Embed Size (px)
Citation preview
Closing the (protective services) gap:why size does matter when determiningoptimal Level 2 service delivery
Michael Townsley† and John W Bond‡
†(Corresponding author) Jill Dando Institute of Crime Science, University College London,Second Floor Brook House, 2–16 Torrington Place, London, WC1E 7HN.Tel: +44 (0)20 7679 0820; Fax: +44 (0)20 7679 0828; email: [email protected]‡Scientific Support Unit, Northamptonshire Police, Wootton Hall, Northampton, NN4 0JQ.Email: [email protected]
Received 11 April 2006; revised and accepted 29 August 2006Keywords: detection, forensic, amalgamation
Michael Townsley is a Senior Research Fellowat the Jill Dando Institute. He completed anApplied Mathematics and Statistics degreebefore embarking on a PhD in Criminology.Having studied the spatial and temporal distribu-tions of burglary, and taken time out to work on anumber of crime prevention projects, he leftAustralia to work in Liverpool, England. Timespent advising Merseyside Police on crime ana-lysis was put to good use when he wasemployed as a manager of a New Deal for Com-munities (NDC) crime theme. Current interestsfocus on problem oriented policing, prolificoffenders and crime analysis. His current pro-jects involve the incorporation of forensic datawith routine crime analysis.John W. Bond is Head of Forensic Science forNorthamptonshire Police. His research interestscentre on the use of forensic science to enhancethe detection of crime, principally through theanalysis of forensic data and the development ofnew techniques for the enhancement of finger-prints. He has research affiliations with a numberof UK universities and lectures on forensic sci-ence at Sussex University.
ABSTRACT
The September 2005 review of policing structurein England and Wales by Her Majesty’s Inspec-torate of Constabulary (HMIC) is considered interms of its impact on forces’ ability to detectvolume crime (crime types that occur in relativelylarge numbers). We demonstrate that there is norelationship between force size and detection rates,however there is a difference with respect to themeans of detecting volume crime. Smaller forcestend to record a greater proportion of their volumecrime detections through forensic means. Weexplore one possibility of this observation —greater forensic deployment due to lower demandor better management — and find some evidenceto support better management of forensic resourcesin smaller forces.
INTRODUCTIONIn England and Wales mainstream policingis oriented around a hierarchy of criminal-ity, labelled Levels 1, 2 and 3. Level 1 isconcerned with local issues and is con-sidered appropriately dealt with at the BasicCommand Unit (BCU) level. Level 3focuses on serious and organised crimeoperating at a national or international leveland is the remit of the Serious Organised
International Journal of Police Science & Management Volume 9 Number 2
International Journal of PoliceScience and Management,Vol. 9 No. 2, 2007, pp. 183–192.© Vathek Publishing,1461–3557
Page 183
Crime Agency (SOCA), the amalgamationof the National Crime Squad, NationalCriminal Intelligence Service, HomeOffice departments responsible for organ-ised immigration crime and Customs staffexperienced in serious drug trafficking andcriminal asset confiscation. Level 2 dealswith ‘cross-border’ crime — essentiallycriminal acts perpetrated between twopolice jurisdictions.
The current regime of police servicedelivery is oriented around Level 1 crimin-ality, with Level 3 the subject of a dedicated(albeit newly formed) agency. Level 2 crim-inality is often the subject of spare capacity,regional teams drawn together on an ad hocbasis or some other non-systemic approach.Apropos of this, O’Connor (2005, p. 24)quotes from Stephenson (2004, p. 13) thatfewer than 6 per cent of the estimated 1,500organised (Level 2) criminal networks activenationally, were targeted by police in anygiven year.
