House of Commons - Caught red-handed: Why we can't count on Police Recorded Crime statistics

Caught red-handed: Why we can't count on Police Recorded Crime statistics - Public Administration Committee Contents

2 Police Recorded Crime (PRC)

Background

4. Crime statistics published by the Office for National Statistics (ONS) are central to the understanding of the nature and prevalence of crime in England and Wales. The statistics are based on two main sources:

· The PRC series: the number of 'notifiable offences'recorded by the police (notifiable offences comprise all offences that could be tried by jury, plus a few additional closely-related offences);

· The Crime Survey for England and Wales (CSEW; known until April 2012 as the British Crime Survey)-a large scale population survey conducted since 1982, which captures crimes experienced by adults, whether or not these crimes were reported to the police, as well as gauging public perceptions of the police.

Each of these sources has its own inherent strengths and limitations but together they should provide a more comprehensive picture of crime than could be obtained from either series alone.

5. Our inquiry examined crime recording practices in England and Wales. Police crime recording arrangements in Scotland and Northern Ireland are the responsibility of the devolved administrations; although the UKSA's jurisdiction is UK-wide. The Police Service of Northern Ireland records crime using the same National Crime Recording Standard as in England and Wales, while Scottish PRC is compiled in accordance with the Scottish Crime Recording Standard, introduced in April 2004. Both jurisdictions also conduct what are known as household victimisation surveys; the equivalent of the CSEW.

6. HM Inspectorate of Constabulary for Scotland published their "review of incident and crime recording" in December 2013.[2] This looked at compliance with crime recording standards since 1 April 2013. It found that 93% of the records examined complied with the standards, stating "Compliance rates varied according to crime type, ranging from 99% for domestic abuse to 89% for sexual offences. We were disappointed that the total proportionof compliant incidents fell below the accepted standard of 95%."[3] The review notes that "the very high compliance rates for domestic abuse illustrates what can be achieved when a focussed and robust approach is taken to attending, investigating and recording a particular crime type".[4]

7. Our inquiry covered crime statistics in England and Wales. However, it would be surprising if similar issues to do with the quality of the statistics did not exist in Northern Ireland and Scotland. HM Inspectorate of Constabulary for Scotland's recent review into compliance of police recording with the expected standards gives cause for concern-the recording of some crime types falls well below the expected standard.

8. We recommend that UKSA urgently investigate the quality of crime statistics in Scotland and Northern Ireland and their compliance with the Code of Practice, in the light of the findings of this inquiry, and UKSA's decision to remove the 'National Statistics' kitemark from crime statistics in England and Wales.

9. The collection and publication of data on crimes recorded by police forces in England and Wales began in 1857, making it one of the longest-running administrative datasets in the country. The main technical guidance document which informs police recording decisions, the Home Office Counting Rules (HOCR), has existed in one form or another since the 1920s.

10. The recorded crime figures are a by-product of a live administrative system which is continually updated as incidents are logged as crimes by police forces, and then investigated. As a result, some offences may change category, for example from theft to robbery. Other incidents initially recorded as crime may on further investigation be found not to be a crime-this is referred to as 'no-criming' (as distinct from 'not-criming', whereby an incident is not recorded as a crime in the first place). The rules stipulate that a recorded crime can be retrospectively 'nocrimed' if 'additional verifiable information' emerges which demonstrates that no crime was committed. Another relevant non-crime incident type is 'crime-related incident' (CRI), used when the balance of probabilities suggest that a crime was committed, but no victim (or representative) can be found to confirm this.

11. Currently, the Home Office is responsible for collating raw data from police forces each month, performing some validation checks and querying outliers with forces, who may then re-submit data. The Home Office statisticians then supply a snapshot of the data each quarter to the ONS for further analysis and then publication.

12. The PRC dataset serves several vital purposes within the landscape of criminal justice statistics. It:

· indicates trends in overall crime levels (in conjunction with the CSEW);

· includes offences (and victims) falling outside the scope of the CSEW, for example, offences such as possession of weapons and drugs, and potential victims such as those living in communal establishments;

· enables detailed analysis of crime incidence at a local level (the CSEW cannot do this, due to sample-size constraints);

· provides detail on the incidence of individual offences and offence types;

· underpins the data on crime detection rates and criminal justice outcomes; and

· gives forces an account of what crimes are happening and where, and provides an important indicator of the size and distribution of police workloads, and so is a crucial factor which determines how police forces identify priorities and deploy resources.

