Improving Performance in a Swedish Police Traffic Unit Results of an Intervention

8/12/2019 Improving Performance in a Swedish Police Traffic Unit Results of an Intervention

1/13

Improving performance in a Swedish police trafc unit: Results of an intervention

Robert D. Pritchard a, Satoris S. Culbertson b,, Kenneth Malm c, Anders Agrell d

a Department of Psychology, University of Central Florida, Orlando, FL 32816-1390, United Statesb Department of Psychology, Kansas State University, 421 Bluemont Hall, 1100 Mid-Campus Drive, Manhattan, KS 66506-5302, United Statesc ProMES International Sweden AB, 70357 rebro, Swedend Department of Social Science Psychology, University of rebro Sweden, 701 82 rebro, Sweden

a b s t r a c t

This article describes the results of a feedback system designed to improve performance for a Swedish trafcpoliceunit andexamined whether such a feedback system was benecial or detrimental tothe attitudes of theofcers. As in many Western countries, government organizations are being required to demonstrate theireffectiveness with quantitative performance measures.An approach called the ProductivityMeasurement andEnhancement System (ProMES) was used with three groups of Swedish Trafc Police to do this. ProMES is amethod for identifying unit's objectives, developing measures for these objectives, and using this informationas feedback.ProMESwas developed withthese police unitsand feedbackfrom thesystemwas usedover a four-year period. Resultsindicatethat thereweresubstantial increasesin performance. Therewerealso decreases inaccidents, injuries, and fatalities compared both to baseline and to comparison groups in Sweden. Theseimprovements were made with fewer and fewer police ofcers each year.

2008 Elsevier Ltd. All rights reserved.

Introduction

As in many Western countries, government organizations are beingrequired to become more accountable by using performance measuresto evaluate the contributions made by the agency. In addition, agenciesare being increasingly asked to monitor and improve their performanceas the need to control budgets and pressures to downsize reminddecision-makers of the need to do more with less. While this need to bemore efcient while maintaining effectiveness is important, however,there are other factors to consider. For example, it is also importantthatevery employee understands the overall organizational mission as wellas the plan to accomplish that mission using departmental goals andobjectives (Harrison, 1996). It is not always clear to departments,however, how to accomplish this without destroying morale.

The primary purpose of this study was to evaluate a specicintervention that accomplished these objectives with a sample ofSwedishtrafc police units.Thisstudy soughtto examine theresults of ameasurement and feedback system for improving performance whilealso increasing alignment of efforts with organizational objectives. Inthis article,performance is denedas how well theoutputsproduced bythe unit meet organizational objectives. The effects of this interventionon attitudes of the ofcers were also examined. A secondary purpose ofthis study wasto exploretwo problematicmeasurementissuescommonto many types of work, but especially to police work: control over

outcomes and detection of negative events. This article rst describes

the theoretical background underlying the intervention, followed by asummary of how the intervention is done and a discussion of specicmeasurement issues. The article concludes with a presentation of theresults of the intervention and a discussion of the ndings.

Theoretical background

The intervention used in this study was the Productivity Measure-ment and Enhancement System, or ProMES (Pritchard, 1990, 1995;Pritchard, Harrell, DiazGranados, & Guzman, 2008; Pritchard, Holling,Lammers, & Clark, 2002). ProMES is a results-oriented measurementand feedback system specically designed to improve performanceover time, while at the same time improving the quality of work life.The theoretical background of ProMES comes primarily from themotivational aspects of the Naylor, Pritchard, and Ilgen (1980)(NPI)theory and a more recent motivation theory (Pritchard & Ashwood,2008) based on NPI theory. These theories are expectancy theories,proposing that individuals are motivated by the anticipation of howtheir effort will lead to the satisfaction of their needs (e.g.,Campbell &Pritchard, 1976; Heckhausen, 1991; Kanfer, 1990, 1992; Latham &Pinder, 2005; Mitchell & Daniels, 2003; Vroom, 1964).

ThePritchard and Ashwood (2008)theory posits that individualshave a certain amount of energy at a given timetheir energy poolthat is used to satisfy their needs for such things as food, water,achievement, safety, and power. This energy pool varies across peopleand across time for any individual. The concept is similar to thatof attention resources (Kanfer & Ackerman, 1989; Kanfer, Ackerman,

Journal of Criminal Justice 37 (2009) 8597

Corresponding author. Tel.: +1 785 532 0620; fax: +1 785 532 5401.E-mail address:[email protected](S.S. Culbertson).

0047-2352/$ see front matter 2008 Elsevier Ltd. All rights reserved.

doi:10.1016/j.jcrimjus.2008.12.008

Contents lists available at ScienceDirect

Journal of Criminal Justice
mailto:[email protected]://dx.doi.org/10.1016/j.jcrimjus.2008.12.008http://www.sciencedirect.com/science/journal/00472352http://www.sciencedirect.com/science/journal/00472352http://dx.doi.org/10.1016/j.jcrimjus.2008.12.008mailto:[email protected]


2/13

Murtha,Dugdale, & Nelson, 1994) in that the energy pool concernstheissue of the limited resources individuals have to devote to tasks.Within this framework, motivation is described as the process thatdetermines how this energy is used to satisfy needs.

According to the Pritchard and Ashwood (2008) theory, themotivation process can be broken down into several key componentsactions, results, evaluations, outcomes, and need satisfaction.Energy is allocated across possible actions (e.g., a police ofcer

patrolling a neighborhood, issuing citations, or writing reports),which generally producesresults(e.g., typing, an action, generates areport, a result). Thus, a result is the individual's output. Whenresults are observed and an evaluator places the measured result ona good-to-bad continuum, evaluations are produced. Multipleevaluators evaluate the output (e.g., the ofcer's report may beevaluated by the ofcer, his or her supervisor, and/or the districtattorney). After these evaluations are made, outcomes occur. Theseare intrinsic outcomes such as a feeling of accomplishment, orextrinsic outcomes such as forms of recognition, promotion, or payraises. Outcomes get their motivating power because of their ties toneed satisfaction. The more an individual's needs are satised, thegreater the positive affect that is experienced.

As with other expectancy theories, the linkages between thecomponents are critical. Within the Pritchard and Ashwood (2008)theory, these linkages are called connections. The rst linkage is theactions-to-results connection, which describesthe individual's perceivedrelationship between the amount of effort devoted to an action and theamount of the result that is expected to be produced. This perceivedrelationshipcan range fromvery strong to nonexistent. The next linkageis the results-to-evaluations connection. This connection reects theindividual's perceived relationship between the amount of a result thatis produced and the level of the evaluation that is expected to occur.There would be such a connection for each different result and for eachindividual who evaluates the result(s) such as the ofcer, colleagues,supervisor, prosecutor, and so forth. The strength of these connectionsvaries. The timeliness of reports (a result), for example, may be morestrongly related to the supervisor's evaluation of the ofcer than theamount of community service. Theevaluations-to-outcomesconnection

is the perceived relationship between the level of the evaluationand thelevel of outcomeexpected. The outcomes-to-need satisfaction connectiondenes the perceivedrelationship between how much of an outcome isreceived and the degree of anticipated need satisfaction that will result.