It was in the light of apparent lack ofcapacity at tackling Level 2 criminality thata ministerial report, Mind the (Level 2) Gap,recommended (among other things) that‘[a] confidential national assessment of pro-tective services should be carried out byHMIC’ (O’Connor, 2005, p. 88). The ful-filment of that recommendation was therecently published, Closing the Gap, HMICreview, authored by Denis O’Connor.Commissioned by the Home Secretary andpublished in September 2005, the report isa review of the organisational structure ofpolicing in England and Wales and considersthe ability of the existing structure of poli-cing to deal effectively with protective servicesas well as volume crime (the focus of thecurrent structure). A secondary objectivewas to determine whether collaborationwas an appropriate method to address ser-vice delivery shortfalls.
Beyond assessing police structure, stake-holders placed a number of priorities on theClosing the Gap review, one of which was to
‘[s]afeguard the resilience of local (BCU/Neighbourhood) services’ (O’Connor,2005, p. 21). In other words, while Level 2issues are undoubtedly important, anyrecommendations for restructuring musttake into account the ramifications foraddressing Level 1 criminality.
Protective services, the focus of theHMIC review, have been identified underthe following seven broad headings:
● major crime (homicide);● serious, organised and cross-border
crime;● counter-terrorism and extremism;● civil contingencies;● critical incidents;● public order; and● strategic roads policing.
The HMIC review considered individualforces’ capability and capacity to provide forprotective services and these were assessedagainst standards on intelligence, preventionand enforcement/resolution. HMIC foundthat very few of those forces assessed metthe required standard and concluded thatforces with over 4,000 police officers (or6,000 staff in total) ‘tended to meet thestandard across the seven protective servicesmeasured’ (O’Connor, 2005, p. 7). Thereview acknowledged the success of thecurrent policing structure in reducing levelsof volume crime, in contrast to the con-tinued rise in many indicators associatedwith more serious crime and the protectiveservices.
The review concluded, therefore, that arestructuring of the police service wouldprovide increased resilience to support pro-tective services and, at the same time, offerresilience in other areas, such as finance,where the review highlighted ‘cost pres-sures’ in the provision of (for example)forensic science and DNA analysis.
The recommendations of the reviewhave received much attention, with some
Why size does matter when determining optimal Level 2 service delivery
Page 184
commentators claiming it represented ‘thebiggest shake-up of police in thirty years’(BBC, 2005). While much of the discussionaround force mergers has focused on norm-ative positions either for or against theHMIC review, precious little has beendevoted to exploring empirically the impactamalgamation may bring.
To our knowledge Harrad (2006) repre-sents the only independent, quantitativetreatment of the implications of force mer-gers. She considered the economic argu-ment for amalgamation and, while notarguing against it, concluded that withoutincreased capacity ‘the case for amalgama-tion would at present need to be made ongrounds other than cost and promises ofbetter performance in those areas measured’(Harrad, p. 26).
In May 2006 John Reid was appointedHome Secretary and in little more than amonth it was announced that he would notbe pursuing compulsory force mergers asindicated by his predecessor. The criticalissues appear to relate to cost, both ofimplementing mergers and to local author-ities within merged forces. Future plans tofill the protective services gap have focusedon assessing the strengths and weaknesses ofa range of different operational configura-tions (such as inter-force operations, thelead force model and shared services).
The original intention of the analysis setout below was to compare and contrastvolume crime performance of differentforces so that mergers could be imple-mented to minimise the impact at Level 1(ie by capturing and therefore not dilutinggood practice). Nevertheless, now that thepolice reform agenda has shifted to con-sidering the benefits of different methods ofworking, we feel that comparing volumecrime performance of different forces willgo some way to informing the debate.
DATA ANALYSISTwo main sources of data were used forthis analysis. The first was the numbers ofrecorded criminal incidents and detectionsfor six key offences in each BCU inEngland and Wales during 2004/05 (seeNicholas, Povey, Walker, & Kershaw, 2005for a description). Three offences — vio-lence, sexual offences and robbery — wereaggregated to produce a proxy for seriousoffending. Similarly, three property crimes— domestic burglary, theft of motor vehicleand theft from motor vehicle — wereaggregated to derive a proxy for volumeoffences. The same procedure was used toyield detections for serious and volumeoffences. It should be noted that somecrime categories (eg murder, drugs) will beexcluded from this count, but the proxiescontain the major offence types in both seri-ous and volume crime. This should ensurethat forces are compared appropriately.