13. Accurate Police Recorded Crime data is essential if Police and Crime Commissioners and Chief Constables are to know what crimes are being committed in their area and therefore how to respond.

14. There have long been concerns about the reliability and consistency of police recording practices. Debates about whether changes in PRC reflected actual changes in crime rather than changes in reporting and recording practices were part of the reason for the introduction of the British Crime Survey in 1982. The last fifteen years have seen a succession of reports examining PRC as part of broader reviews of crime statistics, detailed in the table below.

Previous reports since 2000 documenting concerns about crime statistics
·  HMIC, Povey, K,On the record: Thematic Inspection Report on Police Crime Recording, the Police National Computer and Phoenix Intelligence, 2000
·  Home Office, Review of Crime Statistics, 2000
·  Lynn P and Elliot D, The British Crime Survey: a review of methodology (2000)
Simmons J, Legg C and Hosking R, National Crime Recording Standard: an Analysis of the Impact on Recorded Crime (2003)
·  Statistics Commission, Crime Statistics: User Perspectives-interim report and final report, 2006
·  Smith, A, Crime Statistics: An independent review, carried out for the Home Secretary, 2006
·  Audit Commission, Police Data Quality 2006-07, 2007
·  Casey L, Engaging Communities in Fighting Crime, Cabinet Office, 2008
·  Sir Ronnie Flanagan, Independent Review of Policing, 2008
·  UKSA, Overcoming Barriers to Trust in Crime Statistics England and Wales Monitoring report 5, 2010
·  ONS, National Statistician's Review of Crime Statistics for England and Wales, 2011

15. In the wake of a critical inspection of police recording practices published by HM Inspectorate of Constabulary in 2000, the National Crime Recording Standard (NCRS) was developed and introduced in 2002/03, with the aim of standardising crime recording practices across forces and establishing a more victim-oriented prima faciemodel of crime recording whereby the police are required to record a victim's report if it amounts to a crime in law and there is no credible evidence to the contrary.

16. The introduction of the NCRS led to an immediate structural increase in the number of crimes recorded in the first two years of its implementation (2002-03 and 2003-04). Since this initial bedding-in phase, overall recorded crime levels have fallen in every subsequent year, at a faster rate that the CSEW suggests is credible.

17. In January 2011 the Home Secretary announced an independent review into the collection and publication of crime statistics, to be led by the National Statistician.[5] In order to address concerns that the Home Office's involvement in publishing crime statistics was undermining public confidence in the political independence of the figures, the review was tasked with identifying an independent body which would take over responsibility for publication. The review, published in June 2011, recommended that responsibility for publishing the main crime statistics (PRC and the British Crime Survey, now the CSEW) be transferred from the Home Office to the ONS; this took effect from April 2012.[6] The review also led to the creation of a Crime Statistics Advisory Committee (CSAC) to provide expert advice on methodological issues.

Concerns about data quality

18. The CSEW provides strong evidence that the overall volume of crime has been falling over the past twenty years. The findings of the CSEW broadly parallel the overall trend indicated by the PRC data since 2002-03 (the first year of NCRS implementation)-the overall volume of crime recorded by both measures has fallen by 38% (from 2002/03 to year end September 2013), as illustrated by the chart below.

19. There will be inevitable changes over time in how people report crime-what an independent review of crime statistics, written for the Home Secretary in 2006, called "unknown and uncontrollable variability in the public's reporting of crime to the police".[7]However, we have seen an accumulation of substantial and credible evidence-based on statistical analysis and on authoritative testimony from current and former police officers-indicating that:

· the PRC data doesnot correctly represent the rate of decrease in crime or the composition of crime;

· that the erosion of police compliance with the agreed national standards of victim- focussed crime recording has contributed to this; and

· that monitoring and audit arrangements have been insufficient to ensure acceptable standards of data quality and integrity.

As a result of this evidence, UKSA decided in January 2014 to strip the PRC data of their designation as National Statistics, discussed later in this report.

20. In January 2013 the ONS published an analysis of crime trends which identified a divergence between the PRC data and the CSEW.[8]While both datasets show a clear downward trend over the last decade, the PRC dataset has in recent years shown a faster decline than the CSEW for comparable offences, resulting in a smaller ratio of recorded crimes to CSEW crimes within the comparable subset, shown in the chart below.

