Peter Henderson

I’m reading Keir Irwin-Rogers and Luke Billingham’s Against Youth Violence. The first chapter has a good overview of sources of data on the prevalence of violence. It’s not easy to measure how much violence is happening – all the available sources have their strengths and limitations.

Here’s a summary of the sources, their strengths and limitations.

Data source	Strengths	Limitations
Police recorded crime Crimes reported to and investigated by the police.	More serious crimes are more likely to be reported. This is especially true of homicides – most homicides are reported.	There’s good evidence that a lot of violence isn’t reported to the police. There are many reasons why: – Variation in the effectiveness of police recording practices. – A lack of trust in the police and police legitimacy. – Fear of potential repercussions, from the perpetrators of violence or the authorities There is evidence that the following are particularly under-reported: – domestic violence – violence involving younger people (a MOPAC report suggests only 44% of young victims reported violence to the police), especially when they are from a Black, Asian, or minority ethnic background – violence in deprived areas
Hospital admissions data Data collected when people are admitted to hospital for a violence-related injury	There are likely fewer barriers to victims attending hospital if they need medical treatment. Hospital admissions data is likely a good source for relatively serious injuries that require medical treatment.	There is evidence that some young people might avoid hospital if they can, for fear of contact with the authorities. However, this point shouldn’t be overstated; most victims with serious injuries will seek medical treatment. Hospital admissions data won’t capture incidents leading to less serious harm. Many incidents of violence don’t lead to injuries requiring medical treatment, even if the incident was serious (e.g. involving a knife).
Crime survey for England and Wales A survey that attempts to give a nationally representative picture of crime. The survey is a random sample of 36,000 people from randomly selected households. It asks if they have been a victim of crime in the last year.	About three quarters of households approached agree to take part in an interview. Perhaps people who wouldn’t report their crime to the police feel safe telling a survey team about it.	Some of the limitations relate to variation in how different groups of people might interact with the survey: – Sampling bias – perhaps the people who are more likely to be victims of violence are more likely to be in the quarter of households who don’t take part? – Systematic variation in how people report crime. E.g. There’s some evidence that well-educated people are more likely to report minor assaults. – Variation in how different groups of people interpret important terms like ‘violence’ or ‘threaten’. If different groups of people understand these words differently then perhaps variation in how they report violence is caused by this different understanding rather than real differences in experience. Other limitations relate to more general challenges in survey methods: – Memories are not perfect. People might misremember what happened to them. – People might feel unwilling to answer what are sensitive questions – People’s answers might be affected by their desire to see survey results influence a certain policy outcome.