The second source of information camefrom Home Office National Forensic Per-formance Monitors and Trackers. Thesemeasurements cover, for every police forcein England and Wales, attendance, recoveryand processing of both DNA and finger-prints for residential burglaries and theft ofmotor vehicles using a ‘Forensic Impact’measurement method (Williams, 2004). Italso includes the contribution made byforensic science to detecting residentialburglaries and theft of motor vehicles. Eachperformance measure is graded red, amberor green depending on the performance ofeach force for that measure with respect toall other forces. An overall force grading isthen produced based on an unweightedassessment of 14 contributing gradings. Thisis known as the forensic impact grading.
Scientific support managers in eachpolice force submit to the Home Office arange of monitoring information on for-ensic science activity. Included in this arethe numbers of crimes detected through
Townsley and Bond
Page 185
forensic means. Incidents where forensicinformation contributed to the detection ofan offender are labelled primary detections.Any further detections arising from a pri-mary detection (offences taken into con-sideration, say) are labelled secondarydetections.
In line with the Closing the Gap report,we partitioned forces into ‘small’ (less thanor equal to 4,000 officers) and ‘large’(greater than 4,000 officers). We alsoexcluded the two London police forces onthe following grounds: (a) the MetropolitanPolice Service as it represents an extremeoutlier with an officer strength of over30,000 police officers (as at January 2005);and (b) the City of London Police becausethe residential population they serve is verysmall, about 7,000 (as at January 2005) andconsequently the magnitude of volumecrime dealt with is inconsequential.Excluding the London forces left a samplesize of 5 forces considered large and 36forces considered small.
We begin by comparing the detectionrates for volume and serious crimes inBCUs. BCUs were partitioned into twogroups according to whether they arelocated in forces which are considered small(N = 167) or large (N = 53) as defined
earlier. In order to examine the differencebetween BCUs in small and large forces, weobserved the distributions directly by esti-mating their density using a kernel densityfunction. Essentially this is equivalent to ahistogram for a continuous variable. Theresults are shown in Figure 1.
Figure 1 shows that the respective detec-tion rates for volume and serious crime arebasically the same for both groups, althoughBCUs in small forces appear to be locatedjust to the right for both crime types,indicating slightly higher detection rates.The distributions for the small force BCUsshare very similar ‘spreads’ with large forceBCUs, although small force BCUs have alonger upper tail for both crime types,augmenting somewhat higher detectionrates than BCUs in large forces. Thus, itappears BCUs in both small and large forcesperform identically and where they do notit is BCUs from small forces which per-form better than other BCUs (regardless offorce size).
Our attention turns to whether themethod of achieving detections variesbetween small and large forces. This is achallenging prospect as there are a multi-tude of factors at play that determinewhether a given crime becomes detected:
Figure 1Estimated distribution ofdetection rate of seriousand volume crime ofBCUs by force size
Why size does matter when determining optimal Level 2 service delivery
Page 186
the discriminating modus operandi of theoffender, the quality of recording, theextent of investigations, crime scene exam-iner presence and competence, interview-ing strategy, levels of intelligence, and soon. One area on which there is reliableinformation is extent and outcome offorensic activity.
The only information to which we hadaccess was aggregated at the force level. Asthere are only five police forces in the largecategory, density estimates on a sample sizeof five would be pointless. Instead we havesimply displayed the distributions of pro-portion of primary detections due to for-ensic data type and police force size. Figure2 shows the distributions using strip charts,a one dimensional equivalent of a scatter-plot; each force is represented by an indi-vidual symbol.
