Annex 2: Quality of the statistics
Governance of Data Quality for Police Recorded Crime:
7.16. The recording of crime by the police and the subsequent publication of information through this statistical bulletin is undertaken by those who are professionally independent of wider decision-making and performance processes within their organisations. Within Police Scotland, the Crime Registrars ensure that crimes in Scotland are recorded ethically. They sit within the corporate services business area so that they are removed from direct operational activity and investigation. In turn, the production of the National Statistics on recorded crime is managed in an impartial and objective way, in the public interest, by Scottish Government statisticians.
7.17. A Scottish Crime Recording Board supports the production of accurate and objective statistics on crime in Scotland. It takes into account the needs of both users and providers in the production of crime statistics and ensures that this process is undertaken in a manner consistent with the Code of Practice for Official Statistics. The Board is chaired by the Justice Analytical Services division of the Scottish Government and a wide range of organisations are represented including Police Scotland, the Scottish Police Authority, Her Majesty's Inspectorate of Constabulary in Scotland ( HMICS) and the Crown and Procurator Fiscal Service ( COPFS). Information on the remit and functions of the Board, its actions and outcomes, is available from the following link: http://www.gov.scot/Topics/Statistics/Browse/Crime-Justice/PubRecordedCrime/SCRB.
7.18. A Framework of Assurance is available for Recorded Crime statistics to provide evidence on:
- the suitability of the administrative data for use in producing national statistics;
- factors that the statistical producer needs to take into account in producing the national statistics;
- the information that users need to know in order to make informed use of the statistics.
The Framework of Assurance can be accessed via the link below. The aforementioned Crime Recording Board has responsibility for approving any changes to this product: http://www.gov.scot/Topics/Statistics/Browse/Crime-Justice/PubRecordedCrime/FoA.
7.19. On the 21st September 2016, Her Majesty's Inspectorate of Constabulary in Scotland ( HMICS) published Crime Audit 2016. This audit assessed the extent to which police recording practices complied with the Scottish Crime Recording Standard and Counting Rules.
7.20. HMICS audited just over 7,000 incidents reported between the 1st January and 31st March 2016. Five categories were included, three of which related to specific crime types:
- Sexual crime
- Violent crime
- Vandalism (Damage)
- Non-crime related incidents (i.e. incidents that were potentially crime-related but which were eventually closed as being non-crime related); and
- No-crimes (i.e. cases that were originally thought to be a crime but were later redesignated as not being a crime following additional investigation).
The examination of no-crimes involves an assessment of whether the no-crime decision was correct. Because the no-crime test is different from that applied to incidents, the no-crime results are reported separately.
The following tests were applied to incidents:
- Test 1 involves reviewing the initial report to the police (the 'incident') and assessing whether the incident has been correctly closed. Correct closure means either that (a) the incident was closed as non-crime related and contained sufficient information to dispel any inference of criminality; or (b) the incident indicated a crime had occurred and a crime report was traced. Incidents which result in a crime report proceed to Test 2.
- Test 2 involves reviewing the crime report to assess whether the crimes recorded are correctly classified and counted. Test 2 allows us to consider whether subsuming  has been carried out correctly.
- Timeliness: crimes should be recorded within 72 hours of the circumstances becoming known to the police (or within seven days where the delay is out with police control).
All three tests were applied to the three crime types HMICS examined (sexual crimes, violent crime, vandalism (damage)). Only Test 1 was applied to non-crime related incidents as reviewing them involves a simple assessment of whether the decision not to record a crime was correct.
The full report, including key findings, recommendations and improvement actions, can be accessed from the HMICS website:
7.21. In addition to HMICS, audits of incident and crime records are also undertaken by Police Scotland's Crime Registrars. From April 2015 onwards, and in line with a recommendation made by HMICS in their 2014 Crime Audit, updates on the results of these internal audits and Police Scotland's actions against other recommendations are presented to the Audit and Risk Committee of the Scottish Police Authority. Papers from the Committee's meetings are published, with those from meetings held in 2016 available via the following link:
7.22. In March 2016 the Crime Board considered how the audits by both HMICS and Police Scotland could best collectively support the quality assurance of these National Statistics. It concluded that HMICS should remain the primary source of audit-based assurance, given their role as an independent auditor. The results of Police Scotland's audit process will still be fed back to the Board by the National Crime Registrar. Where statisticians consider that this information has substantial data-quality implications, they will notify users of this through future publications and the SCOTSTAT network. Users will also be kept informed of any Crime Board action(s) taken in response.
Quality Assurance of Submitted Data:
7.23. The data provided by Police Scotland go through a process of quality assurance in which data received are checked across the four quarters of the financial year period, and against the previous year. Anything unusual or which we feel requires further explanation is then fed back to the Police for their attention. Any amendments are carried out and the final data is used to produce a set of data tables which can be used to check the final dataset. If the police are content that the figures are an accurate reflection of the dataset, then the data is officially signed off for use in the publication.
During the quality assurance checking process, it is possible for errors to be found in data for previous years. While we do not routinely revise figures, we are committed to correcting errors in the data and providing suitable explanations for any changes made to previously published data.
7.24. Detailed information on the quality assurance processes applied to recorded crime data is available in the Scottish Crime Statistics: Quality Assurance process document, available via the following link: http://www.gov.scot/Topics/Statistics/Browse/Crime-Justice/PubRecordedCrime/qualityassurance.
7.25. Before the publication of the Recorded Crime in Scotland, 2013-14 bulletin, an extensive quality assurance exercise was carried out to ensure consistency of data pre and post police reform, and to check the quality of recorded crime data supplied by the police. This followed the data being extracted from one central unit within Police Scotland, using their new information and performance management reporting tool ( ScOMIS), instead of being returned by the eight legacy police force areas. ScOMIS collates crime reports from across the variety of systems used by the eight legacy police forces into one central system, and supplies the Scottish Government with recorded crime data.
This exercise raised several data quality issues which led to a number of corrections to legacy data published in previous years.
In terms of comparability, the Scottish Government are satisfied that crime data, Groups 1 to 5, are comparable pre and post reform. This means there is no break in the time series for this data. However the overall comparability for offences data, Groups 6 and 7, has been affected by some of the issues relating to legacy force data and changes in recording practices. For Group 6 - Miscellaneous Offences, all crime codes are fully comparable from 2008-09 onwards. There are greater comparability issues for Group 7 - Motor Vehicle Offences and overall comparisons should not be made prior to 2013-14.
The extent of the amendments made to the data as part of the quality assurance exercise referred to above, along with all findings of this exercise and their impact on comparability, have been published in a Technical Report, available via the following link: