does organizational structure matter investigation centralizat

27
http://pqx.sagepub.com/ Police Quarterly http://pqx.sagepub.com/content/17/3/250 The online version of this article can be found at: DOI: 10.1177/1098611114546229 2014 17: 250 Police Quarterly John D. McCluskey, Jeffrey M. Cancino, Marie Skubak Tillyer and Rob Tillyer Clearances, and Robberies Does Organizational Structure Matter? Investigation Centralization, Case Published by: http://www.sagepublications.com On behalf of: Police Executive Research Forum Police Section of the Academy of Criminal Justice Sciences can be found at: Police Quarterly Additional services and information for http://pqx.sagepub.com/cgi/alerts Email Alerts: http://pqx.sagepub.com/subscriptions Subscriptions: http://www.sagepub.com/journalsReprints.nav Reprints: http://www.sagepub.com/journalsPermissions.nav Permissions: http://pqx.sagepub.com/content/17/3/250.refs.html Citations: What is This? - Aug 12, 2014 Version of Record >> at University of Hong Kong Libraries on October 18, 2014 pqx.sagepub.com Downloaded from at University of Hong Kong Libraries on October 18, 2014 pqx.sagepub.com Downloaded from

Upload: msheerazqaisrani

Post on 12-Nov-2015

220 views

Category:

Documents


1 download

DESCRIPTION

Police Quarterly Volume 17

TRANSCRIPT

  • http://pqx.sagepub.com/Police Quarterly

    http://pqx.sagepub.com/content/17/3/250The online version of this article can be found at:

    DOI: 10.1177/1098611114546229 2014 17: 250Police Quarterly

    John D. McCluskey, Jeffrey M. Cancino, Marie Skubak Tillyer and Rob TillyerClearances, and Robberies

    Does Organizational Structure Matter? Investigation Centralization, Case

    Published by:

    http://www.sagepublications.com

    On behalf of:

    Police Executive Research Forum Police Section of the Academy of Criminal Justice Sciences

    can be found at:Police QuarterlyAdditional services and information for

    http://pqx.sagepub.com/cgi/alertsEmail Alerts:

    http://pqx.sagepub.com/subscriptionsSubscriptions:

    http://www.sagepub.com/journalsReprints.navReprints:

    http://www.sagepub.com/journalsPermissions.navPermissions:

    http://pqx.sagepub.com/content/17/3/250.refs.htmlCitations:

    What is This?

    - Aug 12, 2014Version of Record >>

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • Article

    Does OrganizationalStructure Matter?InvestigationCentralization,Case Clearances,and Robberies

    John D. McCluskey1, Jeffrey M. Cancino2,Marie Skubak Tillyer3, and Rob Tillyer3

    Abstract

    This study examines the reorganization of robbery detectives from a decentralized to

    centralized model in one large department, with the purpose of understanding

    whether organizational structure affects robbery clearance. Time-series regression

    models (Auto-Regressive, Integrated, Moving Average) demonstrate that the

    percentage of investigation-eligible robberies cleared by arrest increased and

    reported robberies decreased subsequent to the reorganization. Additionally, inter-

    views indicated organizational changes in case processing with respect to information

    collection and use, cooperation among detectives and other police, and the police

    prosecutor interface. The organization of robbery detectives appears to be import-

    ant to case clearances and crime reduction. A more comprehensive research

    program on police investigations is recommended to fully assess the costs and bene-

    fits of detective configurations.

    Keywords

    investigations, case clearances, organization, robbery

    1College of Liberal Arts, Rochester Institute of Technology, NY, USA2Department of Criminal Justice, Texas State University-San Marcos, TX, USA3Department of Criminal Justice, University of Texas-San Antonio, TX, USA

    Corresponding Author:

    John D. McCluskey, College of Liberal Arts, Rochester Institute of Technology, 93 Lomb Memorial

    Drive, Rochester, NY 14623-5603, USA.

    Email: [email protected]

    Police Quarterly

    2014, Vol. 17(3) 250275

    ! The Author(s) 2014Reprints and permissions:

    sagepub.com/journalsPermissions.nav

    DOI: 10.1177/1098611114546229

    pqx.sagepub.com

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • Introduction

    There is little consensus whether organizational structure inuences the e-ciency and eectiveness of police work, especially with regard to detectivework (Horvath, Meesig, & Lee, 2001). In particular, it is unclear whethercentralizing investigative units substantially impacts case processing byimproving police clearance rates and impacting crime more broadly. Giventhat previous research (Manning, 1992; Sanders, 1977) suggests that much ofwhat investigators do is information processing, recent advances in informa-tion technology may facilitate this process within a centralized environment.The centralization of personnel and resources, for example, may help to pro-mote more connections among investigators, thus improving pattern recogni-tion. Alternatively, a community policing model suggests that decentralizeddetective units oer the potential to develop stronger ties between detect-ives and the community (Wyco & Cosgrove, 2001). Such relationships,coupled with closer proximity to patrol ocers, might oer detectives beatknowledge about crimes and repeat oenders, thus enhancing crime-controloutcomes.

    Beyond any impact on clearance rates, there appears to be little condenceamong criminologists that investigations (or the organizational structure of theseunits) matter much in terms of crime reduction (Bayley, 1998; Greenwood &Petersilia, 1975; cf., Eck, 1992). In other words, even if centralization improvesclearance rates, it is unclear whether such organizational changes can impactcrime. Recently, however, Braga, Flynn, Kelling, and Cole (2011) challenged thisnotion by arguing that investigators do have the ability to control recurringcrime problems. In short, the existing research on the consequences of investi-gative unit organizational structureboth with respect to outputs (i.e., clear-ances) and impacts (i.e., crime rates)is inconclusive and dated. Given theremarkable advances in information technology, it is plausible that the impactof police organizational structure has changed in recent years. Coupled with therising public pressure to improve performance measures within an environmentof declining resources (Wexler, 2010), an examination of whether organizationalchange (i.e., centralization) can improve case clearances and reduce crime isparticularly timely.

    The present study explores whether changes in investigations organizationalstructure can improve police investigators outputs (i.e., clearances) and impacts(i.e., robbery reduction). Data to address these questions are drawn from the SanAntonio Police Department (SAPD), which recently centralized its robberydetectives. An interrupted time-series design is used to examine the inuenceof robbery unit centralization on case clearances and robberies. We supplementthese ndings with qualitative data collected through a series of interviews con-ducted with a sample of robbery detectives and supervisors. This informationprovides some context for the questions addressed with quantitative data and

    McCluskey et al. 251

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • provides insight into potential mechanisms by which organizational changesmay enhance clearance and crime reduction.

    Literature Review

    As with most American law enforcement functions, the origin of police investi-gation can be traced to the United Kingdom (Klockars, 1985). Over the years,the investigation detail has become a permanent xture and comprises a consid-erable portion of police organizations. The Law Enforcement Management andAdministrative Statistics (LEMAS) survey shows approximately nine of 10 localpolice departments had investigative responsibilities (Hickman & Reaves, 2006),and estimates indicate 16% of sworn police ocers were assigned to investigativeroles (Horvath et al., 2001). Notwithstanding the role of investigations in policework, research on the eectiveness of police investigation is sparse comparedwith the abundance of studies focusing on patrol functions for more than thepast 30 years (Harper, 1991; W. Skogan & Frydl, 2004). One area that hasreceived empirical attention is the central role of information in the investigativeprocess.