Figure 2 shows two features worth not-ing. First, large forces, relative to smallforces, do not vary much in their propor-tion of volume crime detected primarilythrough forensic means. Small forces are
much more variable, ranging from 0.5 to3.3 per cent for DNA-led detections com-pared with a range of 1 to 1.2 per cent forlarge forces. The pattern is repeated forfingerprint detections (1.1 to 2.7 per centcompared with 1 to 7.1 per cent). Second,the majority of small forces have a greaterproportion of primary detections fromforensic means than do large forces. Smallforces are either as good as or better thanlarge forces at detecting volume crime withforensics.
A further feature of the data not apparentfrom Figure 2 is that generally forces withhigh primary detection rates through finger-prints are the smallest forces. The correla-tions between force size and primarydetections through fingerprint is –0.49 (p <0.002) for all 41 forces used in the sampleand –0.36 (p < 0.033) for the 36 smallforces. The corresponding correlation coef-ficients for DNA primary detections werenon-significant.
How can the features of Figure 2 beinterpreted? If it is known that detection
Figure 2Distributions of primaryvolume detections byforensic type for differentforce types. (The actualvalues are jitteredvertically to distinguishindividual values and tominimise over-plotting ofadjacent symbols.)
Townsley and Bond
Page 187
rates of serious and volume crime at theBCU level are largely independent of forcesize (analyses at the force level yield thesame pattern) but at the force level there is adifference in the nature of volume detec-tions, what could explain a higher apparentutilisation of forensic science among smallforces? One hypothesis is that small forcesdo not possess the depth of investigatorexperience that large forces do. Perhapslarge forces with higher numbers of casesare better placed to systematise procedureswhich have not yet transferred to smallforces, or even that individual officers areexposed to a wider array of criminalbehaviour on which they ‘cut their teeth’.This could compel officers in small forcesto rely on forensic solutions for detectionsto a greater degree than their counterpartsin large forces.
An alternative explanation is that Scien-tific Support Units (SSUs) in small forcestend to collect more useful forensic materialfrom crime scenes because they are man-aged in a different way, or the demandprofile is different enough that allows themto do so.
It is inconceivable to test the formerhypothesis (differential investigative expert-ise) using aggregate data. There are simplytoo many variables that will be inadequatelycaptured by performance data that areeasily available. In the not too distant futurethis may possible with the realisation ofthe Professionalising Investigation Process(Home Office, 2001). Instead, we focus onthe second hypothesis regarding the vari-ation of SSU performance.The possible factors that allow small forcesto collect more or different forensic mater-ial might be:
1. Superior scene attendance. Small forces mayhave greater resources at their disposal incertain situations. All things being equal,forces should have similar ratios forcrime scenes to Crime Scene Examiners
(CSEs). This is likely to break down atthe extremes, for either very small forces(there must be a minimum staffingthreshold above which SSUs must main-tain) or for large forces (when crimerates outstrip CSE recruitment, trainingand development). Substantially differentratios may result in some forces’ attend-ing a greater proportion of scenes;having more time to spend examiningeach scene; collecting more material/scenes; or some combination of these.
2. Better evidence processing. The only area offorensic evidence that is conductedwithin forces and is common enough toyield a difference between detections isfingerprint matching. Thus, we areinterested in the number of fingerprintexperts each force possesses, their work-load and their respective matchingrates.
3. Some combination of the above.
The following analysis is carried out at theforce level, as some forces do not have CSEsdedicated to BCUs (and, for reasons statedabove, the Metropolitan Police Service andthe City of London Police have beenexcluded from this analysis). In order todetermine whether small forces enjoy percapita greater resources per scene, we calcu-lated for each force the average number ofcrime scenes a CSE will attend annually.This was computed by summing the num-ber of domestic burglaries attended, thenumber of theft of motor vehicle incidentsattended and all serious crimes recorded foreach force, and then divided by the numberof CSE staff. We only counted attendedvolume criminal incidents as there is a fairdegree of variation in attendance propor-tions. We assume that all serious criminalincidents are subject to some forensicresponse due to their rareness and nature.Note we have omitted scenes of theft frommotor vehicle incidents in our calculationsonly because these are not included in the
Why size does matter when determining optimal Level 2 service delivery
Page 188
Home Office forensic impact measurementsavailable to us.