21. While the ONS analysis has said that "the data can't tell us why the police appear to be recording a lower proportion of crime reported to them than in previous years", it did suggest that declining standards of compliance with the established recording rules may have contributed to this divergence:

Among the possible drivers for this divergence, the ONS suggested the following:

· lack of awareness or adequate understanding of the NCRS as time passes from its launch leading to some officers recording 'as charged' or 'if detected' which might result from staff turnover and lack of sufficient on-going training;

· performance pressures associated with targets (for example, to reduce crime or increase detection rates) acting as perverse incentives for some crimes to be downgraded from notifiable into non-notifiable categories or as anti-social behaviour or as crime-related incidents (which are not captured in data returned to the Home Office);

· though forces have continued with their own internal audits, the cessation of independent audits from 2006-07 onwards may have reduced the focus on addressing non-compliance;

· the move to Neighbourhood Policing in recent years may also have led to more low-level crimes being dealt with informally and outside the formal crime recording system; and

· in the context of pressure on police budgets and a general policy shift to promote greater officer discretion, a return to a more evidential recording model.[10]

22. The ONS's analysis was one of the most recent to signal a problem with the PRC data. However, as the ONS report concedes, such high-level analysis:

23. To gain a deeper insight into the reality of police forces' recording practices, we are heavily reliant on the testimony of concerned officers and staff working on the 'frontline' of the crime recording system. This inquiry was prompted by the concerns expressed by PC James Patrick, a serving officer in the Metropolitan Police with involvement in data analysis.PC Patrick is also a constituent of the Chair of PASC, Bernard Jenkin MP.

24. In his written and oral evidence to this inquiry, PC Patrick raised a number of specific and serious concerns relating to crime recording practices in the Metropolitan Police-such as the downgrading of offences to a less serious offence category (for example, from robbery to theft from the person) and also particularly troubling evidence in relation to the potential misrecording of sexual offences, which we consider later in this Report.[12]

25. Such concerns have been reinforced by an array of evidence and comments from serving and former officers, including those who have served at the most senior levels. For instance, Lord Stevens, former Metropolitan Police Commissioner, told the Home Affairs Committee in January 2014 that "ever since I have been in the police service, there has been a fiddling of figures" and that it was still going on.[13]Evidence from Dr Rodger Patrick (no relation of PC James Patrick), former Chief Inspector at West Midlands Police, sets out his research showing how the perverse incentives embedded in quantitative performance management regimes encourage a range of 'gaming' behaviours that result in under-recording of crime.[14] We discuss the issue of performance culture and gaming later in this Report.

26. It is suggested that deliberate misrecording of crime is one source of under-recording. However, under-recording of crime can of course come about as a result of police officers' misunderstanding or ignorance of the established rules and principles of crime recording, which would be a particular problem in forces where crime recording training is inadequate or where there is insufficient communication of the core principles of crime recording to officers. For example, officers may erroneously set the evidential bar too high when making a recording decision, based on their perception of the likelihood of a Crown Prosecution Service charge, rather than using the victim-focussed standard prescribed by the NCRS.[15]Witnesses have also pointed out that lack of understanding of the counting rules can in some instances lead to over-counting of crime.[16]

27. Gwent Chief Constable Jeff Farrar, in his capacity as National Policing Lead for Crime Statistics, also raised the possibility of tension between compliance with official rules and the common-sense exercise of professional discretion in the public interest:

To illustrate this point, Chief Constable Farrar used the example of a parent contacting the police for help in disciplining an 11-year-old son who had stolen money from her purse:

28. Olivia Pinkney, then Assistant Inspector of Constabulary at HMIC (now Deputy Chief Constable at Sussex Police), explained to us that the new revised framework for recorded crime outcomes gives officers scope to apply a non-punitive outcome to a recorded crime where a prosecution or other criminalising sanction-detection outcome may not be in the victim's or the public's interest. She added that there will be "a much greater breadth of explanation for the public" about this.[19] The revised framework aimsto ensure that police officersknow, and the public understand, that the policehave the discretion to take a victim-focussed, common-sense approach, within the standards for recording crime.[20]

29. It is not credible to suggest that sensible resolution of the tensions between a rigid compliance with the recording rules and a common-sense approach can explain the exaggerated decline of Police Recorded Crime.Our witnesses provided a wealth of insight into the various ways in which crime data accuracy and integrity can be and have been compromised. However, the lack of regular and rigorous audit of crime recording practices in recent years makes it impossible to assess the extent of any compromise andthe relative importance of these factors. The re-establishment of regular annual external audit of forces from this year onwards, which we discuss later in this Report, provides a vital opportunity to fill this gap in the understanding of the problem and to contribute towards a durable solution.