    Information is widely recognized as fundamental to much police work(Manning, 1992). W. G. Skogan and Antunes (1979), for example, notethat information is a key determinant in the apprehension of suspects andforms the foundation for investigation. The literature on detectives informa-tion work has generally equated detective work with time spent on dierentactivities. The 1975 Rand study (Greenwood, Chaiken, Petersilia, & Pruso,1975) is usually credited as being the touchstone research in the area of policeinvestigations. The study involved on-site observations across 25 departmentswhere researchers interviewed, observed, and tracked cases of individualinvestigators and supervisors for the purpose of describing daily activities.Results indicate that detectives case work was consumed by administrativeactivity. As Greenwood et al. (1975) conclude, . . .an investigators time ispreponderantly consumed in reviewing reports, documenting lings, andattempting to locate and interview victims on cases that experience showswill not be solved (p. viii). Similar ndings were reported in the UnitedKingdom where much of detectives activities were conducted from theirdesk (Tarling, 1988). A more contemporary detective time-allocation studyby Womack (2007) involving self-report time cards for case work performedby 24 detectives in ve units with the Richardson (TX) Police Departmentrevealed the following:

    Detectives spent the majority of their time (65.6%) on six activities: contacts with

    victims and witnesses (19.9%); supplemental report preparation (17.0%); login and

    review of case (9.2%); computer searches for suspect information (6.7%); suspect

    interviews/interrogations (6.4%); and case preparation (6.4%). (p. 80)

    252 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • This time spent model may have been very eective in describing solo detectiveoperators in a paper world but may no longer reect some detective work in themodern police bureaucracy.1

    Beyond a simple assessment of time spent, Eck (1983) introduced the idea thatcases vary in terms of their solvability. Thus, burglary and robbery investigativeunits ought to engage in a system of triage to eectively manage investigatorstime. More precisely, he conceptualized three information-centered investigationresource scenarios where cases cannot be solved, can be easily solved, and couldpossibly be solved (Eck, 1983, 1992; also see Gaines, Lewis, & Swanagin, 1983).Eck identied information as a central factor in making key resource allocationdecisions in case processing. For example, in cases where no suspect is immedi-ately identied, detectives traditionally focused on the usual suspects, personswhose past behavior and associations led investigators to believe they might beinvolved in the current case, or might know who was involved (M. Maguire,2003). Additionally, Eck reported that, on average, detectives spent 34 hours onrobbery and burglary follow-up investigations and showed that detectives inter-view witnesses in about 75% of cases (see Eck, 1992). More recent studies con-rm that detectives in a medium-sized agency focused signicantly more time(as a proxy for investigative vigor) on those cases that yielded more evidence(i.e., information) from the outset (Brandl, 1993; Brandl & Frank, 1994). Thus,information availability and sharing of this information are crucial to a success-ful investigation, and cases that have better evidence would appear to be a usefulmechanism for targeting police resources.

    The value of information is further heightened given the aordances of tech-nology. Aordances are most easily understood as new patterns of action thatcan emerge from the conguration of organizations and technology (Zammuto,Grith, Majchrzak, Dougherty, & Faraj, 2007). For example, technology hasallowed for visualization of work processes, virtual collaboration, mass collab-oration, simulation, etc. This has transformed the nature and content of workand likely transformed the possibilities of detective work, such that priorresearch describes a paper and typewriter world that no longer exists.

    Despite the dearth of studies focusing on the intersection among technology,detectives, and investigation units, there is a contemporary recognition thatpolicing has undergone a technological revolution involving computerization(Chan, 2001; Danziger & Kraemer, 1985) and the strategic deployment ofresources driven by philosophies/practices linked to geographic information sys-tems, Compstat, intelligence-led policing, and predictive policing (Knobler &Bratton, 2009; Weisburd, Mastrofski, McNalley, Greenspan, & Willis, 2003).This sociotechnical revolution, which postdates the two seminal pieces ofresearch in this area (Eck, 1983; Greenwood & Petersilia, 1975), provides thecontemporary backdrop against which to examine police investigations.Regardless of the period studied, a key premise is that information is fundamen-tal to many stages of police work (Brown, 1988; Chan, 2001; Dean, Fahsing, &

    McCluskey et al. 253

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • Gottschalk, 2007; Manning, 1992; Reiss & Black, 1967), and the digital storage/retrieval of information for investigations should increase eectiveness (seeNorthrop, Kraemer, & King, 1995). Research conducted more recently indicatesthat changes in technology are likely to aect how and what resources aredevoted to investigations and how investigation units may capitalize onnew technology in managing information (Wyco & Cosgrove, 2001). It isfrom the perspective of information and technological advances in gathering,organizing, retrieving, and sharing information that we approach the issue oforganizational structure as a potential inuence on clearance rates and crimereduction.

    Centralized Versus Decentralized Investigations

    American policing originated from centralized authoritarian military preceptscharacterized by routinized supervisorsubordinate relationships that excludedsubordinate decision making (Angell, 1971; Klockars, 1985; Smith, 1960). Thetenets of classical organizational theory were, historically, considered the bestunied approach to the administration of justice (Angell, 1971). Researchershave described the centralized closed perspective as a rational model (Weber,1946), goal model (Packer, 1968; Skolnick, 1967; Smith, 1960), or rationalgoal model (Feeley, 1973), respectively. They share the implication that cal-culable rules (not people) attain means and goals and assume an elaborateapparatus that, for example, processes arrests consistent with well-denedrules and procedures (Feeley, 1973, p. 410). For example, Smith (1960)argues that an overarching benet for centralized police is the pooling ofmanpower under a unied structure that stimulates coordination of eortand reduction of personnel demands. He cautions, however, that centraliza-tion is likely to be most benecial when reorganization eorts aect smallindependent units conducting similar activities. Thus, success of centralizationdepends on whether (a) smaller units engage only the partial services ofindividuals whose spare time would not be devoted to other tasks if thestructure were centralized, (b) several small, related units are brought undersingle supervision to reduce lost motion, and (c) administrative focus providesa channeling eect on activities (p. 258).

    Centralized investigative units (i.e., detectives are housed in a single location)are the norm within current law enforcement agencies (Horvath et al., 2001).Although centralized structures (e.g., rational/goal models) provide a degreeof objectivity for evaluating the eectiveness of organizations, centralizationskey weakness is hypothesized to be the disruption of complex behavioral pro-cesses. Disruption occurs through overemphasis on rules and end-goals andde-emphasis of the human element in shaping the behavior of actorswithin organized social units (Etzioni, 1960, 1961; Simon, 1947). This realizationinuenced the community policing era suggestion that decentralization of

    254 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • neighborhood investigations might be a useful strategic change for police depart-ments. The trend of geographically decentralizing detective units has been moti-vated at least in part by the goal of developing stronger links betweendetectives and the community, both directly and by working more closelywith the patrol ocers who serve a particular area of the city (Wyco &Cosgrove, 2001). The recognition of the importance of patrol ocers contri-bution to investigation has led to eorts in cross-training patrol ocers andencouraging greater involvement in investigative work (e.g., Kenney, White,& Runengo, 2010). Given that much of the information about crimes,repeat oenders, and beat knowledge is found at the level of patrol ocerswithin police districts, centralization may eliminate important networks ofknowledge and information from percolating to investigative units. Roberts(2008), however, used multilevel event history analysis to examine robbery,forcible rape, and aggravated assault and reported that the number of inves-tigators and community policing ocers did not have a statistically signicantimpact on clearances for rape, aggravated assault, and robbery.2 In theory,decentralized police experience relaxed rules, thereby placing decision makinginto the hands of subordinates who collaborate as socially organized units toexpand organizational rationality. Therefore, one could argue that centraliza-tion may yield a net negative outcome for clearance rates, if, in fact, theproximity to beat-level information is vital to investigations.

    Conversely, centralization arguably encourages investigators to share infor-mation about patterns across a larger tapestry and not be restricted to ageographically circumscribed subset of cases. This format would be consistentwith a sense of investigation that relies on identifying patterns and recognizesthat oenders may travel across the geographic boundaries specied by manypolice agencies as beats, patrol areas, or neighborhoods. Therefore, central-ization of resources may serve an important purpose of setting up a moreholistic perspective of crime patterns, investigative resource capacity, andconstraints on action. In the context of contemporary detective work, onemight speculate that collaborative, centralized workspace and command ofdetectives, focused on similar cases, may yield a dierent outcome than thepessimistic detective work adds little value prediction. Braga et al. (2011) fore-shadow this thinking in their analysis of changes in Milwaukee, New YorkPolice Department (NYPD), and other departments that worked to integratetechnology and adopt more of a problem focus aimed at crime control out-comes. Ultimately, the benets of centralization would be apparent in greaterclearance rates and perhaps a reduction in crime.