The mean number of scenes attendedannually for small forces was 469.7 scene(per examiner, SD 128.9), compared withan average of 497.3 scenes per examiner forlarge forces (SD 82.3). In practical terms, ifthis mean difference was distributed uni-formly throughout the year, CSEs in largeforces would, on average, attend an extracrime scene every ten calendar days com-pared with their small force counterparts.Unsurprisingly then, a non-parametrichypothesis test failed to reject the nullhypothesis of equivalent central tendency(p = 0.38). Figure 3 shows the two distribu-tions using strip charts.
Figure 3 has a number of notable fea-tures. Within the two classes of police forcesthere is evidence of a large degree of vari-ation, with CSEs in some forces potentiallyresponding to almost double (large forces)or almost treble (small forces) the number ofcrime scenes as their counterparts in otherforces. For instance, CSEs in small forcescould potentially respond to between 250crimes to nearly 800 annually.
There is little evidence that small forcesenjoy greater per capita resources than largeforces. While it is difficult to be prescriptivewith a sample size of five, the distribution
of the large forces’ scene to examiner ratiois not dramatically different from that of thesmall forces.
The second factor feasibly influencinggreater forensic utility among small forces isthat of superior evidence processing. Forthis we rely on fingerprint expert perform-ance as this function is housed intra-force.As all experts have identical training, differ-ential performance of fingerprint bureauxindicates something about the organisation.Forces generally possess fingerprint expertsin direct proportion to their officer strength(correlation of 0.88). Workload for finger-print experts is measured by the number ofcases (crimes) from which fingerprints arerecovered and submitted to bureaux. Ifsmall forces are gathering information atmore scenes, then we would expect the caseto expert ratio to be significantly higherthan for large forces. This ratio was com-puted for each force and the distribution isdisplayed in Figure 4.
The distributions shown in Figure 4show a great deal of variation within forcesof a given size. There is a three- and two-fold difference between the highest andlowest ratios for small and large forcesrespectively. Importantly, though, there isno indication that fingerprint experts insmall forces enjoy a lighter workload than
Figure 3Strip chart of scene toexaminer ratio for largeand small police forces.(The actual values arejittered vertically todistinguish individualvalues and to minimiseover-plotting of adjacentsymbols.)
Townsley and Bond
Page 189
experts in large forces. If anything, expertsin small forces appear to have a heaviervolume crime case workload.
The next area of performance examinedis the rate at which individuals are identifiedfrom crime scene fingerprint lifts. Here, wecomputed the number of identificationsfrom fingerprints divided by the number ofexperts for each force. Figure 5 contains theresults.
The features of Figure 5 worth notingare that small forces have a greater rangethan large forces and that this range islargely in a positive direction, ie towardsidentifying offenders for a greater proportionof criminal incidents.
To recap, small forces are similar to largeforces with respect to the detection ofvolume and serious crime, crime sceneexaminer workload and fingerprint expertworkload, but different in terms of methodof volume detection and fingerprint expertperformance.
DISCUSSIONWhilst a ‘critical mass’ of 4,000 policeofficers has been proposed by the HMICreview to provide adequately for protectiveservices, it is apparent from this analysis thatperformance at detecting Level 1 criminal-ity differs considerably within the existing
Figure 4Strip chart of case tofingerprint expert ratiofor large and small policeforces. (The actual valuesare jittered vertically todistinguish individualvalues and to minimiseover-plotting of adjacentsymbols.)
Figure 5Identifications perfingerprint expertannually. (The actualvalues are jitteredvertically to distinguishindividual values and tominimise over-plotting ofadjacent symbols.)