According toPritchard and Ashwood (2008), the result of thesemotivation components is the intent to behave. This intent leads toactual behavior, or the application of energy to actions, which in turnleads to actual results, evaluations, outcomes, and need satisfaction.These actual events have a feedback relationship with the variousmotivational components. For example, actual outcomes receivedinuence subsequent evaluations-to-outcomes connections.

In addition to Naylor et al. (1980)theory and the Pritchard andAshwood (2008)motivational theory, the development of ProMES hasbeen inuenced by several other bodies of literature. These include the

literature on feedback, goal setting, participation, roles and roleambiguity and conict, and team effectiveness. How these literaturesinuenced thedesignof ProMES isdescribedbelow, after thedescriptionof ProMES.

The ProductivityMeasurement and Enhancement System (ProMES)

The next section describes the specic intervention used. This isfollowed by a discussion of how the intervention operationalizes thetheory and other literatures.

Development of the system

The ProMES intervention is typically done in a series of steps,

described in greatest detail in Pritchard (1990). Tosummarize, a design

team is formed composed of people from the target unit, one or twosupervisors, and a facilitator familiar with ProMES. This design teammeets to identify the objectives of the unit and correspondingquantitative measures (indicators) that assess how well the unit ismeeting the objectives. The objectives and indicators are thenapproved by higher management in a formal meeting between thedesign team and higher management where management reviews,and if necessary, works with the design team to revise the objectives

and indicators.

Objectives and indicators

The objectives and indicators might look like the following.(In most actual cases, four to six objectives and eight to fteenindicators are developed, but to keep the example manageable, only asubset was used.)

Objective 1: respond to emergency calls

Indicator 1: average number of minutes to respond to emergency

calls

Objective 2: investigate crime

Indicator 2: percentage of violent crimes leading to arrest

Indicator 3: percentage of violent crimes handled within thirtydays

Objective 3: aid in the prosecution of crime

Indicator 4: percentage of arrests transferred to the prosecutor

Objective 4: facilitate crime prevention programs

Indicator 5: number of ongoing prevention programs

Indicator 6: percentage of ofcer time for educational crime

prevention programs

Contingencies

Once the objectives and indicators are approved, the design teamdevelops what are known as contingencies. Contingencies are a typeof graphic utility function relating variation in the amount of the

indicator to variation in unit effectiveness. In other words, it is afunction that denes how much of an indicator is how good for theorganization.Fig. 1shows examples of contingencies for four of theindicators above. The upper left quadrant ofFig. 1is the contingencyfor the rst indicator, average number of minutes to respond toemergency calls. Varying amounts of this indicator are shown on thehorizontal axis, ranging from a slower response time of fteenminutes to a quicker response time of three minutes. The vertical axisis the effectiveness score. Effectiveness is dened as the amount ofcontribution being made to the organization. It ranges from -100,through 0 to +100. The zero point is dened as the amount of theindicator just meeting minimum expectations. Indicator amountsabove this expected level get a positive effectiveness score. The higherthe unit is above this expected level, the higher the effectiveness

score. Indicator amounts below the expected level receive a negativeeffectiveness score.

The contingency relates indicator amounts to the effectivenessscores. For example, the contingency for the rst indicator (averagenumber of minutes to respond to emergency calls) shows that theminimum expected level is nine minutes. It is not expected thataverage response time can be lower than three minutes because oftime to transmit the call and the distance to be traveled. Thecontingency indicates that responding nine minutes and more getsprogressively worse. Going from an average of nine minutes to veminutes produces a substantial increase in effectiveness, but respond-ing in less than ve minutes produces a much smaller improvement ineffectiveness. This is because, although responding faster than veminutes is of value, it could also produce some negative consequences

such as driving recklessly to the site. Unit members create a separate

86 R.D. Pritchard et al. / Journal of Criminal Justice 37 (2009) 8597


3/13

contingency for each indicator, so in the current example with its fourindicators, there are four contingencies.

A formal step-by-step process is followed to develop thecontingencies. This procedure is described inPritchard (1990)and inPritchard et al. (2008). It essentially consists of group discussion toconsensus where contingency development is broken into discrete

steps executed by the design team. Each of the different parts of thecontingencies are agreed upon, and then put together into a whole.Once the design team has come to agreement on the contingencies,they are presented to higher management for review and approval.This is similar to the step done for objectives and indicators.

Importance of contingencies

Three things are particularly noteworthy about the contingencies.First, they essentially scale the level of output (the indicator level) tohow good that is (the effectiveness score). In doing this, they formallydene what is considered good, adequate, and poor performance oneach indicator. This allows the feedback system to provide bothdescriptive feedback from the indicator level and evaluative feedback

from the effectiveness score. With the contingencies agreed upon, the

individuals in the unit and their management know in advance howgood or bad each level of output is considered. If a unit gets aneffectiveness score above zero, the unit has exceeded minimumexpectations. The higher the score, the more they have exceededexpectations. Negative effectiveness scores mean the unit is perform-ing below expectations.

A second feature of the contingencies is that they capturedifferential importance. Not every indicator is equally important,and the overall slope or range of the effectiveness scores captures thisdifferential importance. For example, Fig.1shows that percent violentcrimes handled within thirty days is the most important indicatorbecause it has the steepest slope; it ranges from an effectiveness scoreof - 100 to +100. Number of ongoing crime prevention programs is theleast important indicator with a range from - 30 to + 30.

The third noteworthy feature of the contingencies is that theycapture nonlinearity. The relationship between how much anorganizational unit does on an indicator and the amount ofcontribution (effectiveness) that level of the indicator makes to theoverall functioning of the organization is frequently not linear(Campbell, 1977; Campbell & Campbell, 1988; Kahn, 1977; Pritchard,

Jones, Roth, Stuebing, & Ekeberg, 1989; Pritchard, Youngcourt, Philo,

Fig. 1.Example ProMES contingencies.

87R.D. Pritchard et al. / Journal of Criminal Justice 37 (2009) 8597


4/13

McMonagle, & David, 2007). It is common, for example, that once theunit's level of quality reaches a point that satises the customer,further improvements in quality are not especially valuable. That is, apoint of diminishing returns is reached. The contingencies in ProMEScapture this nonlinearity. For example, the contingency inFig. 1forpercentage of violent crimes handled within thirty days shows a pointof diminishing returns after 80 percent. Such a contingency mightreectthebeliefthatbecausesomecrimesarenotevergoingtobefully

resolved, devoting the resources to go above 80 percent is not aneffective use of resources. The contingency for percent of ofcer timefor educational crime prevention programs (in the lower right-handquadrant)is a special type of nonlinearityshowing that if thepercent ofofcer time goes above 5 percent, thevalueto theorganizationactuallydecreases. It is also important to note thatthese nonlinearities are verycommon, with the vast majority of the contingencies developed inProMES having some degree of nonlinearity.

Feedback using the system

Upon approval of the contingencies by upper management, thefeedback system is nished and ready for implementation. Unitpersonnel collect data on the indicators, and along with thecontingency information, feedback is provided to each member ofthe unit after each performance period during regular feedbackmeetings. An example of the basic information in the feedback reportis shown inTable 1.The indicators are listed along with the score onthat indicator for the unit for that period. For example, the unitresponded to emergencies in an average of 5.0 minutes for this period.Based on the contingency, this correspondsto an effectiveness score of+20, above minimum expectations. The contingencies rescale all theindicators to a common metric of effectiveness. Thus, they can beadded together to produce an overall effectiveness score for thatperiod. In the example, the unit's overall effectiveness score for thisperiod is +43, well above minimum expectations. Other aspects of thefeedback included plots of the overall effectiveness score over timeand changes in indicator scores from the past month to the currentmonth.