    The pervasive belief that the organizational structure of investigative unitshas little to no impact on clearance rates may explain why the inuence ofinvestigation on crimes has not received much contemporary attention inthe criminological literature. Clearance patterns, proactive police, and theirassociation with crime over time have been investigated at the city level

    McCluskey et al. 255

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • (Kubrin, Messner, Deane, McGeever, & Stucky, 2010; MacDonald, 2002;Wilson & Boland, 1978), but there is scarce information about how themechanism of investigation translates (or fails to translate) into changes incrime. There are at least two potential mechanisms by which organizationalstructure might indirectly impact robberies via increased clearance rates. First,incapacitation of high-rate robbers would reduce future robberies. Withgreater apprehension, the capture of high-rate robbers is likely to occurmore frequently and, as long as these individuals are in custody, their eect-ive rate of oending falls to zero. Literature on the rate at which robberscommit oenses suggests that high-rate oenders are especially prolic. Forexample, inmate surveys have indicated that in Michigan, California, andTexas, imprisoned robbers had annual oending rates of 77, 53, and 9,respectively (Chaiken & Chaiken, 1982; cf., Visher, 1986). Similarly, researchon active robbers interviewed in St. Louis, Missouri, suggested that about onein ve committed at least one or more robberies per week, yielding an annualrate of 52 or more (Wright & Decker, 1997).

    The second possibility is that oenders are deterred from committing crimesas the possibility of apprehension increases. As more robbers are arrested fortheir crimes, the perceived probability of capture increases among the oenderpopulation. Thus, some number of robbers may abandon the crime and somerobbery opportunities that are deemed too risky will also be abandoned. Durlaufand Nagin (2011) have argued that the increase in probability of being caught,which essentially manipulates the concept of certainty in deterrence theory, is aplausible mechanism for reducing crimes.

    In summary, an optimistic perspective on prior research suggests thattriage/screening routines should increase productivity on cases worked (Eck,1983), and technology applied to investigations should make for more ecientand eective use of information. Given the modern aordances accompanyingtechnology, we posit that the centralization of robbery investigations willimprove both outputs (i.e., clearances) and impacts (i.e., robbery levels).The centralization of resources and personnel should promote more connec-tions among investigators and form greater pattern recognition when com-pared with geographically decentralized robbery investigations. In eect, thedecentralized model operates with investigators possessing a one-to-manyrelationship with cases, whereas the centralized model encourages a many-to-many orientation between investigators and cases. Nevertheless, the relativeecacy of these two strategies (decentralized vs. centralized) has not beenthoroughly explored in the context of contemporary practice, and an alter-native interpretation of the extant literature might suggest centralization willlead to lower clearance rates and increases in robbery incidents. In 2009, theSAPDs robbery unit moved from a decentralized structure to a centralizedstructure, providing an opportunity to observe how such organizationalchange might inuence both police outputs and impacts.

    256 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • Data and Methods

    Site

    San Antonio is the primary city within Bexar County, Texas and covers522 square miles. According to the 2010 Census, San Antonio has 1.3 millionresidents of which 230,000 were added in the past decade. San Antonio is apredominately Hispanic city (61.5%), with White citizens representing 28.5%of the citys population, and Black and Asian citizens comprising 6.9% and2.1% of the population, respectively.

    The SAPD has an authorized sworn strength of 2,374 ocers, with over360 detectives. In recent years, calls for service to the SAPD have exceeded1 million. According to the Texas Department of Public Safety, the SAPDaveraged 2,469 reported robberies per year from 2005 through 2009, and theclearance rate during the period was 18%. By comparison, police in the nearbycity of Austin averaged 1,349 robberies recorded per year and cleared approxi-mately 29% during the same period. Robbery clearance rates for the entire stateof Texas and the United States were approximately 25% and 26% duringthat same period. Thus, the robbery clearance in San Antonio substantiallyunderperformed other agencies prior to the 2009 centralization.

    Under the decentralized investigation system in place prior to October 2009,robbery detectives were assigned to investigations in each of the departments sixsubstations. Robbery investigations usually comprised three or four detectiveshoused along with property detectives and a contingent of more than a dozenother investigators who were responsible for investigations of burglaries andother crimes. All detectives were managed by a sergeant, typically without asubstantial investigation background.3 In October 2009, the robbery investiga-tion was centralized and consisted of 21 detectives, supervised by two sergeants,one lieutenant, and one captain.

    Data

    Data on 15,475 robberies were recorded between January 1, 2005 and November30, 2010 to determine whether robbery investigation patterns changed after theconsolidation of the robbery unit in October 2009. Thus, there is a 14-monthfollow-up period to explore whether a signicant change occurred in clearingcases by arrest, of at least one suspect. The purpose for isolating arrests in theprimary analysis is twofold. First, it represents a completed investigation thatpasses an individual to the next stage of criminal justice processing. Second,arrest and punishment are more likely to be associated with theoretically pro-posed crime control eects that can come from investigations (Braga et al.,2011), whereas closures resulting in no arrest are less likely to yield such eects.Specically, apprehension could result in a specic deterrent eect for the indi-vidual as well as the incapacitation of potential high-rate oenders. Changes in

    McCluskey et al. 257

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • the probability of arrest, from an economic perspective, are precisely the leversthat policy analysts suggest manipulating (Durlauf & Nagin, 2011; Sherman,1990, 2011) to increase the certainty of punishment. Hence, it is arguably abetter test of the eectiveness of investigations to limit analyses to those caseswhere an arrest has been recorded as the measure of clearance that is mostclosely tied to crime control.

    Cases with more than six identied suspects were excluded from analysis ofrobbery patterns to standardize robbery investigations. Those cases with morethan six perpetrators identied represented 134 cases with a mean of more than10 individuals involved. Though classied as robberies in a technical sense, thecut point of six was chosen because there was a sharp drop-o in cases at thatpoint, the nature of these events seemed rare, and these cases created dicultiesfor compiling electronic les to analyze the data. Additionally, it is plausible thatgroup or crew robberies, involving 10 individuals, for example, are substantiallydierent from those robberies that have ve or fewer perpetrators involved. Thendings reported below should contain no biases, as these types of events werespread across all observed periods.

    Next, cases were examined to determine whether an on-scene arrest wasmade. Such incidents are largely due to the eort of patrol, and the apprehensionand arrest of suspects on scene is not creditable to robbery investigators. Thisdoes not discount the possibility that information ow from the robbery unitmay make on-scene apprehension more likely, but it is reasonable to removethese cases, as a class, from the primary consideration of investigative work assolving the case. There were 856 instances of on-scene arrest among the 15,475incidents, or approximately 5.5% of cases. Similarly, 400 unfounded cases wereexcluded from consideration, though unfounding represents investigative work inthe sense that the legal elements, from the report, are determined to have notbeen satised. A further 35 cases were also eliminated from consideration due toambiguous administrative status in the SAPD records management system.A total of 1,291 cases were excluded by adopting the aforementioned decisionrules, or 8% of all incidents. This left a total of 14,184 investigation-eligiblerobberies during the 71 months of observation, including 14 months of postcen-tralization robbery investigation. The percent of cases cleared was calculated forthe cases recorded every month, such that if a case was cleared 6 months after theinitial report, it was counted toward the proportion for that original month inwhich it was rst reported, not the month in which it was cleared. This is asubstantial advantage over using monthly Uniform Crime Report data, as themonthly clearance and reported crime data represent work achieved during theparticular month.

    Figure 1 illustrates the monthly percentage of cases that resulted in investi-gator-initiated clearance by arrest. The 57 months prior to centralization areseparated from the 14 postcentralization months by the vertical line.A comparison of mean levels of the two series indicates that for the 57

    258 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • months prior to centralization, an average of 8.3% of monthly investigatedrobberies were cleared by arrest. In contrast, the 14 months following central-ization indicate that the mean clearance rose to 11.6%, an increase of 3.3% inthe clearance rate. If this is interpreted as the risk of capture, the increase from8.3% to 11.6% represents a nearly 40% increase in the probability of capture(3.3/8.3 0.4) under the centralized investigation format.