Why size does matter when determining optimal Level 2 service delivery
Page 190
policing structure. For residential burglaryand theft of motor vehicle offences, theHome Office data show that a significantpercentage of detections for these crimesare brought about through DNA or finger-prints (ranging between 35 per cent and 7per cent of total detections depending onforce size).
In order to address the shortfall in per-formance amongst forces with more than4,000 police officers, it is necessary tounderstand what small forces do thatenables them to outperform their largeneighbours. A clue to this may lie in theabove analysis. The correlations observedmay well be influenced by the opportunityin small forces to manage the forensic pro-cess positively and to focus on scenes mostlikely to provide forensic evidence able todetect the crime. Williams (2004) con-cluded that CSE performance was partiallyconditioned by the management of CSEswithin the approach to volume crimeinvestigation. Williams also stated that inforces where scientific staff were seen as‘expert collaborators’, managers were ableto exercise the greatest control over theCSE resource. Although Williams’s researchdid not sample enough forces to commenton CSE deployment as a function of forcesize, his findings would be in keeping withthe suggestion that the profile of forensicscience in a force dictates the amount ofCSE resource and its deployment.
The difference in the magnitude of cor-relation between DNA and fingerprintdetections with force size may be furtherevidence of ‘process management’. Fromthe above analysis, the correlation betweenfingerprint primary detections and forcesize was deemed significant (–0.49),whereas the equivalent for DNA primarydetections was not (–0.11). This differencemay well be attributable to the different
processing methods for DNA and finger-print evidence after the evidence has beencollected at the crime scene. Fingerprintevidence is processed (and any identifica-tions made) in-house by fingerprint staff (insome cases a neighbouring force). As such,there is more opportunity for managers toinfluence the workflow compared withDNA processing which is performed byone of a small number of service providers(independent of the police service)throughout the UK. Doubtless, in a smallforce, there is more opportunity to inter-vene in the fingerprint processing for indi-vidual crimes and to manage the processproactively.
It is unlikely that there is anything inher-ent in a small force that prevents its goodpractice being transferred to a large forcebut in implementing the HMIC review, andin considering different collaborationoptions for filling the protective servicesgap, the Home Office will need to paycareful consideration to the differences inforensic performance of different forces.Not to do so may have a considerableimpact on the resiliency of domestic pro-tective service capability.
ACKNOWLEDGEMENTS
The authors acknowledge the assistance ofMrs Trudy Loe (Northamptonshire Police)with the analysis of the data presented inthis paper. One of us (JWB) gratefullyacknowledges the support of the chiefofficers of Northamptonshire Police.
REFERENCES
BBC. (2005). Clarke Attacked on Police Reform.Retrieved March 24, 2006 from http://news.bbc.co.uk/1/hi/uk_politics/4541012.stm.
Townsley and Bond
Page 191
Harrad, L. (2006). Are Larger Police ForcesBetter than Smaller Police Forces? PoliceProfessional, 44, 24–26.
Home Office. (2001). Policing a New Century:A Blueprint for Reform (White Paper,Cm 5326). London: Home Office.
Nicholas, S., Povey, D., Walker, A., & Kershaw,C. (2005). Crime in England and Wales2004/2005 (Home Office StatisticalBulletin 11/05). London: Home Office.
O’Connor, D. (2005). Closing the Gap: AReview of the ‘Fitness For Purpose’ of theCurrent Structure of Policing in England and
Wales. London: Her Majesty’s Inspectorateof Constabulary.
Stephenson, P. (2004). What Would a Level 2Model of Policing Look Like and What WouldBe the Specifications of the Services that it WouldOffer? (Report to ACPO Police ReformSteering Group (Restricted Access)).London: ACPO.
Williams, R. (2004). The Management of CrimeScene Examination in Relation to theInvestigation of Burglary and Vehicle Crime(Home Office Online Report 24/10).London: Home Office.
Why size does matter when determining optimal Level 2 service delivery
Page 192
