity (summarised by detection rates for example) as well the socio-economic factors such as employment opportunities and the distribution of wages that are widely thought to affect criminal behaviour. Putting them together into a model allows us to understand how each affect crime rates.

Raw figures can be misleading, as it masks the differences that arise between, say, those who go to prison and those who don’t, by not accounting for differences in say socio-economic status between the two groups. Such differences, rather than whether they have or have not been given a prison sentence, may explain differences in behaviour of the two groups. This is something that will be overlooked if we simply look at raw data.

Interesting patterns

Using a large dataset on recorded crime in England and Wales over a 20 year period, we found some interesting patterns.

First, apart from violence against the person, detection rates have a crime lowering effect for all crime categories. Policing activity seems to matter.

However, prison does not fare too well in the analysis. Broadly speaking, prison does not do much to reduce crime, though when one considers variations in sentencing length, there is some impact for longer sentences. It fares worst against young offenders.

Does this mean non-custodial sentences work? In this case the answer varies: on juveniles, non-custodial sentences do not seem to work either, but for adults it seems to be effective in reducing violence against the person and sexual offences.

Does this hide more than it reveals? For sure, it cannot answer the question around the effectiveness of either conclusively as looking at numbers on either type of sentence cannot answer questions around its quality: how effective is probation for instance or how well supervised is community punishment. Yet, as it currently stands while non-custodial sentences do not have a huge impact on crime, neither does prison. And this is particularly true for young offenders.

The results around youth offending are of particular note and makes us realise we must move beyond debating how effective are post-crime interventions. If being in contact with the criminal justice system increases criminal activity, we need to look at how to prevent youth from coming into such contact in the first place.

The broad criminal justice model that criminalises youthful activity must be reconsidered, along with a more positive agenda that can help divert youth from pursing a criminal career. There is scope for fruitful debate and evidence based research of what such an agenda will look like. More contact with the criminal justice system does not look like the answer though.

Dr Siddhartha Bandyopadhyay is a senior lecturer in economics at the University of Birmingham