    Analysis

    Comparing the raw percentages indicates that there was some change in clear-ance by arrest from precentralization to postcentralization eorts. A formalstatistical test is required to determine whether this change in clearance is dif-ferent from chance variation. First, a t test was estimated across the two periodsto test whether the means were dierent. This yielded a statistically signicantcontrast between the periods (t4.81, p< .05), which indicated that the dif-ference was likely a real change in the probability of arrest.

    As the data are arrayed over time, there is the possibility of trend and auto-correlation (i.e., observations are not independent of one another). If this is thecase, then the t test would be inappropriate for drawing an inference about thedierences between pre- and post-centralization of investigation. A set ofmodels, Auto-Regressive, Integrated, Moving Average (ARIMA), allow forthe identication and removal of trends and autocorrelation from the data.These more rigorous models are suggested for testing the hypothesis of whetherthe change in organizational structure to a centralized unit substantially inu-enced arrest clearances.

    0

    2

    4

    6

    8

    10

    12

    14

    16

    1 3 5 7 9 11131517192123252729313335373941434547495153555759616365676971

    % C

    lear

    ed b

    y A

    rres

    t

    Month

    Centralizaon

    Figure 1. Monthly percent of eligible cases yielding an arrest, January 2005 to November

    2010 (with trendline).

    McCluskey et al. 259

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • McCleary and Hay (1980, see also, Box & Jenkins, 1976) suggest an identi-cation, estimation, and diagnosis approach to ARIMA, which is followed here.First, the raw series was subjected to a DickeyFuller test and determined to bestationary. Next, the autocorrelation function (ACF) was examined and indi-cated that an ARIMA (1,0,0) process was likely associated with the pattern.More simply, a term was required because prior monthly realizations of theseries inuenced the current month; thus, this would correct for that potentialproblem. An ACF was computed for the residuals for ARIMA (1,0,0) model,and the BoxLjung Q-statistic (2 23.3, 24 df, p .50) indicated that the sys-tematic variation was removed by including this term. Next, a regression wasestimated including a transfer function set at zero for precentralization monthsand one for postcentralization periods. This represents the change in the meanlevel of the percentage of cleared cases across periods. The ARIMA model,summarized in Table 1, conrms both the simple percentage comparisondescribed earlier and the t test. There was a signicant increase in the percentageof cases cleared by arrest for the 14 months observed under centralized investi-gations that persisted once systematic components were removed. More pre-cisely, it appears that the 3.3 dierence is not due to chance but insteadrepresents a signicant change in the probability of arrest for robberies com-mitted after the investigations were centralized. Thus, the centralized robberyunit added substantial value to the criminal investigation of robberies in its newconguration.

    Does centralization have a crime control effect? An analysis of robbery incidents. Giventhe increase in the probability of apprehension accompanying centralization ofrobbery in San Antonio, we next explored whether a detectable decline in rob-beries accompanied this change. To this end, we compiled all robberies on aweekly basis for the 59 full weeks following the October 2009 formation of thecentralized robbery unit. The rst full postcentralization observation week com-menced on Saturday, October 3, 2009 and ended the following Friday. The last

    Table 1. ARIMA Model of Postcentralization Monthly Clearance

    Percentage by Arrest.

    b SE t CI

    Intercept 8.30* 0.36 22.72

    Centralized 3.30* 1.08 3.04 [1.2, 5.4]

    AR(1) 0.22 0.15 1.46

    Model 2 11.1 (2 df)*

    Note. ARIMAAuto-Regressive, Integrated, Moving Average; CI confidenceinterval.

    *p< .05.

    260 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • observation week was the last full week (Saturday through Friday) in November2010. Unfounded robberies were removed from the weekly totals. This yielded59 total postcentralization observations consisting of counts of all recordedrobberies for each week. For a precentralization series, a period twice as long,or 118 weeks, was observed. This period commenced on Saturday, June 30, 2007and encompassed the following 118 weeks of precentralized investigations as acontrol series.

    The weekly time aggregation was chosen to maximize the postcentralizationobservations and allow for the eects of deterrence and incapacitation to playout over a greater number of observations. Because this is a theoretical eectbased on robbery detectives increased eectiveness, it would permeate into thecommunity of potential robbers in the form of incapacitation (some arrested thatotherwise would have remained active) and deterrence (a choice to abandon amarginal robbery opportunity may be made due to greater risk of apprehen-sion). These microdecisions and variations would seem to be more detectable insmaller units of analysis, and the pool of active robbers would be more com-parable across a 2-year period (approximately covered by our 118 weeks)as compared with the long monthly series used in the previous analysis.

    For the weekly observations prior to centralization, shown in Table 2, themean number of recorded robberies was 53.90 and the postcentralizationmean was 47.07 recorded per week, which represents 6.83 fewer per week. Thedierence is statistically dependable according to a t-statistic (t 3.87, p< .05)calculated for this mean comparison.

    As discussed earlier in the arrest clearance analysis, data arrayed over timepresent the possibility of trend and autocorrelation. Therefore, we employed thesame identication, estimation, and diagnosis approach to ARIMA describedearlier to test the inuence of centralization on the number of recorded rob-beries. First, the raw series was subjected to a DickeyFuller test and determinedto be stationary. This is the desirable property of a series, which indicates it doesnot exhibit a substantial trend, and can be analyzed for autoregressive andmoving average components. Next, the ACF was examined and indicated thatan ARIMA (2,0,0) process was likely associated with the pattern. More simply,two autoregressive terms were required because prior monthly realizations of theseries inuenced the current week; thus, this would correct that problem and

    Table 2. Mean Levels of Robbery in Pre-and Postcentralization Periods.

    N Minimum Maximum M SD t-statistic

    Precentralization 118 20.00 82.00 53.90 11.29 3.87*

    Postcentralization 59 22.00 73.00 47.07 10.60

    *Note. p< .05.

    McCluskey et al. 261

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • allow for an estimation of the centralization eect on recorded robberies.Consistent with identication and diagnosis practices, an ACF was computedfor the residuals for the ARIMA (2,0,0) model, and the BoxLjung Q-statistic(2 43.8, 52 df, p .78) indicated that the systematic variation was removed byincluding these terms. Next, a regression was estimated including a transferfunction set at zero for precentralization months and one for postcentralizationperiods. This represents the change in the mean level of weekly recorded rob-beries across the two periods. The ARIMA model, summarized in Table 3,conrms a signicant decline in recorded robberies subsequent to centralization.There was a signicant decrease in weekly robberies recorded for the 59 weeksobserved under centralized investigations that persisted once systematic compo-nents were removed. More precisely, it appears that there was a substantialdecrease of robberies recorded (i.e., 7.06 incidents per week), and this dierenceis not due to chance. This estimate, during a typical year (52 weeks of observa-tion), yields an expected net decrease of 367 robberies annually.

    Given that the internal validity of our time-series analysis may be threatenedby unknown history eects (Shadish, Cook, & Campbell, 2002), we also usedresidential burglary as a comparison series. Consistent with Wilsons (1968)work on police styles/organization, we are keenly aware that clearances varyby crime type. Thus, we conducted an additional analysis to observe the rela-tionship between centralization and residential burglary to ensure that anychange in robbery following centralization was not reecting broader crimetrends in San Antonio during the study period. Residential burglary was selectedbecause, like robbery, police presumably exercise less discretion in addressingthis serious property oense. Stated dierently, police tend to exercise consider-ably more discretion for victimless crimes associated with drunkenness, dis-orderly conduct, and driving while intoxicated (Wilson, 1968). A morecomprehensive discussion on internal validity and police discretion is outlinedlater.

    Table 3. ARIMA Model of Postcentralization Weekly Recorded Robberies.

    b SE t CI

    Intercept 53.99* 1.37 39.33

    Centralized 7.06* 2.36 2.99 [11.7, 2.4]AR(1) 0.12 0.08 1.40

    AR(2) 0.15* 0.07 2.14

    Model 2 17.9 (3 df)*

    Note. ARIMAAuto-Regressive, Integrated, Moving Average; CI confidence interval.*p< .05.