The feedback meetings are held with all unit personnel and theirsupervisor to review the feedback report, identify ways to makeimprovements, and evaluate previous improvement attempts. Wherebythe unit personnel essentially design the system, and managementapproves it,an understanding and alignment of organizationalgoalsandobjectives is more likely because any misunderstandings or misalign-ments would be discussed andresolved.This processof regularfeedbackreports and meetings goes on over time in a continuous improvementmodel. The feedback system is reviewed periodically to determinewhether changes are necessary.

How ProMES operationalizes the theory

ProMES was designed to operationalize the Naylor et al. (1980) and

Pritchard and Ashwood (2008)motivation theories and capitalize ontheir implications. Results in the theory are the ProMES indicators and

results-to-evaluations connections are the contingencies. The actions-to-results connections can be thought of as deningwork strategies inthat they identify how effort should be allocated across actions. Thefeedback reports and feedback meetings focus on developing betterwork strategies, i.e., a more optimal set of action-to-results connec-tions. The feedback over time allows unit personnel to evaluate howwell the new strategies are working and to rene them as needed,further rening the actions-to-results connections. The process of

developing these connections and having them approved by manage-ment ensures role clarity and reduces role conict because doing thecontingencies means thatdifferent evaluators agree on what is valued.As employees experience improvements in performance, outcomesare added in the form of recognition from supervisors and feelings ofpersonal accomplishment.

ProMES was also designed to operationalize factors from otherliteratures that show performance and attitude improvements. Theliterature on feedback (e.g., Bobko & Colella, 1994; Ilgen, Fisher, &Taylor, 1979; London, 2003; Murphy & Cleveland, 1995; Smither,London, & Reilly, 2005; Taylor, Fisher, & Ilgen, 1984; Wall & Lischeron,1977; West & Anderson, 1996) hassuggested a number of featuresthatan ideal feedback system should include. One such feature is includingboth descriptions and evaluations of performance, which ProMESdoes through indicator (description) and effectiveness (evaluation)scores. Anotherimportant feature is making the system transparentsopeople know what their evaluations will be.ProMES accomplishes thisthrough formal agreement of the performance measurement systemby all individuals involved in making ProMES the evaluation system.Validity of the measurement system and perceived validity of thesystem, additional features described in the feedback literature, aremaximized through a careful review of the indicators and contingen-cies by the design team, unit members not on the design team, andupper management. Finally, accuracy of the feedback is ensuredthrough efforts to ensure validity, maximize participation, make thesystem transparent, and give regular feedback.

ProMES also includes aspects ofgoal setting (Latham & Pinder,2005; Locke & Latham, 2002). While goal setting clearly includesformal, relatively public, agreed-upon levels of output to strive for

(formal goal setting), it also includes less formal processes such asprivate and public intentions to act (Frese & Zapf, 1994; Locke &Latham, 2002). ProMES provides feedback for learning goals, andfeedback meetings focus on behaviors necessary to attain thoselearning goals that will help performance (Brown & Latham, 2002;Latham, Mitchell, & Dossett, 1978).

ProMES also includes features from the participation literature.There is considerable evidence that participation on important issuescan have positive effects on employee performance and attitudes(Cawley, Keeping, & Levy, 1998; Dipboye & de Pontbriand, 1981; Locke& Schweiger, 1979). Participation enhances perceptions of proceduraljustice and voice (Cawley et al., 1998; Lind, Kanfer, & Earley, 1990).ProMES design teams participate heavily in formulating the measure-ment system and are encouraged to discuss the development process

with unit members outside of the design team. In addition, the entireunit participates in feedback meetings.

The literature on roles is also addressed within ProMES. Rolesin Naylor et al. (1980) and Pritchard and Ashwood (2008) are dened asthe set of results-to-evaluations connections.These connectionsidentifyexpectedoutputs, indicate their relative value of different measures, anddene how level of output is related to value and to evaluations. Roleconict and ambiguity inuence performance and attitude variables(Fisher & Gitelson,1983; Jackson & Schuler, 1985; Tubr & Collins,2000).Role ambiguity is reduced by identifying the results-o-evaluationsconnections; role conict is reduced by gaining agreement on connec-tions by unit personnel, supervisor(s), and upper management.

Finally, theteams literature is also relevant. In a study examiningthousands of teams, West (2007) found teams performed better

and were more satised when there were clear objectives, members

Table 1

Basic feedback information

Indicators Indicator value Effectiveness score

Average minutes to respond to emergencies 5 .0 +20Percent violent crimes handled within

thirty days47% -30

Number of ongoing crime preventionprograms

15 +38

Percent ofcer time spent onprevention programs

6% +15

Overall effectiveness +43



5/13

worked together to achieve objectives, and when they met regularlyto review team effectiveness and how it could be improved. ProMESincludes all three of these characteristics. Additionally, Salas, Kosar-zycki, Tannenbaum, and Carnegie (2004) and Salas, Rosen, Burke,Goodwin, and Fiore (2006)identied numerous factors affect teameffectiveness including holding shared mental models, having clearroles and responsibilities, engaging in a prebrief-performance-debriefcycle, cooperating and coordinating, and using multiple criteria.

ProMES creates shared mental models of the work through thedevelopment of the system. Roles and responsibilities are clariedduring the development of the system. Additionally, the ongoingfeedback meetings are a type of prebrief-performance-debrief cycle,and cooperation and coordination are encouraged through the feed-back meetings. Finally, multiple criteria of performance are included inthe multiple indicators.

ProMES and other interventions

There are other interventions that have similarities to ProMES suchas total quality management (e.g., Mahoney & Thor, 1994) and thebalanced scorecard (Kaplan & Norton, 1993, 1996). ProMES has anumber of features, however, that make it different from theseapproaches. First, ProMES has a clearunderlying theorythat led to itsdevelopment, Naylor et al. (1980) and the Pritchard and Ashwood(2008)theories. It is also based on other literatures such asfeedback,

goal setting, roles, and participation. ProMESis alsoa bottom-up strategyrather than the top-down approach of the balanced scorecard andrelies heavily on theparticipationof the people doing the work. Thishigh level of participation means helps improve the acceptanceof theresulting system. The combination of people doing the work develop-ing the system and their ideas for the measurement system reviewedand approved by higher management helps improve alignmentwithbroader organizational objectives. The most unique structural featureis the use of contingencies in ProMES. These capture differentialimportanceof the measures, dene what level of performance is ex-

pected, andcapture nonlinearities. These nonlinearities are especiallyimportant because they can be used to quantify and thus clearly

communicate the organizational priorities for improvement strategies.Forexample, consider the twotop contingencies in Fig.1. Suppose for agiven month, response time averaged nine minutes and percentviolent crimes handled was 80 percent. Improving from nine to sevenminutes response time adds aboutfteen effectiveness points whileimproving violent crimes handled from 80 percent to 90 percent addsonly about seven effectiveness points. So even though violent crimeshandled is overall the more important measure, it is better to focusimprovement efforts on response time. The determination of effec-tiveness improvement for different measures allows for improvementstrategies to be made where they can have the most positive effect onoverall effectiveness. Finally, as described in the next section, there isclear evidence that ProMES has a positive effect on performance basedon applications by many scholars, in different types of jobs, different

types of organizations, and in different countries. There is surprisinglylittle scientically acceptable evidence that other interventionsactually improve performance.

