1. A Tale of Two Sources Bringing Together Scotland’s Crime Statistics Trish Campbell, Justice Analytical Services @ SGJusticeAnalys

2. Crime statistics come from two sources Police recorded crime captures crimes that are reported to, and recorded by, the police. The Scottish Crime and Justice Survey, a large survey of people aged 16 and over living in private households.

3. How do we communicate the two sources? The Main Findings of the SCJS are published biennially. Police recorded crime figures are published annually. The annual bulletin presents statistics on crimes and offences recorded and cleared up by the police in Scotland, disaggregated by crime/offence group and by local authority.

4. Why do we need two sources? Police recorded crime and the Scottish Crime and Justice Survey (SCJS) are essential complementary sources of data that, together, present a fuller picture of crime in Scotland. 815,000 273,053 Incidents of crime estimated by the SCJS in 2012/13 Crimes and offences recorded by police in 2012/13 Estimates produced by crime surveys are higher than the volume recorded by the police, emphasising that, for many reasons, not all crimes come to the attention of the police.

5. Strengths and Weaknesses: Police Recorded Crime Covers the entire range of crimes and offences and covers the whole population. Able to present information for lower geographical areas (local authority and national level and at legacy police force area prior to police reform). It is dependent on the public reporting crimes to the police. Can be affected by police recording/counting practices and changes may not reflect changes in underlying crime trends. Does not provide information on the characteristics of victims such as age and gender or the circumstances of the crime, or provide data on public attitudes.

6. Can capture crimes that are not reported to, and therefore not recorded by, the police. Can analyse victimisation against other demographic variables (e.g. age, gender, socio-economic group, tenure status and area deprivation) Able to capture attitudinal data, such as perceptions of crime rates, fear of crime and attitudes towards the criminal justice system. It doesn’t capture whole population (e.g. children) or the full range of crimes and offences. Survey estimates are subject to a margin of quantifiable and non- quantifiable error. Strengths and Weaknesses: SCJS

7. Challenges in Making Direct Comparisons Unable to ‘match’ survey responses to recorded crime records. Changing survey methodology. Consistent crime survey data available since 2008/09. Timescales. SCJS is a continuous survey, published biennially whereas recorded crime is released annually. Population covered. Crimes and offences covered. Uncertainty around estimates for some crime groups.

8. Making Comparisons Covered by both sources. Can be coded in the same way. Adjustments are made to maximise comparability. Just under 2/3 of the SCJS 2012/13 can be compared with police recorded crime.

9. Recorded crime figures and SCJS estimates both show downward trends in the extent of crime since 2008/09 (-33% and -28% respectively). Uncertainty around SCJS estimates. What can the comparable subset tell us?

10. Comparisons can be made between crime estimated to have been reported to the police in the SCJS, and police recorded crime data. “Did the police come to know about this?”

11. ScotlandEngland and Wales

12. (in)conclusions We need two sources of crime statistics to help present a fuller picture of crime in Scotland. Relatively early stages of developing analysis of comparable crimes in the coming years. Limited conclusions can be drawn with only four survey sweeps and with uncertainty around changes between sweeps. Report forthcoming and new section in annual recorded crime statistical bulletin that brings together SCJS & RC. Plans to continue this analysis in future.

13. Do we do enough to explain the differences between the sources? Could we present the two sources better? – If so, how? Questions to Consider Is this analysis using the comparable subset useful? Can we use the comparable subset better? How can we deal with uncertainty?