A new study examining conflicting trends in official violent crime stats reveals that the leading measure in England and Wales underestimates the level of violent crime - and the findings challenge official confidence in a downward trend.

The study’s authors, from Lancaster University and Royal Holloway, University of London, argue that too much confidence is placed in the Crime Survey for England and Wales’ conclusion that violent crime is in decline, and that improvements in police statistics mean both datasets should be published in parallel to give a more rounded picture on violent crime.

Their paper published today by the journal PLOS One, examines why the two leading measures of crime, the Crime Survey for England and Wales (CSEW), which is produced by the Office for National Statistics (ONS), and police recorded crime data, show conflicting trends in violent crime.

The CSEW is currently recommended by the Office for Statistics Regulation as the best source for violent crime statistics.

The ‘main part’ of the CSEW, used by the ONS to determine trends, suggests a declining trend for violent crime in England and Wales. This dataset has been used by the ONS to conclude that violent crime is decreasing.

However, at the same time, police recorded crime statistics, which is recorded by police forces across England and Wales, show an increase in violent crime reported to the police.

The researchers investigated these two main sources of statistics on violent crime to explore why they paint very different pictures.

“The findings from the Crime Survey for England and Wales that shows violent crime to be in decline contrasts sharply with the data recorded by the police, which is showing an increase,” said Brian Francis, Professor of Social Statistics at Lancaster University and lead author of the study. “It’s important for a wide range of reasons that we have an accurate picture of violent crime in Britain. Our study shows why the datasets differ and why we need to question the reliance of the UK Office for Statistics Regulation on a single dataset for violent crime.”

The two datasets gain their information in different ways: the police data is from those reporting crime to them, and the survey from a sample of the residential population. The two datasets also measure violence in different ways, with some violent offences, such as stalking, excluded from the survey.

In order to make a statistically valid comparison the researchers aligned both series to measure the same narrower set of violent crimes (assaults) in England and Wales going back to 2010 and limited the analysis in both sources to crime reported to the police.

The comparison of the aligned datasets shows that in 2022-23 there were nearly 1,300,000 reported assaults reported to the police, compared with just over 400,000 reported assaults estimated by the Crime Survey- more than three times as large.

They found that the two diverging trends still existed in the aligned datasets. The recorded police crime figures show an upward trend in violent crime from 2010 to 2022-23, whereas the Crime Survey of England and Wales estimates a downward trend. The cross-over where the police figures begin to show higher rates of violent crime occurs in 2015/16.

The researchers examined possible reasons for why the two datasets are trending in different directions.

They conclude that improvements in police accuracy in recording crime explains part of the difference. The researchers believe differences in data-collecting methods offers other explanations.

Researchers believe other key reasons for differences in trends shown by the datasets is that the survey population is narrower than in police records because the CSEW excludes vulnerable groups such as children, the homeless and those in temporary accommodation like hostels and refuges.

These exclusions from sampling of vulnerable groups by the CSEW as well as the recent reduction in the survey’s response rate to 42% also concern the researchers.

Sylvia Walby, Professor of Criminology at Royal Holloway, and co-author of the study, said: “Our assessment of the methodology is that the quality of the data on violent crime in police recorded crime has improved while that in the Crime Survey for England and Wales has declined.”

The researchers also believe that the CSEW is significantly underestimating violent crime in Britain.

The survey shows that 40.9% of respondents who were victims of crime reported it to the police. In 2022-23 the police reported more than 2.1 million incidents of violent crime. Given the 40.9% reporting rate shown by the crime survey, the researchers estimate that the number of violent incidents of crime in Britain that year is likely to have been more than 5.1 million.

This is more than double the number of police recorded violent crimes and more than five times the number of violent crimes estimated by the CSEW.

“Based on our estimates, the CSEW seriously underestimates the volume of violent crime in England and Wales”, said Professor Walby. “We believe the main part of the CSEW has always underestimated violent crime, and this has become visible now that police data has improved.”

“Instead of relying on one dataset, data and trends from both the survey and the police should be presented alongside each other to provide a fuller picture of violent crime,” said Professor Francis.

The study is detailed in the paper ‘Conflicting trends in violent crime measured by police recorded crime and the Crime Survey in England and Wales since 2010’ which is published in PLOS One.

DOI: https://doi.org/10.1371/journal.pone.0324272