Past experience with ProMES

Although ProMES has been shown to be effective in many settings(Pritchard et al., 2008), prior to this study it had not been usedspecicallywith law enforcement personnel. Nevertheless,Pritchardet al. (2008)reported results from eighty-three ProMES interventionswhere the overall effectiveness score duringbaseline was compared tothat during feedback. Results showed large improvements underfeedback. The mean effect size from baseline to feedback was 1.16 and1.44 when weighted by number of time periods in each study. Put

another way, performance improved under feedback by 1.16 standard

deviations, 1.41 when weighted. These effect sizes are very large, andassuming performance over time forms roughly a normal distribution,mean that overall effectiveness during feedback equals what was theeighty-eighth percentile under baseline. The weighted effect size of1.44 is the ninety-third percentile under baseline. Performance datafor control groups who did not get ProMES feedback showed nochange in performance during the time ProMES was done. Thesendings led to the prediction that ProMES will improve performance

in criminal justice settings.

Hypothesis 1. Performance will be higher during ProMES feedbackthan during the baseline prior to feedback.

Specic measurement issues

Two specic measurement issues must be addressed in anymeasurement of police performance, whether ProMES or anotherapproach is used. A special focus of this project was to attempt to dealwith these issues in a novel way.

Control over outcomes

The ultimate goals of police work are often not under the control ofthe ofcers. For example, typical goals include preventing crime,enhancing public safety through education and enforcement, andsafeguardingproperty and individualrights. There arenumerous factors,however, that police ofcers cannot control that contribute to crimerates, such as the socioeconomic status of individuals within their beats(Bukenya, 2005; Rutter, Giller, & Hagell, 1998), pollution levels (Rotton,1983; Rotton & Frey, 1985), population density of their city (Christens &Speer, 2005), the local job market (Wang, 2005), or even the weather(C. A. Anderson,1989; Hipp, Bauer, Curran, & Bollen, 2004). Furthermore,asEisenberger, Fasolo, and Davis-LaMastro (1990)noted, patrol ofcerstypically lack discretionary control over many aspects of theirperformance such as the number of occasions they are called to thescene of accidents or the time spent in court waiting to testify. Thiscreates a dilemma when assessing police performance. Ofcers want to

be evaluated on factors under their control, but there may be few thattruly fall under their control, andthe factors under their control may notdirectly inuence the ultimate goals of public safety.

This importance of being evaluated on measures that arecontrollable is supported by research and theory. A variety of sourceshave suggested that the less control one has over outputs, the lowerhis or her motivation is to improve those outputs (Frese & Zapf, 1994;Muckler, 1982; Spector, 1986; Wall, Corbett, Martin, Clegg, & Jackson,1990). For example, thePritchard and Ashwood (2008)theory wouldpredict that if the relationship between effort and performance is low(action-to-results connections), motivation will be low. Not havingcontrol over the measure means personnel cannot impact themeasure by their efforts. Furthermore, the importance of controlover outputs goes beyond motivation. Indeed, in a meta-analytic

review of the literature on control, Spector (1986)found that highlevels of perceived control are positively related to motivation, jobsatisfaction, commitment, involvement, and performance.

The dilemma,therefore, is whether to measure those things that arethe most important to the organization and let motivation, jobsatisfaction, commitment, work involvement, and performance suffer,or measure those things the people doing the work have control over,but which are not of primary value to the broader organization. Thisdilemma was present in this project. For the Swedish government andNational Board of the Police Force (Rikspolisstyrelsen), the mostimportant outcomes for trafc police were trafcaccidents, injuries,and fatalities.These objectives,however,are noteasilyunder the controlof the ofcers.

To deal with this issue, the design team specied objectives and

indicators under their control, but which they believed would



6/13

subsequently lead to the ultimate goals of decreased accidents,injuries, and fatalities. To address these ultimate goals, they agreedupon three overall objectives: enforce compliance with speed limits,decrease motorists' use of alcohol or other drugs, and decreasehazardous driving behavior. These objectives, however, were still notunder their control. Therefore they created indicators under theircontrol that would ultimately help them reach their objectives, andsubsequently the goals set by the Swedish government and National

Police Board. For example, although they had no control overmotorists' actual drug or alcohol usage (Objective 2), they had controlover how frequently they monitored specic roads at times whenpeople were most likely to be driving under the inuence of drugs oralcohol. This led to the creation of indicators under their control butthat they theorized would lead to attainment of the overall objectiveof decreasing accidents, injuries, and fatalities. The objectives andindicators developed by the trafc unit are shown inTable 2.

Thus, the Swedish trafc ofcers designed their system using whatthey perceived to be controllable measures. The plan, however, alsocalled for monitoring the less controllable measures: accidents,injuries, and fatalities. Additionally, some indicators were decidedlynot under the control of the ofcers. For example, as shown in Table 2,another indicator designed to measure Objective 2 (decrease motor-ists' use of alcohol or other drugs) that was not within their controlincluded the number of motorists failing alcohol or drug tests in thelast four months.The logic was the ProMES system could help themmakeimprovementson the things theycould control. If they were ableto improve on the controllable measures, and theultimateoutcomes ofaccidents, injuries, and fatalities decreased, this would tell them theirtheory was correct and would meet the goals of the government andthe National Police Board. In contrast, if they improved on thecontrollable measures, but the ultimate outcomes did not improve, itwould indicate their theorywas not correctand they would then needto change their measurementsystem. They wouldhave to change theirtheory of how to have positive impact on the ultimate outcomes. Thisresolution was accepted by ofcers and higher management.

Hypothesis 2. If overall effectiveness scores improve during ProMESfeedback, external measures of accidents, injuries, and fatalities willalso improve.

Detection of negative events

The other measurement challenge addressed in this project dealswith measures that detect negative events. Examples are crimes

detected by police, near misses detected by air traf

c controllers,inappropriate ndings detected by accountants, and errors made inclerical work. If the unit being evaluated is responsible for detectingthese negative events, this creates a problem because if the unitpersonnel do a better job of detecting these negative events, the rateof negative events would be expected to increase, at least in the shortterm. For example, if air trafc controllers do a better job of detectingand reporting near misses, more near misses will likely be reported.This gives the impression the controllers are doing a poorer jobbecause the rate of near misses goes up. A similar issue is present inpolicework. Examples are frequencyof drivers exceeding speed limits,number of motorists driving under the inuence of illegal substances,and frequency of robberies. Better detection of these events has theapparent effect of increasing their frequency.This apparent increase inthe frequency of negative events can lead to perceptions of poorerofcer performance. This creates the dilemma for the ofcers thatdoing a better job of detecting negative events may lead to theperception of poorer performance.