    262 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • A parallel ARIMA analysis involving residential burglaries reported acrossthe 118 weeks preceding centralization and 59 postcentralization weeks is pre-sented in Table 4. Examination of the series indicated it was stationary, and anARIMA (1,0,0) was sucient to prewhiten the model consistent with theapproach taken with weekly robbery series. More simply, the ARIMA compo-nents remove systematic time variations from the series that would distort thestatistical test of the intervention analysis. The analysis of the control seriesyielded a positive nonsignicant slope for the intervention period (b 0.76,p .94) indicating a nominal weekly increase in the postcentralization period.It should be noted that this increase, relative to the mean of the series, is less thanone half of a percent (0.76/242) of the weekly mean. This result from the controlseries further supports the inference that the observed eect in the robbery modeldoes not represent an underlying change in overall crime patterns inSan Antonio. Rather, the robbery series distinct contrast from residential bur-glaries buttresses the link between changes in apprehension and resulting incrime-specic changes in robbery documented earlier.

    To our knowledge, there was no other intervention that coincided with thecentralization of the robbery unit that might explain the signicant relationshipbetween centralization and the increase in clearances and the decline in rob-beries. However, one threat to the research design adopted here is that a con-founding variable accounts for the results obtained. With regard to clearancesrising in the monthly series, we argue that the centralization of detectives couldhave led to a decrease in clearance rates; unless systematic change was under-lying the reorganization of detectives, it would be unlikely for the investigatedcases to yield a statistically reliable increase in the short monthly postcentraliza-tion series. Additionally, with the time-series analysis approach, the estimatedmodels conservatively assign months prior to October 2009 to the pretreatmentperiod. However, a separate research project has uncovered robberies in thepreintervention period cleared by the centralized detectives. With regard to rob-bery, it seems that linking the centralization to robbery declines is more tenuous

    Table 4. ARIMA Model of Postcentralization Weekly Recorded

    Residential Burglaries.

    b SE t CI

    Intercept 242.72* 5.35 45.32

    Centralized 0.77 10.48 0.07 [19.7, 21.3]AR(1) 0.52 0.06 9.09

    Model 2 83.9 (2 df)*

    Note. ARIMAAuto-Regressive, Integrated, Moving Average; CI confidenceinterval.

    *p< .05.

    McCluskey et al. 263

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • for two reasons. First, the mechanism leading from clearance increases to rob-bery decreases is not clear, but the trace of something of consequence happeningto the weekly robbery series postcentralization is present in the data and absentin the burglary control series. Is it a message of deterrence or incapacitation?This is unknown and limits us to speculation regarding the exact cause within therobbery units centralization. In the following, we consider the reorganizationfrom the detectives perspective with an aim at learning how reorganization mayhave aected outputs and crime rates.

    Putting Centralization in Context: Robbery Investigator Interviews

    Understanding how the centralization of investigations aects the work ofdetectives vis-a`-vis the day-to-day case processing is an important considerationnot captured easily by institutional record management systems and the preced-ing analyses. Only investigators themselves or persons observing daily routinescould oer a perspective on what the work of the units were as decentralizedunits and how it transformed under centralization. To that end, a series ofinterviews with six of the 21 robbery detectives in the centralized SAPDRobbery Unit were conducted in July 2011, and a subsequent interview withone supervisor provides a sense of what, if anything, changed under reorganiza-tion. The sample is best considered a convenience sample, but we arrangedinterviews with robbery detectives who came from four of the subcommandsunder the decentralized system, so that we could inquire about idiosyncrasiesregarding robbery investigation under that model. An interview schedule draw-ing elements from prior investigation research focused detectives on broad areasof changes from decentralization to centralization that had positive and negativeimpacts on the investigation process. Specic questions also probed the eect oninformation ow and supervision routines.

    The following should be considered as providing a context, or an enumerationof elements, that may explain why centralization might yield a change in clear-ance rate in the study period and suggest areas for future research on policeinvestigations. These interviews should not be treated a formal theoreticaltesthowever, they do oer some insight into the black box of organizationalchange. More simply, they provide further context for understanding the qualityof the changes underlying the reorganization of robbery detectives in thedepartment.

    A series of questions solicited responses that would help understand theperceived benets centralization oered and to consider the relative benets ofdecentralization. Common themes emerged from the conversations with vecommenting that communication and information sharing improved greatly.

    The general concept of communication was related to an expanded know-ledge base, and two respondents specically noted that centralization allowed forimprovement with information sharing across substations and ease of identifying

    264 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • serial or spree robbers. Hence, having all robbery detectives together changedthe nature and ow of information available to any particular robbery detectivein the unit. One respondent commented that the decentralized system had noenvironment for sharing. . . information ended up in a dead end and that thecurrent system amounted to a pool of information among detectives(Respondent A). As Manning (1992) and others have noted, information isthe root of police work, and especially investigations (Sanders, 1977). In sum,the responses indicated that the increase in information quantity, quality, andsharing was nearly universally recognized among the detectives interviewed.

    Two other themes that emerged from the concept were team motivation andsupervision. Among the specic elements related to teamwork, several noted thatthe consolidated unit, with greater manpower (21 detectives vs. three in a sub-station), allowed for cases to be workedmore intensely to the point of arrest. Onerespondent noted that everybody helps each other out and everybody wants tobe here (Detective B). There was consensus that the teamwork, focused task,and supervision that stayed on task were important to the consolidated unit.Detective C mentioned that, under the decentralized system, supervisors didntgive a ip about focusing on robbery. Under that system, if shorthanded,detectives were assigned to phone duty in the substation. This team approachis embodied by supervisors in the centralized unit who mentioned that decen-tralized cases were strictly an individual detectives. If a detective was out onvacation, no one has access to the case. If he was not working on, no one is!Under the centralized system, detectives did have responsibility for cases, butinformation was more readily accessed and cases could be worked without thelead investigator present.

    The overarching eects of centralization on the entire units function weregleaned from a follow-up interview with one supervisor. He reinforced thechanges discussed by the detectives in terms of developing internal routinesfor information gathering and shared ownership of cases. In addition to theseinternal changes in information ow and routine, there were also signicantchanges in interorganizational or external functions related to robbery case pro-cessing that were realized with the consolidation of robbery detectives. Therobbery unit interfaced directly with crime scene technicians and provided expli-cit instruction and training protocols for evidence collection at robbery scenes.With respect to external changes, the robbery unit developed prosecution guidesfor cases resulting in arrest. These were systematic case evidence presentationguides that detectives provided for prosecutors. Furthermore, attorneys from theprosecutors oce established email feedback with the unit on cases statuses,met with the centralized robbery unit intermittently on successful and unsuccess-ful cases, and suggested improvements that would increase the probability ofconviction. Such eort would, potentially, translate into the possibility forgreater incapacitation eects, if, in fact, centralization facilitated a moretightly coupled criminal justice response toward arrested robbery suspects.

    McCluskey et al. 265

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • The preceding analysis of data on case clearances and recorded robbery suggeststhe promises of centralization, as voiced by detectives and their supervisor, thatmay have led to detectable changes in investigation outputs and impacts.

    Centralization should not been seen as wholly without a degradation of somefacets of the detectives work. Interviewees who had worked under decentralizedcommands mentioned it was convenient to deal with cases that were in close geo-graphic proximity to robbery incidents. Several respondents also mentioned acloser working knowledge of, and rapport with, substations patrol, andgiven the importance of this aspect in discussions of community investigations(Wyco & Cosgrove, 2001), the result is unsurprising. However, one ocerdirectly noted that Patrol, as an asset for solving crime, is not that big. Inanother segment of the interview, we queried directly the inuence of patrolocers on investigation and the consensus was that they were somewhat helpful,but this assistance was not greatly impaired by the centralized conguration.

    When asked about the weakness within the centralized system, two respond-ents mentioned the proximity to cases they were working, one mentioned geo-graphic specialization, and another mentioned limited personal interaction andknowledge of ocers in substations. A follow-up question regarding the fre-quency and quality of patrol interaction solidied the sense that it was muchcloser and personal under the decentralized system, but that information andcommunication by radio meant that information still owed to investigationsfrom patrol areas. Consistent with Wyco and Cosgroves (2001) hypothesis,there was a sense that decentralized detectives had a greater sense of ownershipin an area, familiarity with patrol, and knowledge of local problems in SAPD.