30. It is vital that the Government ensures the accuracy and reliability ofPolice Recorded Crime.Police Recorded Crime provides a crucial intelligence resource for the police and informs the operational deployment of police resources. Lax supervision of recorded crime data risks reducing the police's effectiveness in their core role of protecting the public and preventing crime because they cannot deploy resource effectively if they are not aware of the true level and nature of crime.

31. Under-recording or miscategorising crime erodes public trust in the police and undermines the trust and confidence of frontline police officers in police leadership: it creates doubt that the public will be taken seriously when they report a crime.

Misrecording of sexual offences

32. A particular troubling aspect of the evidence heard by the Committee related to the misrecording of sexual offences by means of excessive recourse to 'no-criming' decisions and classifying cases as 'crime-related incidents' (CRI), rather thanrecorded crimes. The IPCC's critical report on the Southwark police's Sapphire Unit's recording of sexual offences in 2008-09,found "officers of all ranks [...] felt under pressure to improve performance and meet targets".[21] It stated that no-criming "benefited the unit's performance statistics" and "the number of serious sexual offences classified as a 'no crime' or as a 'crime related incident' was consistently higher than the MPS [Metropolitan Police Service] average".[22]PC Patrick described to us a more recent analysis of sexual offence recording decisions he conducted in 2013. He suggested that his findings indicatedcontinuing excessive 'no-criming' and CRI-ing of sexual offences.[23]

33. Sir Bernard Hogan-Howe, the Metropolitan Police Service Commissioner, gave oral evidence to the Home Affairs Committee two weeks after our session with PCPatrick (largely in connection with their inquiry into counter-terrorism). He said then that "[HMIC] inspected our systems in 2012 and found them to be competent and reliable" and "we think that some of the comments that this officer made to the Public Affairs Committee relate to a period of over two years ago when the no crime issue was around 25%".[24] However, when he gave oral evidence to us in January 2014, he added that "some of the concerns that were expressed—for example about the nocriming of rape—are things that for police, and for others, have been a real issue over many years".[25]PC Patrick submitted further evidence,which argued that the concerns were recent, not historic, giving data up to 2012-13.[26]

34. Data subsequently obtained from the Metropolitan Police under Freedom of Information by the Bureau of Investigative Journalism, and included in PC Patrick's evidence, show the 'no-crime' and CRI rates in relation to alleged rapes reported to the Metropolitan Police in 2008-09 to 2012-13. The figures show that a decline in the 'no-crime' rate after 200809 was accompanied by an offsetting increase in the CRI rate in the same period, leaving the overall no-crime-and-CRI rate within the range of 25%-30% over the five years in question. This gives rise to the suspicion that even though police forces may have succeeded in reducing instances of unjustified 'no criming', any gain in overall accuracy of the recording was offset by an increase in those reported crimes which were categorised as CRIs. PC Patrick pointed out in his written evidence that during the period in question "significant efforts were made to specifically reduce 'no-criming'-the central issue raised by the IPCC in the Southwark report".[27]

35. We heard evidence that the desire to avoid unsolved reported sexual offences remaining on the system can go as far as trying to justify 'no-crime' on the basis of "mental health or similar issues of vulnerability" and that "what happened in Southwark is still happening."[28]In the wake of these allegations, the Metropolitan Police ServiceCommissioner Sir Bernard Hogan-Howe announced before the Home Affairs Committee on 3 December that the Metropolitan Police had commissioned a piece of academic research examining the force's no-criming decisions in respect of sexual offences, in order to investigate whether police officers put pressure on women to withdraw their allegations.[29]

36. There remain wide disparities in the no-crime rates for reports of rape crime. In January 2014, HMIC, on behalf of the Rape Monitoring Group, released a compendium of statistics on recorded rapes in each force over the previous five years. This revealed wide disparities between forces in the no-crime rate for reported rapes and in the rates of recorded rapes per 100,000 adults.According to these figures, in Lincolnshire, for example, 26% of all reported rapes were no-crimed in 2012-13 and 20% were no-crimed in 2011-12. This contrasts with Merseyside, where 4% reported rape crimes were no-crimed in 2012-13 and 9% were no-crimed in 2011-12.[30]

37. In the chart below shows how far the no-crime rate for reported rape incidents differs from the average no-crime rate for England and Wales, aggregating the data from April 2008 to March 2013.[31] The national average no-crime rate for that period was 11.9% - that is, an average of 11.9% of reported rape incidents were no-crimed over that five year period. The chart shows that some forces, such as Cleveland, Surrey and Lincolnshire, had a far higher no-crime rate than the national average, while others, such as South Yorkshire, South Wales and Essex, had far lower no-crime rates than the national average.