This was an issue in this project because several indicatorsmeasured negative behaviors: percent of vehicles driving within thespeed limit, number of motorists failing alcohol or drug tests in thelast four months, and number of reported hazardous drivingincidents in the last four months.The ofcers were concerned aboutincluding these measures because the better job they did detectingthem, the worse their performance would appear, at least in the shortterm. The solution they developed was to measure both the frequencyof these negative behaviors and directly measure the quantity andquality of their detection efforts. Examples of such detectionindicators, shown in Table 2, are percent of actual surveillance during

critical time periods for dened roads, number of alcohol and drugtests per month,and percent of total alcohol and drug testing timedone at high risk places or on high risk people. Their logic was that byincluding their detection efforts and the frequency of the negativeevents,the measurementsystemwould showthat increased detectionwas due to better monitoring. This solution was accepted by everyoneinvolved. There were no specic hypotheses about the negative eventsissue, but this issue is discussed in detail in the Discussion section.

Team climate perceptions

Although performance improvement is certainly important, itshould not be at the expense of employees. Furthermore, not everyindividual or departmentmay be open to change. If individuals are not

opento changeor ifthey donotbenet from the introduction of a newintervention, then large gains in performance would not be expected.Therefore, whether ofcer attitudes toward their unit and thedepartment and the climate for innovation were improved orworsened following the implementation of ProMES were examined.In order to do this, the team climate inventory (TCI) developed byWest (1990)was used, which measures four dimensions related togroup process and the climate for innovation (N. Anderson & West,1998; West, 1994). Innovation is dened as the intentional introduc-tion and application within a role, group, or organization of ideas,processes, products or procedures, new to the relevant unit ofadoption, designed to signicantly benet role performance, thegroup, the organization or the wider society (West & Farr, 1989, p.16).In this case, the innovations were the changed ways the unit did its

work because of ProMES.

Table 2Trafc police objectives and indicators

Objective 1: enforce compliance with speed limitsPercent time checking effectively for speeding out of total amount time devoted to

trafc controlPercent of speeding violations less than 15 km/h above limit, in relation to total

violationsPercent actual surveillance time to planned surveillance time on dened roadsPercent surveillance during critical time periods for dened roadsPercent of vehicles driving within the speed limit

Objective 2: decrease motorists'use of alcohol or other drugsNumber of alcohol and drug tests per monthPercent of total alcohol and drug tests done between 9:00 p.m. on Friday and noon on

SundaysPercent of total alcohol and drug testing time done at high risk places or on high risk

peoplePercent of alcohol and drug tests done when stopping a driver for a trafc offenseNumber of motorists failing alcohol or drug tests in the last four months

Objective 3: decrease hazardous driving behaviorPercent of total trafc monitoring time spent monitoring hazardous passingPercent of total trafc monitoring time spent monitoring driving toocloseto thevehicle

in frontPercent of total trafc monitoring time spent monitoring vehicles crossing roads

unsafelyNumber of reported hazardous driving incidents in the last four months

Objective 4: do efcient police administrationPercent of correctly and legibly completed trafc violation reportsPercent of time spent on internal administration in relation to total amount of time

available (lower is better)Percent of correctly and legibly written reports from vehicle inspectionsPercent of investigations completed within thirty days



7/13

Four dimensions of the TCI were expected to be inuenced by theprocess of doing ProMES. The rst dimension, vision, refers to howclearly dened, valued, shared, and attainable are the team'sobjectives (N. Anderson & West,1996). By clearly dening and sharingthe organizational vision, employees become more empowered toassume authority over their own actions toward the vision ( Nanus,1992), which are directed toward relevant aspects of performance andaway from irrelevant or undesired aspects (Kouzes & Posner, 1995;

Locke et al., 1991). The process of doing ProMES was hypothesized tobe likely to inuence vision and shared objectives. In a typical ProMESintervention, unit personnel agree on the measures used to assesswhetherobjectives are being met. Once completed, theyare presentedto higher-level management for approval. This process therefore wasexpected to produce a climate where goals are clear, agreed, accepted,and seen as valuable. Thus, it was expected that vision would improvefrom before the start of the design team's work to after feedback hasbeen in place.

Participative safety, the second dimension of the TCI, reects thedegree to which group members feel comfortable and secure withtheir fellow group members to share information and involve eachother in decision-making (N. Anderson & West, 1996). It was expectedthat participative safety increases in a ProMES project because of thehighly collaborative nature of the system. Specically, the majority ofthe design team members were actual employees in the unit. Duringthis process, the facilitator was responsible for ensuring all teammembers had an opportunity to express their views. By providing thesupport to contribute to discussions, the facilitator was likely creatingan atmosphere in which team members feel safer in expressing theirown views. Furthermore, developing and using the system typicallyuncovers conicts and the facilitator helps the group work throughthese conicts in a constructive way. After feedback starts, all unitmembers tend to be heavily involved in the feedback meetings. Thus,there is usually ample opportunity over an extended period of time tolearn participative skills and become comfortable with them. This ledto the prediction that ProMES would have a positive impact onparticipative safety.

The third factor is task orientation, which is the commitment of

the team to achieve the highest possible standards of task perfor-mance, including the use of constructive progress monitoringprocedures(N. Anderson & West, 1996, p. 59). ProMES was expectedto inuence this dimension because shared concern for performancewould be fostered by identifying and agreeing on the objectives, theindicators, and the contingencies. Furthermore, during the develop-ment process, the facilitator is typically responsible for ensuringdiscussions focus on task relevant issues. This process producesconsiderable involvement by unit personnel and considerable interestin how well they will perform on the group level measures. Appraisaland constructive challenges to group performance would likely occurin the feedback meetings. Thus task orientation was expected toincrease as a result of ProMES.

The nal factor predicting team innovation is support for

innovation. This dimension includes expectations and approval forand support of attempts at innovation. According toN. Anderson andWest (1996), this is the degree of practical support for innovationattempts contrasted by the rhetoric of professed support by seniormanagement (p. 59). This dimension was expected to be higherbecause of the commitment needed by both upper management andthe employees actually designing the system. By committing to doingProMES, the group typically goes through a process that shoulddevelop group norms for innovation because the major reason for thediscussions in the feedback meetings are to make such changes andinnovations. There is support for this not only by the previousexperience of working together to develop a measurement andfeedback system that everyone accepts, but by having the time, place,and the information to develop and evaluate these innovations. Thus,

support for innovation was expected to increase in a ProMES project.

Hypotheses 3-6. Employees have higher scores on the vision(Hypothesis 3), participative safety (Hypothesis 4), task orientation(Hypothesis 5), and support for innovation (Hypothesis 6) dimensionsduring feedback than during baseline.

Method

Participants and procedure

The participants in this study constituted the entire trafc policepersonnel in rebro County, an area near the center of Sweden, duewest of Stockholm. The implementation of ProMES started during theautumn 1996, and was done with three teams or shifts of policeofcers in the trafc division, a total of thirty-three people at the startof the project. The ofcers in the three separate shifts wereresponsible for monitoring compliance with speed limits, investigat-ing trafc accidents, checking for motorists' substance abuse, andgenerally maintaining safety on the highways in this county.