    However, the value of these advantages may be oset in the comments of allsix detective respondents and their supervisor who noted that the critical mass ofdetectives allowed for making in-house arrests, setting up surveillance, and shar-ing information across geography. More specically, in the decentralized system,detectives could rarely make arrests, but instead led warrants. Now, the unitmembers access information from a shared drive, can work collectively on casesrequiring surveillance, and directly arrest known suspects.

    Thus, the picture that emerges operates on two levels. First, information ismore permeable in the centralized command, with shared computing archivesand personal interaction. Second, the supervision built the units work around ateam ethos for working cases and collective responsibility (investigative) andaction (arrest teams) for executing their mandate.

    As mentioned, the tightened linkage between prosecution and robbery ishypothesized to have led to substantially higher quality arrests and presumablymore serious consequences for oenders. Anecdotal evidence from conversationswith police supervisors indicates this to be the case. A postcentralization robberydecline may have been initiated by the change in the interorganizational inter-face, which yielded incapacitation and deterrent eects. Therefore, robbery cen-tralization may only be necessary to the extent that systematic feedback in the

    266 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • local criminal justice system required the mass of detectives to easily interfacewith a small number of prosecutors. From this perspective, the logical next stepis to examine if and how centralization transferred its change in quality andquantity of arrests to prosecution outcomes in Bexar County Texas courts.Testing how prosecution (and the prosecutiondetective interface) changedempirically, after this reorganization, is beyond the scope of the current articlebut consistent with the idea of the new criminal justice and a focus on the func-tion of interlinked systems at the local level (Klofas, McGarrell, & Hipple, 2010).

    Discussion

    The study of a police unit in transition from one conguration (decentralized orcommunity allocation) to another (centralized or a hive model) is an importantcontribution during a time when police organizations are seeking to optimizeresources as public revenues shrink. The present study observed how the restruc-turing of an investigations unit inuenced police outputs and impacts from theperspective of detectives and supervisors as well as empirically testable changes.Specically, we examined whether the transformation of the SAPD robbery unitfrom decentralized to centralized improved case clearances and reduced robberyincidents. Historically, police scholars have been rather skeptical about theimportance of police investigations in solving cases and reducing crime.Similarly, organization theorists and the community policing movement havemade strong arguments consistent with decentralization of police functions. Wehave argued, however, that in an era of technological aordances, robbery unitcentralization may aid detectives in their information work.

    Our ndings challenge the existing notion that police organizational changesmatter little in terms of police outputs and impacts. Not only was robbery unitcentralization associated with a statistically signicant increase in case clear-ances, but there was also a signicant reduction in robbery observed. Further,additional analysis revealed that residential burglaries experienced a nonsigni-cant increase postcentralization, lending modest support to the conclusion thatthe observed reductions in robbery were not part of some larger city-wide crimetrend. This suggests that investigation centralization, at least with regard toserious or major crimes like robbery, is an approach in need of deeper scrutinyin the current policing environment.

    Interview data indicate that detective contact with the community, undertypical command structures, does not seem to have been vital under the decen-tralized system. Imagining that there is a rote set of skills for investigatorsignores that much of police work is learning by doing, and in the consolidatedunit, best practices and the advantages of technology can be shared. In decen-tralized units, knowledge and investigative skills did not appear to cross sub-station boundaries. These are tradeos recognized by prior research (Wyco &Cosgrove, 2001), but with unknown impacts on the quality and eectiveness of

    McCluskey et al. 267

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • investigative work. In an age of expanding technology, it seems that the cross-pollination of good ideas, routines, and processes for managing and executingarrests of robbery suspects in a centralized unit represents a qualitative improve-ment in investigative eciency. It should also be noted that the sketch of rigidcentralized police hierarchy drawn from the literature (e.g., Smith, 1960) wastransformed into something more akin to a team environment in the recentrali-zation of the robbery unit. The statistical analyses do not clearly demarcate whataspect of centralization is most vital to increasing clearance rates. Further, asingle quasi-experiment oers only tenuous support for centralization being theprimary causal mechanism at work in the results reported earlier.

    Though the analyses suggest that investigative centralization inuences clear-ances, and in turn, robbery, the research methodology employed herein does notpermit us to draw rm conclusions about the mechanism by which improvedcase clearances might reduce robbery. As we discussed earlier, both deterrentand incapacitation eects are possible. That is, the improved performance of therobbery unit might alter oender decision making, as potential robbers note thechanges in risk associated with this particular crime type. It is also plausible thatthe increased clearances led to the incarceration of high-rate oenders, therebyreducing the overall city-level robbery postcentralization. Though our city-leveltime-series analyses do not allow us to examine the extent to which either or bothof these mechanisms might be operating in San Antonio, we do speculatethat the latterincapacitationis more plausible. Case clearances did signi-cantly increase following centralization; we question, however, whether the aver-age oender would be aware of such changes in risk. There were no publicservice announcements or contact with the oender population to advertisethe new strategy. In fact, the SAPD was uncertain of whether the organizationalchange would improve case clearances, consistent with the mixed ndings onederives from the extant literature. Though the increase in probability of capturefollowing centralization rose by 40%, the change in risk does not seem substan-tively meaningful (from 8.3% to 11.6%) when considered in the context ofoender perceptions about the risk of detection and apprehension. We aredoubtful that such a marginal change in riskalbeit, a statistically signicantonecould be perceived by oenders in the absence of some noticationstrategy.

    Conversely, it seems more plausible that the centralized format, which allowsdetectives working cases distributed across city to share information andimprove pattern recognition, would enhance clearance of serial robberies (i.e.,repeat oenders). Future research is needed to examine whether the additionalcase clearances associated with centralization were primarily part of serial rob-beries. In other words, it might be that centralization allows detectives to makeslightly more arrests, and that these arrests are of highly active oenders, thusaccounting for the substantial drop in robberies. This idea is consistent withBraga et al.s (2011) assertion that police investigators have the potential to

    268 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • serve in a problem-solving capacity if they, in fact, target repeat oenders whoare responsible for a disproportionate number of crimes.

    As mentioned earlier, centralization may serve as an information hub fordetectives ability to recognize an active (or repeat) oenders crime series.Centralization may facilitate recognition of the characteristics associated withthe crime series, which in turn, can inuence serial clearances. The benets ofcentralization may resonate more so when considering commercial serial rob-bers. While research indicates that high-volume versatile oenders (less experi-enced) have a greater probability of getting arrested, specialized oenders(expertise in certain types of crimes) yield a smaller probability of being arrested(Clare, 2010). In essence, experience associated with a certain type of crime leadsto greater expertise. It is within this context that centralization might be a moresuitable organizational approach toward achieving increased clearances, andfewer robbery incidents regarding both versatile and specialized robbery oen-ders. Detectives working collectively may more easily distinguish criminal mar-kers and characteristics within- and between-serial strings, thereby increasingeciency and eectiveness (Woodhams & Toye, 2007).

    Despite these promising ndings, we caution against generalizing to investi-gative units focused on other crime types. The status of robbery as a relativelyrare serious crime makes for a case ow that is manageable and a triage systemfor the allocation of investigative resources possible. Crimes that occur withgreater frequency (larceny, aggravated assaults) and where victims are unlikelyto have contact with the oender (auto theft, burglary) represent a dierent setof circumstances than the face-to-face nature of robbery. Thus, centralizinginvestigation of other crimes may require caution and consideration of howand whether technology can increase eciency.

    Whether operating under the paradigm of community policing, intelligence-led policing, problem-oriented policing, or some combination thereof, the newreality for police managers is one of an entrepreneur (Ackroyd, Harper, Hughes,Shapiro, & Soothill, 1992; Chan, 2001; Wexler, 2010). Managers must initiateorganizational change, while assuming some level of risk due to budget cuts thathave handicapped manpower (e.g., ocer layos). Use of technology has helpedfacilitate this entrepreneurial spirit by encouraging managers to employ tech-niques of private for-prot corporations, emphasizing cost control, performanceindicators, and risk assessment; but technology also imposes new forms ofaccountability in terms of cost eectiveness, probity, and procedural regularity(Chan, 2001). Our ndings suggest law enforcement can leverage existingresources organizationally in a way that improves outputs and impacts.The ndings support the contention that modern aordances accompanyingtechnology, coupled with the centralization of resources and personnel,may establish more connections among investigators and promote greater pat-tern recognition when compared with geographically decentralized robberyinvestigations.

