Violent crime, including knife crime, is the biggest crime problem of recent years, causing serious and long-lasting harm to victims, friends, families and communities. The UK’s new government has pledged to reduce knife crime by 50% over the next ten years. Understanding how to measure violence accurately is critical to achieving this goal.

A new study, published in the International Journal of Population Data Science (IJPDS), reveals that the way we collect data—whether through police records or self-reports—affects our understanding of who commits serious violence and when it occurs, often undercounting violence by those most at risk.

Most information on violence comes from police records, but this only includes crimes reported to or seen by the police. However, most violence is not reported and because police spend more time in higher crime areas, violence data is incomplete and can reinforce discrimination and unfairness. An alternative method, asking people directly if they have committed violence through a survey, can also be inaccurate because people may withhold information or may not take part at all.

Researchers from the University of Bristol and the University of Hull linked data on young people involved in the “Children of the 90s” (also known as the Avon Longitudinal Study of Parents and Children, or ALSPAC), which has been tracking the health and development of over 14,000 people since birth, with local police records. This allowed them to compare what violence people say they have done with the violence that police know about.

They found that police records only included a small proportion of the violence mentioned in surveys and that the surveys also missed some of the violence in police records, because the people most prone to violence tended to drop out of the study. The study also found that undercounting of violence seemed to be most severe in boys in late adolescence, the group and time when risk of violence is highest.

Using a method called multiple imputation to “fill in” missing information, researchers concluded that violence rates in these groups may be higher than previously thought. As violence interventions typically target groups most at risk, there is potential that strategies are not always aligned with the prevalence of the problem.

Lead author, Dr Rosie Cornish added, “Further work could explore whether there are other groups in which this undercounting is a particular problem. Without understanding and addressing the gaps in our knowledge about violence, interventions and strategies risk falling short or reinforcing incorrect assumptions about who and when violence occurs. Adjusting our data on violence to accurately reflect the true risk and patterns of violence is crucial to ensuring the government’s ambitious ten-year target is met.”

 

Click here to view the full article

Dr Rosie Cornish, Senior Research Fellow, Bristol Medical School, University of Bristol

Cornish, R., Teyhan, A., Tilling, K., Macleod, J. and Brennan, I. (2025) “Measuring serious violence perpetration: comparison of police-recorded and self-reported data in a UK cohort”, International Journal of Population Data Science, 10(1). doi: 10.23889/ijpds.v10i1.2391.