The ProMES process used with the three trafc units was thatsummarized in the introduction earlier and followed the proceduresoutlined in Pritchard (1990). The specic details of the processand theresulting system are described in Agrell and Malm (2002) in moredetail. The single design team was composed of representatives fromthe three teams/shifts and the chief constable, and worked closelywith the National Board for Road Maintenance (Vgverket), which isin charge of road safety. This design team met four hours a week forthree months to develop a single set of objectives, indicators, andcontingencies that would apply to all three groups. The system wasthen approved by management and the police ofcers in the units aswell as by the Vgverket representatives. The actual feedback wasseparate for each of the three units. Each reported their own indicatorscores, each had their own feedback reports, and each had separatefeedback meetings. The four objectives and eighteen indicatorsdeveloped by these units are shown in Table 2. Feedback was giventhrough a computer program designed to be used with ProMES calledProMES Navigator (more information is available from the Web sitehttp://promes.cos.ucf.edu/computerprogram.php). This program pro-

videsboth for entry of indicator data and a varietyof types of feedbackreports.

The three units received their rst feedback reports in January1997. Thus, data collected during 1996 were considered baseline dataand data collected in 1997 and after were considered data under thefeedback condition. The units used ProMES feedback for approxi-mately ve years, until the end of 2001.

Measures

Team climate for innovationTeam climate for innovation was measured with the Swedish

version of the TCI (Agrell & Gustafson, 1994; N. Anderson & West,1998). TCI questionnaires were given to ofcers in the three units

before the ProMES design team started their work (in 1996), againafter one year, at which point ProMES feedback had been operating forsix months (in 1997), andnally after several years of feedback (2001).

Each questionnaire contained thirty-eight items related to the fourfactors. Example items measuring vision included, How worthwhiledo youthink these objectives are to the organization? and How clearare you about what your team objectives are?Sample participativesafety items included, Peoplefeel understood and accepted by eachother,and Everyone's view is listened to, even if it is in a minority.Sample task orientation items included, Do members of the teambuild on each others ideas in order to achieve the best possibleoutcome?and Do you and your colleagues monitor each other so asto maintain a higher standard of work? Sample support forinnovation items included, People in this team are always searching

for fresh, new ways of looking at problems,

and Team members

http://promes.cos.ucf.edu/computerprogram.phphttp://promes.cos.ucf.edu/computerprogram.php


8/13

provide practical support for new ideas and their applications.Respondents were required to indicate the extent to which theyagreed with each item on a scale ranging from one (strongly disagree)to ve (strongly agree). The Swedish TCI is completed individually butis conceptualized and operationalized as a group-level measure. Thus,the team climate score for each of the four factors was the mean scoreacross the individuals in that group for that factor.

Trafc accidents, injuries, and fatalitiesData on trafc accidents,injuries, and fatalities were collected fromrebro County as well as the rest of Sweden. Data were collectedduring the baseline period (1995 and 1996) and during the feedbackperiods from 1997 to 1999. After 1999, however, the data were nolonger comparable to earlier years. New methods of collecting anddening these events were put into place. In addition, the construc-tion of new highways substantially changed trafc patterns makingcomparison with previous data problematic.

Results

Effects of the ProMES intervention

The rst hypothesis dealt with whether ProMES feedback wouldimprove performance. Fig. 2 shows these results. The horizontal axis istime showing baseline data available for ve months, from August1996 to December 1996. Feedback started in January 1997 and datawere available from then until October 2001.

The vertical axis is the overall effectiveness score averaged acrossthe three units. This is the sum of the effectiveness scores from all theindicators. Fig. 2 indicates that in the months prior to the start ofProMES feedback, overall effectiveness was low. That is, their overalleffectiveness scores were negative and well below the minimumexpected level: an overall effectiveness score of zero. In fact, theaverage baseline score is - 520.4. Once feedback started, however,there was a rapid rise in overall effectiveness during the rst six toeight months of feedback. This increase was maintained with somevariation throughout 1998, 1999, and the rst months of 2000. The

mean overall effectiveness score for the feedback period was 276.7.Each of the three trafc police units showed results similar to thoseshown in the gure.

Another way to look at the magnitude of the improvements is tolook at effect sizes. Effect size used was d(Hunter, Schmidt, & Jackson,1982). To calculate the effect size, the mean difference in the overall

effectiveness score between feedback and baseline was calculated,which was thendivided by the pooled standard deviation. These effectsizes were 2.78, 4.19, and 3.22 for the three units. Put another way, themean overall effectiveness score with ProMES feedback was 2.8 to 4.2standard deviations higher than it was during baseline. An effect sizeof .8 is considered large (Cohen, 1992), so the values obtained in thisstudy were very large.

Accidents, injuries, and fatalities

Hypothesis 2 stated that if overall effectiveness scores improveduring ProMES feedback, external measures of accidents, injuries, andfatalities also improve. Figs. 3 and 4 show the changes in trafcaccidents, injuries, and fatalities over the time periods from 1995 to1999 for rebro County and the rest of Sweden. Fig. 3shows theseoutcomes of interest in terms of the mean number of accidents,injuries, and fatalities for rebro County compared to the rest ofSweden.Fig. 4shows the same information as a percent of baseline.

As shown inFig. 3, the data indicate that the number of accidents,injuries, and fatalities decreased after feedback started and continuedto further decrease each year of the feedback. First, regardingaccidents, the fact that the average Swedish county had moreaccidents is not surprising because rebro County is smaller thanthe average Swedish county. What is important, however, is thechange in the number of accidents from 1995-96 to 1997-99.Specically, the number of accidents during the feedback period was88 percent of what it was during baseline for the average Swedishcounty but was 82 percent of what it was during baseline for rebroCounty. Thus, the decrease in accidents in rebro County decreased abit more than did the accidents throughout Sweden.

This can also be viewed as a percentage of baseline, which makes iteasier to compare the baseline to feedback and the rebro Countyresults with the rest of Sweden. To compute this, the mean number ofaccidents during the two-year baseline (1995-96) was calculated.Then the values for each of the subsequent years were calculated as apercentage of this baseline. This was done separately for all theoutcome measures for rebro County and for the rest of Sweden. As

shown inFig. 4, both the rebro County units and the rest of Swedenstart, by denition, with a value of 100 percent in the baseline years(i.e., the mean of the two years of baseline for both is set at 100percent). Then, the ProMES units and the rest of Sweden both decreasein accidents and decrease the same percentage the rst year offeedback. After that, the ProMES units continue to decrease, whereastherest of Sweden stays roughly equal to the 1997 values.By 1999, theProMES units' accident rate is 72 percent of what it was in baseline,while it was 87 percent for the rest of Sweden. This decrease in trafcinjuries for rebro County was signicant (2=34.82,pb .001). Thus,Hypothesis 2 was supported in terms of an improvement in reportedaccidents.

Similar results emerged for injuries. Specically, for rebroCounty, injuries under the feedback period were 90 percent of

what they hadbeen duringbaseline. Forthe rest of Sweden, themeannumber of injuries increased; during feedback injuries were 102percent of what they were in the baseline years. Thus, the number ofinjuries decreased with ProMES feedback and decreased relative tothe rest of Sweden.Fig. 4shows the same results, but viewed as apercentage of baseline. rebro County had fewer and fewer trafcinjuries each year of feedback compared to their baseline andcompared to the rest of Sweden. For injuries in 1999, the ProMESunits in rebro County were at 87 percent of their baseline valueswhereas the trafc units in the rest of Sweden were at 106 percent oftheir baseline. So whereas injuries in rebro County decreased overtime, the number of injuries in the rest of Sweden increased. Thisdecrease in trafc injuries for rebro County was signicant(2=15.62, pb .001), thus supporting Hypothesis 2 in terms of an

improvement in reported injuries.Fig. 2.ProMES overall effectiveness over time.