    McCluskey et al. 269

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • Limitations and Future Research

    The primary objective of the current research was to examine how a single policeorganization changed investigative practice over time and whether such changeyielded a detectable inuence on robbery clearance rates. Consequently, our unitof analysis was the organization and specic points in time (e.g., months andweeks). Understandably, our research question, units of analysis, and measuresyield limitations pertaining to generalizability inasmuch as there are no com-parison cities and departments. While the inclusion of comparison cities anddepartments is essential to understanding the complex nature of police organ-izations and structure, police organizations possess similarly generic character-istics, yet exhibit a wide spectrum of structural and operating processes(Scott, 1992).

    Internal validity of the results is threatened by a combination of organiza-tional and individual police discretion that can yield biased robbery clearances.Directly accounting for the discretionary nature of police work would furtherhelp to theoretically conrm whether classical models of centralized police struc-tures (Feeley, 1973; Packer, 1968; Smith, 1960) minimize ocer discretion com-pared with decentralized police organization (Wyco & Cosgrove, 2001).The existence of such discretionary bias would undermine the causal relationshipbetween centralization and robbery clearance rates (e.g., McCleary,Nienstedt, & Erven, 1982). Unlike random bias in police performance measurestypically found in arrest policies and recording practices, systematic bias (error)can present itself in a variety discretionary stages of individual police work, notjust arrest (Goldstein, 1977). Furthermore, aggregate patterns that accompany achange may be a reection of the overall organizational style (McCleary,Nienstedt, & Erven, 1982; Wilson, 1968). This suggests that a more systematicaccount of individual and organizational discretion is required. More specic-ally, a longitudinal multidepartment study of detectives and their organization isnecessary to root the practice in SAPD within the range of detective routines andorganizational congurations available to American municipal police and tomore carefully track and benchmark changes in internal processing. To befair, selecting appropriate measures continues to be a challenge for policeresearchers. As Maguire and Uchida (2000) succinctly state, if measurementis a problem within individual police agencies, then developing measures thatallow us to compare multiple police agencies is like herding cats (p. 491).

    Despite the current limitations, the qualitative and quantitative inquiry intowhether and how organizational change aects robbery detectives, robberyclearances and crime, is likely to stimulate further research among scholarsand practitioners. The contextualizing comments of the robbery detectives, forexample, suggest that future research in this area might further contribute to ourunderstanding of current information usage. This is particularly important giventhe technological advances in policing. Given the ubiquity of information from

    270 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • video, social networking sites, pawn shops, cell records, geolocation associatedwith networked devices, and other emerging technology, any forecast of howbest to approach investigations in terms of triage, time investment, and resourcesseems premature. Perhaps the wisest approach would be to invest in a survey ofpotential information sources (much like has been done with DNA) and estab-lish a best practices portfolio for information processing investigations at thenational level. Additionally, such eorts would oer valuable, updated informa-tion regarding the issues discussed in classic works by Eck (1983) andGreenwood et al. (1975).

    This research represents a tentative rst step in identifying a substantial gap incontemporary detective work research in describing what contemporary investi-gators do. The results are suggestive, but certainly require further multisiteexperimental tests with varying congurations of detective organization andmanipulation of a variety of conditions in the work environment. Given thata substantial percentage of police resources are consumed by investigations, abroader programmatic line of inquiry describing contemporary investigationsand testing propositions regarding investigative eectiveness is warranted.

    Acknowledgments

    The researchers acknowledge the San Antonio Police Departments cooperation in col-lecting data used in this research. The points of view in this document are the authors and

    are not intended to represent those of the Bureau of Justice Assistance or the San AntonioPolice Department.

    Declaration of Conflicting Interests

    The authors declared no potential conicts of interest with respect to the research,authorship, and/or publication of this article.

    Funding

    The authors disclosed receipt of the following nancial support for the research, author-

    ship, and/or publication of this article: This research was funded by Bureau of JusticeAssistance grant #2009-SC-B9-0101.

    Notes

    1. Others have alluded to metaphors in describing detective activities and their work. For

    example, Harper (1991) characterized detective tasks/activities as transforming variousfeatures of reported crime into a Weber-like bureaucratic phenomenon, whereas Irving(1980) equated detective work to a production-line quality involving processes, out-

    puts, production of records, and successful prosecution.2. Data were drawn from the National Incident Based Reporting System (NIBRS), the

    2000 Law Enforcement Management and Administrative Statistics (LEMAS) data-base, and the 2000 Census. Measures of investigators and community policing were

    calculated as the ratio of officers to serious violent crimes in the jurisdiction.

    McCluskey et al. 271

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • 3. This sketch is derived from interviews with detectives with experience across four ofthe six substations and one supervisor.

    References

    Ackroyd, S., Harper, R., Hughes, J. A., Shapiro, D., & Soothill, K. (1992). New tech-nology and practical police work. Buckingham, England: Open University Press.

    Angell, J. (1971). Toward a alternative to the classical police organizational arrange-ments: A democratic model. Criminology, 9, 185206.

    Bayley, D. A. (1998). What works in policing? New York, NY: Oxford University Press.

    Box, G., & Jenkins, G. (1976). Time series analysis: Forecasting and control.San Francisco, CA: Holden Day.

    Braga, A., Flynn, E. A., Kelling, G. L., & Cole, C. M. (2011). Moving the work of the

    criminal investigator towards crime control. Washington, DC: United StatesDepartment of Justice.

    Brandl, S. (1993). The impact of case characteristics on detectives decision making.Justice Quarterly, 10, 395415.

    Brandl, S., & Frank, J. (1994). The relationship between evidence, detective effort, andthe disposition of burglary and robbery investigations. American Journal of Police, 13,149168.

    Brown, M. K. (1988). Working the street: Police discretion and the dilemmas of reform.New York, NY: Russell Sage.

    Chaiken, J., & Chaiken, M. (1982). Varieties of criminal behavior. Santa Monica, CA:

    Rand Corporation.Chan, J. (2001). The technological game: How information technology is transforming

    police practice. Criminology and Criminal Justice, 1, 139159.

    Clare, J. (2010). Examination of systematic variations in burglars domain-specific per-ceptual and procedural skills. Psychology, Crime & Law, 17, 199124.

    Danziger, J. N., & Kraemer, K. L. (1985). Computerized data-based decision systems andproductivity among professional workers: The case of detectives. Public

    Administration Review, 45, 196209.Dean, G., Fahsing, I. A., & Gottschalk, P. (2007). Creativity as a determinant of thinking

    style in police investigations. International Journal of Police Science & Management, 9,

    112121.Durlauf, S., & Nagin, D. (2011). Imprisonment and crime: Can both be reduced?

    Criminology & Public Policy, 10, 1354.

    Eck, J. E. (1983). Solving crimes: The investigation of burglary and robbery. Washington,DC: Police Executive Research Forum.

    Eck, J. E. (1992). What works in policing? Operations and administration examined.In G. Cordner & D. C. Hale (Eds.), Criminal investigations (pp. 1934). Cincinnati,

    OH: Anderson.Etzioni, A. (1960). Two approaches to organizational analysis: A critique and a sugges-

    tion. Administrative Science Quarterly, 5, 257278.

    Etzioni, A. (1961). A comparative analysis of complex organizations. New York, NY: FreePress.

    Feeley, M. M (1973). Two models of the criminal justice system. Law & Society Review, 7,

    407425.

    272 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • Gaines, L. K., Lewis, B., & Swanagin, R. (1983). Case screening in criminal investiga-tions: A case study of robbery. Police Studies, 6, 2229.