38. When we asked him about these issues, Sir Bernard told us that the data accuracy for rape and sexual offences was "a lot better than it was, if we took it back five to 10 years" but did not think that it was entirely reliable.[32]He agreed that there was a "cause for concern" and told us that "there is clearly something that PC Patrick raises that we need to get to the bottom of."[33] He said he was taking three actions to address these issues: (1) invite academics to review quarterly the Metropolitan Police's nocrime reports (2), invite a public figure to look at how the Metropolitan Police looks at sexual offences generally; and (3) change the standard for no-criming sexual offences to "beyond reasonable doubt".[34]

39. Any instance of deliberate misrecording of sexual offences is deplorable, but especially so if this has been brought about by means of improperly persuading or pressurising victims into withdrawing or downgrading their report.

40. The disparities between different police forcesin the 'no-crime rates' for rapes and sexual offences aresufficient in our view to raise serious concerns about the varying approaches taken by police forces to recording and investigating these horrendous crimes. We look forward to the outcome of the research commissioned by the Metropolitan Police examining the force's 'no crime' decisions in respect of sexual offences.

41. The fact that this research is necessary, following the 2008 Independent Police Complaints Commission report into the Sapphire Unit is a damning indictment of police complacency, inertia and lack of leadership.However, the data indicates that the Metropolitan Police Service is unlikely to be the only force of concern.

42. The Home Office must undertake a comprehensive analysis in order to explain the extraordinary disparities in no-crime rates for sexual offences across all police forces. We expect this to be completed within two months and included with the response to this Report.We also recommend that the devolved administrations undertake analogous work. This should lead to work to improve the accuracy transparency and reliability of police recorded sexual offences so that a table of no crime rates does not suggest systemic inconsistency in recording practices.

The role of the Crime Survey for England and Wales (CSEW)

43. The CSEW is a crucially important counterpart to PRC-not least as a benchmark against which the PRC trends can be compared. It is vital therefore that UKSA and the ONS ensure that the CSEW continues to meet the National Statistics standard.However, as it stands, the CSEW cannot give a detailed indication of crime trends at a local level. Although it is a substantial survey, we understand that the sample size is still too small in each force area for the local area survey statistics to give meaningful results for most crimes.

44. We heard that the CSEW could only replace PRC at police force, let alone local or neighbourhood, level at significant cost: an additional £13.7 million would be required in order to increase the sample, on top of the existing annual cost of CSEW of £3.8million.[35]From 2004-05 to 2011-12 the CSEW's core sample size was 46,000 adults; as of 2012-13 this has been reduced to 35,000.[36] In respect of interviews with children, the sample has been reduced from 4,000 to 3,000. According to the Home Office, the reduction in the core adult sample resulted from the scrapping of a target in the old Police Performance and Assessment framework, which had required a boosted sample size to achieve 1,000 interviews in each police force area.[37] The Technical Report to the CSEW does no more than state that there has now been a reduction in sample size; it does not give contextual information as to the impact of the sample size on the reliability of the statistics relating from the survey.[38]

45. We note the reduction in the sample size of the Crime Survey for England and Wales.Police Recorded Crime is the only detailed indicator of crime trends at local level, enabling police forces, Police and Crime Commissioners, local authorities, the public and the Home Office to keep track of crime in different force area. The Crime Survey for England and Wales is no substitute for Police Recorded Crime in respect of monitoring crime trends in local areas.

46. We recommend that theONS review and then publish,alongside the Crime Survey for England and Wales, information about the nature of the sample, including the impact of the reduction in sample size on the reliability of the statistics, its cost over time, and an explanation of what statistics might be published at a sub-national level, for example for the larger police forces.

2 HM Inspectorate of Constabulary for Scotland, Review of incident and crime recording, December 2013 Back

3 HM Inspectorate of Constabulary for Scotland, Review of incident and crime recording, December 2013, p1 Back

4 As above, p1 Back

5 HC Deb, 20 Jan 2011,col 49WS Back

6 Office for National Statistics, National Statistician's Review of Crime Statistics: England and Wales, June 2011 Back

7 Professor Adrian Smith, Crime statistics: an independent review, carried out for the Home Secretary, 2006 Back

8 ONS, Methodological note: analysis of variation in crime trends, 24 January 2013

Back

9 As above, p10 Back

10 As above, p10 Back

11 As above, p11 Back

12 CST02, CST34, CST73, Q6-64 Back