9/13

Finally, similar results emerged for trafc fatalities, although thedecreases were not statistically signicant (2=2.38,ns), possibly dueto the relativelysmall numbers of fatalities in general. Nevertheless, asshown in Figs. 3 and 4, the number of trafc fatalities in rebro Countydecreased with feedback compared to baseline, with the number offatalities during feedback being 72 percent of what it had been duringbaseline. For the rest of Sweden, however, the mean for the baseline

period was equal to the mean for feedback, with an increase to 118percent of baseline by 1999. Thus, fatalities for the ProMES unitsdecreased during feedback whereas the fatality rate for the rest ofSweden did not change or got worse. Thus, Hypothesis 2 was notsupported in terms of an improvement in reported fatalities in astatistical sense, but it could be argued that there was an improve-ment in a practical sense.

Number of ofcers

Although no hypotheses were made related to the number ofofcers during the intervention the data were instructive. The bottomright quadrants inFigs. 3 and 4show the number of police ofcers inrebro County over the same time period as the outcome measures.

As with the rest of Sweden, the number of police ofcers was

decreasing due to budget decreases. In rebro County, the number oftrafc police ofcers decreased from baseline (M=32.5) to thefeedback period (M=26.7). By 1999, the number of ofcers wasdown to twenty-three. Although not shown on the gure, the numberof police in the rest of Sweden also decreased, although actualnumbers were not available. As shown in Fig. 4, by 1999 thenumber ofofcers was 71 percent of what it had been during baseline. Thus,

rebro County had a substantial loss of ofcers, yet substantiallyimproved their performance.

Team climate perceptions

Hypotheses 3-6 concerned the effects of the ProMES interventionon the four TCI dimensions. For each dimension, planned pairwisegroup comparisons were conducted to compare the team climateperceptions during baseline and feedback. A prioricomparisons weremade for eachof the fourdimensions between scores obtained in 1996(baseline) and the average of those obtained in 1997 and 2001(feedback). For each of the four dimensions, it was hypothesized thatthe scores during feedback would be greater than those duringbaseline. Results yielded support for all dimensions except support

for innovation. Specically, for participative safety, the mean score

Fig. 3.Effects of ProMES on external measures: comparison to baseline and the rest of Sweden.



10/13

increased from 3.63 to 4.10, t=3.62, pb .001. Vision also increased,from a mean score of 3.67 to 3.98, t=2.96, pb .01. Task orientationincreased from a mean score of 3.57 to 3.81, t=2.03,pb .05. Supportfor innovation, although increasing from a mean score of 3.27 to amean score of 3.44 was not a signicant increase, t=1.26, ns . Thus,Hypotheses 3, 4, and 5 were supported whereas Hypothesis 6 was not

supported.In addition totesting forstatistical signicance, thetreatmenteffect

omega-squared (2) was computed to examine whether differenceswere meaningful between the scores obtained during baseline andfeedback periods. Scores of .01, .06, and greater than .15 representsmall, medium,and large treatment effects,respectively (Cohen,1977).The effect size was large for participative safety (2= .15) medium-to-large for vision (2=.12), and medium for task orientation (2=.06).Interestingly, although not statistically signicant (pN .05), the effectsize for support for innovation was medium-to-large (2=.08).

Discussion

The primary purpose of this study was to determine the effects of

the ProMES intervention on performance and attitudes. The ProMES

effectiveness scores indicated large gains in performance and themeasures external to the system (accidents, injuries, and fatalities)decreased relative to baseline and relative to the rest of Sweden,although the reduction in the number of fatalities was not statisticallysignicant. This was all occurring despite the decreases in the numberof ofcers over time.

The results suggest that ProMES is an effective way of respondingto the governmental requirements for measuring police effectivenessand also increasing performance in the face of reduced personnel. Italso helps align the efforts of the ofcers with the broader goals of theorganization by the way the measurement system is developed. Theobjectives, indicators, and contingencies are reviewed by higher levelsof management and a key issue is how well they are aligned withbroader organizational goals. Once the measurement system isapproved, the resulting feedback system provides information onhow to allocate resources so as to maximize contribution to theorganization.

Ofcer attitudes toward their unit and their department and theclimate for innovation showed that the scores during feedback weremore positive than during baseline, with the exception of the support

for innovation dimension. Thus, the feedback provided by the ProMES

Fig. 4.Effects of ProMES on external measures: percent of baseline information.



11/13

intervention led to increases in vision, participative safety, and taskorientation.

The problematic measurement issues of control over outputmeasures and detection of negative events were explored. Controlover outcomes was addressed by having the design team developindicators they did control and that they believed would lead to theultimate organizational goals of decreasing accidents, injuries, andfatalities. They did improve on their measures and as they predicted,

the less controllable measures of accidents, injuries, and fatalities alsoimproved. The ofcers were satised that they had control over theirmeasures and managers in the broader organization were satisedwith the improvements in ultimate objectives.

Although no actual data on the issue of detection of negativeevents were presented, information was presented regarding how theofcers addressed this in their own situation, by including theirdetection effortsand the frequency of negative events. This solutionwas accepted by all involved, and may prove to be a useful solution topractitioners facing similar dilemmas.

Practical considerations

It is useful to consider why these improvements occurred. Whilethe information available is anecdotal, it is quite consistent. This wasthe rst time these ofcers took the time to sit down and clarify theirvision and what objectives were really important in doing the work.The combination of doing this and actually developing the measureswas difcult and time-consuming. By the end of this process, however,the ofcers had quite a different idea of how to do their work. Bypooling their knowledge and experience, they realized that a betterstrategy would be to focus on those things that would have thegreatest impact such as patrolling more frequently on some roads, atcertain times, and on certain days. They also saw the necessity forcompleting paper work accurately and minimizing the time spent onadministration.

They were quite surprised when the rst feedback indicated thatthey were not actually doing these things very well. This led to manyattempts to change the way they did the work. They were then able to

assess how good the new strategies were by studying the subsequentfeedback reports. Changes in strategy that led to improvements werekept, changes that did not have an effectwere revised. This processledto large improvements in their feedback scores. The steadilyincreasing feedback scores led to considerable positive affect amongthe ofcers. Over time when it became clear that accidents, injuries,and fatalities were going down, this produced further positive affectand increased their desire to perform well.

Anotherissue to consider in evaluating the results is the time takento develop the system. The design of the measurement system, gettingmanagement approval, and nalizing the feedback system tookapproximately three months, with the design team meeting fourhours a week. This is longer than usual, with the average design teammeeting time thirty to thirty-ve hours (Pritchard, 1995). There were

three different units, however, represented in the design team ratherthan the usual one unit. Thus, the longer time is not surprising.