    Goldstein, H. (1977). Policing a free society. Cambridge, MA: Ballinger.Greenwood, P. W., Chaiken, J., Petersilia, J., & Prusoff, L. (1975). The criminal investi-

    gation process: Observations and analysis (Vol. 3). Santa Monica, CA: RandCorporation.

    Greenwood, P. W., & Petersilia, J. (1975). The criminal investigation process: Summary

    and policy implications (Vol. 1). Santa Monica, CA: Rand Corporation.Harper, R. R (1991). The computer game: Detectives, suspects, and technology. British

    Journal of Criminology, 31, 292307.

    Hickman, M. J., & Reaves, B. A. (2006). Sheriffs offices, 2003. Washington, DC: Bureauof Justice Statistics.

    Horvath, F., Meesig, R., & Lee, Y. (2001). National survey of police policies and practices

    regarding the criminal investigations process: Twenty-five years after Rand. EastLansing: Michigan State University.

    Irving, B. (1980). Police interrogation: A case study of correct practice. Royal Commissionon Criminal Procedure (Research Study No. 1, pp. 75153). London, England: Her

    Majestys Stationery Office (HMSO).Kenney, D., White, M., & Ruffinengo, M. (2010). Expanding the role of patrol in crim-

    inal investigations: Houstons investigative first responder project. Police Quarterly,

    13(2), 136160.Klockars, C. B. (1985). Idea of police. Thousand Oaks, CA: Sage.Klofas, J., McGarrell, E., & Hipple, N. K. (2010). The new criminal justice. New York,

    NY: Routledge.Knobler, P., & Bratton, W. (2009). The turnaround: How Americas top cop reversed the

    crime epidemic. New York, NY: Random House.Kubrin, C. E., Messner, S. F., Deane, G., McGeever, K., & Stucky, T. (2010). Proactive

    policing and robbery rates across U.S. cities. Criminology, 48, 5797.MacDonald, J. M. (2002). The effectiveness of community policing in reducing urban

    violence. Crime & Delinquency, 48, 592618.

    Maguire, E. R., & Uchida, C. D. (2000). Measurement and explanation in the compara-tive study of American police organizations. In D. Duffee, D. McDowall, B. Ostrom,R. Crutchfield, S. Mastrofski & L. Mazerolle (Eds.), Criminal justice 2000:

    Measurement and analysis of crime and justice [NCJ 182411] (Vol. 4, pp. 491557).Washington, DC: Office of Justice Programs, U.S. Department of Justice.

    Maguire, M. (2003). Criminal investigation and crime control. In T. Newburn (Ed.),

    Handbook of policing (pp. 430464). Devon, England: Willan.Manning, P. K. (1992). Information technology and the police. In M. Tonry & N. Morris

    (Eds.), Modern policing (pp. 349398). Chicago, IL: University of Chicago Press.McCleary, R., & Hay, R. (1980). Applied time series analysis. Beverly Hills, CA: Sage.

    McCleary, R., Nienstedt, B., & Erven, J. (1982). Uniform crime reports as organizationaloutcomes. Social Problems, 29(4), 361372.

    Northrop, A., Kraemer, K. L., & King, J. L. (1995). Police use of computers. Journal of

    Criminal Justice, 23, 259275.Packer, H. (1968). The limits of the criminal sanction. Stanford, CA: Stanford University

    Press.

    McCluskey et al. 273

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • Reiss, A. J., & Black, D. J. (1967). Interrogation and the criminal process. Annals of theAmerican Academy of Political and Social Science, 374, 4757.

    Roberts, A. (2008). The influences of incident and contextual characteristics on crimeclearance of nonlethal violence: A multilevel event history analysis. Journal of Criminal

    Justice, 36, 6171.Sanders, W. B. (1977). Detective work: A study of criminal investigations. New York, NY:

    Free Press.

    Scott, W. R. (1992). Organizations: Rational, natural, and open systems (3rd ed.).Englewood Cliffs, NJ: Prentice Hall.

    Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-

    experimental designs for generalized causal inference. Boston, MA: Houghton Mifflin.Sherman, L. W. (1990). Police crackdowns: Initial and residual deterrence. In M. Tonry &

    N. Morris (Eds.), Crime and justice: A review of research (pp. 148). Chicago, IL:

    University of Chicago Press.Sherman, L. W. (2011). Al Capone, the sword of Damocles, and the policecorrections

    budget ratio. Criminology & Public Policy, 10, 195206.Simon, H. (1947). Administration behavior. New York, NY: MacMillan Press.

    Skogan, W., & Frydl, K. (2004). Fairness and effectiveness in policing: The evidence.Washington, DC: The National Academies Press.

    Skogan, W. G., & Antunes, G. E. (1979). Information, apprehension, and

    deterrence: Exploring the limits of police productivity. Journal of Criminal Justice,7, 217241.

    Skolnick, J. (1967). Social control in the adversary system. Journal of Conflict Resolution,

    11, 5267.Smith, B. (1960). Police systems in the United States. New York, NY: Harper Press.Tarling, R. (1988). Police work and manpower allocation. London, England: Home Office

    Research and Planning Unit.

    Visher, C. A. (1986). The Rand inmate survey: A reanalysis. In A. Blumstein,J. Cohen, J. A. Roth & C. A. Visher (Eds.), Criminal careers and career criminals(pp. 161211). Washington, DC: National Academies Press.

    Weber, M. (1946). Bureaucracy. In H. Gerth & C. Mills (Eds.), From Max Weber: Essaysin sociology (pp. 196244). New York, NY: Oxford University Press.

    Weisburd, D., Mastrofski, S. D., McNalley, A. M., Greenspan, R., & Willis, J. (2003).

    Compstat and organizational change: Findings from a national survey. Criminology &Public Policy, 2, 421456.

    Wexler, C. (2010). Is the economic downturn fundamentally changing how we police?

    Critical Issues in Policing. Washington, DC: Police Executive Research Forum.Wilson, J. Q. (1968). Varieties of police behavior: The management of law and order in eight

    communities. Cambridge, MA: Harvard University Press.Wilson, J. Q., & Boland, B. (1978). The effect of the police on crime. Law and Society

    Review, 12, 367390.Womack, C. L. (2007). Criminal investigations: The impact of patrol officers on solving

    crimes (Masters thesis). University of North Texas, Denton.

    Woodhams, J., & Toye, K. (2007). An empirical test of the assumptions of case linkageand offender profiling with serial commercial robberies. Psychology, Public Policy, andLaw, 13, 5985.

    274 Police Quarterly 17(3)

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from

  • Wright, R. T., & Decker, S. H. (1997). Armed robbers in action: Stickups and streetculture. Boston, MA: Northeastern University Press.

    Wycoff, M., & Cosgrove, C. (2001). Investigations in the community policing context.Washington, DC: Police Executive Research Foundation.

    Zammuto, R., Griffith, T., Majchrzak, A., Dougherty, D., & Faraj, S. (2007).Information technology and the changing fabric of organization. OrganizationScience, 18, 749762.

    Author Biographies

    John D. McCluskey is a professor in the department of criminal justice at theRochester Institute of Technology. His current research areas include proceduraljustice, policing, and evaluation research focusing on process and outcomeassessment within local criminal justice systems.

    Jeffrey M. Cancino is an associate professor in the School of Criminal Justice atTexas State University-San Marcos. His current research areas include homicide,immigration, driving while intoxicated, policing, and more recently, serial com-mercial robbery.

    Marie Skubak Tillyer is an associate professor in the Department of CriminalJustice at the University of Texas at San Antonio. Her research interests includeviolence, victimization, and crime prevention. Her recent research has appearedin Crime and Delinquency, Criminal Justice and Behavior, Justice Quarterly, andJournal of Research in Crime and Delinquency.

    Rob Tillyer, PhD, is an associate professor of criminal justice at the University ofTexas at San Antonio. His research interests include decision-making within thecriminal justice system, crime prevention, and victimology. His recent journalarticles have appeared in Justice Quarterly, Criminal Justice and Behavior, andCrime and Delinquency.

    McCluskey et al. 275

    at University of Hong Kong Libraries on October 18, 2014pqx.sagepub.comDownloaded from