Although the time to develop the systemwas longerthan usual,thetime it took to enter the data during feedback was minimal, especiallyonce the department implemented the use of software designed tohelp manage the process and provide reports in a variety of formats.Anecdotally, the feedback was also more meaningful, as the feedbackwas more readily available and the variety in how the data could beviewed (e.g., overall effectiveness scores, separate indicator scores,shifts combined or separate) allowed greater understanding.1

Limitations and future directions

Although the team climateperceptions of vision, participativesafety,

and taskorientation improved during ProMES comparedto baseline, the

reason they increased can only be speculated. It would seem that theProMES process itself, with its participative aspects and process ofclarifying roles and expectations would be an inuencing factor in theseimprovements, but these claims cannot be made without furtherexamination of the intervening variables and thought processes of theofcers themselves. Future researchers and practitioners who useProMES, or any intervention aimed at increasing performance or teamclimate perceptions (or both), would do well to consider some of the

explanatory variables in their design of the system.The focus of the current intervention was on the overallperformance improvement of the group of ofcers, and not onindividuals within the department. Thus, no individual differenceswere assessed in this study's measurement. Individual differences,however, could have impacted some of the results found within thecurrent study. For example, individuals vary on motivation, intelli-gence, and personality traits. Each of these could inuence how mucheffort they put into a task or their perceptions with an outcome.Furthermore, it is not possible to say whether the individuals who leftand those who remained (or joined the department) during theintervention were different from one another. Nevertheless, resultswere separated by time period (baseline versus feedback), and hadthere been an inuencing effect of individual differences, it would beexpected that they would counter each other as such variables arefrequently normally distributed and likely to appear in both condi-tions. Thus, although these issues could not be addressed in thecurrent study, this was not perceived to be a substantial limitation tothe current study's design, nor did it detract from the study's ndings.

Additionally, although the number of ofcers was modest, thisshould not discredit or diminish the ndings of this study for severalreasons. First, although there were only thirty-three ofcers at thestart of the project, this was all of the trafc police from this county(so not an issue of response rate). Second, there was no reason tobelieve that the make-up of the ofcers in this particular county wasany different than the make-up of the rest of Sweden. Third, acritical issue was whether the ndings would generalize to otherofcers within Sweden, but also to other ofcers and departmentsoutside of Sweden. As noted, ProMES has been effective in many

settings in many countries. Thus, the positive results found withinthis study were consistent with a broad pattern of similarly positiveresults. Therefore, although the specic variables that wereexamined in this study were different than those examined inother studies, the odds of these ndings generalizing to othersituations and departments was higher based on the similarlypositive results in other studies (e.g., comparably high effect sizes).Finally, information was presented in terms of actual numbers (e.g.,accidents, injuries, and fatalities) for both rebro County as well asthe rest of Sweden, but also in terms of percent of baseline. This wasdone in order to account for the fact that rebro County is smallerthan the average county, as well as to make meaningful comparisonsthat were independent of sample size.

Finally, despite the improvements in the performance of the

ofcers, the intervention was ultimately discontinued, and thuslimited thelength of time that couldbe examined.Therewereseveralreasons the project ended, including a new supervisor whosemeasurement system included a number of new projects (so ProMESdid not count), and most importantly, when extra money was madeavailable for salary increases, the units doing ProMES did not get anyextra for their improvements in performance with fewer personnel.This caused a negative reaction and loss of enthusiasm for the effortleading to ending it. Based on this, organizations that choose toimplement any innovation should consider the reward structurethat is set in place to support and encourage effective performancebased on the innovation. Clearly, alignment of performance andgoals is important; similarly, alignment between performanceand reward structures is clearly essential for continued success of

an intervention.



12/13

Conclusions

In conclusion, this article describes the results of a feedback systemdesigned to improve performance for a Swedish trafc police unit andexamineswhethersuch a feedback system is benecialor detrimentaltotheattitudesof theofcers.Results indicated thatthere were substantialincreases in performance and signicantdecreases in accidents, injuries,and fatalities compared both to baseline and to comparison groups in

Sweden, all while number of of

cers was decreasing.

Acknowledgements

This article expands on work presented earlier in a chapter byAgrell and Malm (2002). Additional data are presented, moreappropriate analyses are used, additional measurement issues areaddressed, and implications of the results for criminal justice settingsare presented.

Note

1. For further information on the specics of the program ProMES Navigator,contact Kenneth Malm [email protected].

References

Agrell, A., & Gustafson, R. (1994). The team climate inventory and group innovation: Apsychometric test on a Swedish sample of work groups.Journal of Occupational andOrganizational Psychology,67, 143151.

Agrell, A., & Malm, K. (2002). ProMES in a Swedish trafc police department and itseffects on team climate. In R. D. Pritchard, H. Holling, F. Lammers, & B. D. Clark(Eds.),Improving organizational performance with the Productivity Measurement andEnhancement System: An international collaboration (pp. 5368). Huntington, NY:Nova Science.

Anderson, C. A. (1989). Temperature and aggression: Ubiquitous effects of heat onoccurrence of human violence.Psychological Bulletin,106, 7496.

Anderson, N., & West, M. A. (1996). The teamclimate inventory:Developmentof the TCIand its applications in teambuilding for innovativeness. European Journal of Workand Organizational Psychology,5, 5366.

Anderson, N., & West, M. A. (1998). Measuring climate for work group innovation:Development and validation of the team climate inventory. Journal of Organiza-tional Behavior,19, 235258.

Bobko, P., & Colella, A. (1994). Employee reactions to performance standards: A reviewand research propositions.Personnel Psychology,47, 129.

Brown, T. C., & Latham, G. P. (2002). The effects of behavioral outcome goals, learninggoals,and urgingpeople todo their best on anindividual'steamwork behaviourin agroup problem-solving task. Canadian Journal of Behavioural Science,34, 276285.

Bukenya, J. O. (2005). Crime trend and socio-economic interaction: A county-levelanalysis. Criminal Justice Studies: A Critical Journal of Crime, Law and Society, 18,365378.

Campbell, J. P. (1977). On the nature of organizational effectiveness. In P. S. Goodman,J. M. Penning s, & Associates (Eds.), New perspectives on organizational effective-ness(pp. 13-55). San Francisco: Jossey-Bass.

Campbell, J. P., & Campbell, R. J. (1988). Industrial-organizational psychology andproductivity: The goodness oft. In J. P. Campbell & R.J. Campbell (Eds.), Produc-tivity in organizations (pp. 8294). San Francisco: Jossey-Bass.

Campbell, J. P., & Pritchard, R. D. (1976). Motivation theory in industrial andorganizational psychology. In M. D. Dunnette (Ed.), Handbook of industrial andorganizational psychology(pp. 63130). Chicago: Rand McNally.

Cawley, B. D., Keeping, L. M., & Levy, P. E. (1998). Participation in the performance

appraisal process and employee reactions: A meta-analytic review of eldinvestigations.Journal of Applied Psychology,83, 615633.Christens,B., & Speer, P. W. (2005). Predicting violent crime using urban and suburban

densities.Behavior and Social Issues,14, 113127.Cohen, J. (1977). Statistical power analysis for the behavioral sciences. New York:

Academic Press.Cohen, J. (1992). A power primer.Psychological Bulletin,112, 155159.Dipboye, R. L., & de Pontbriand, R. (1981). Correlates of employee reactions to

performance appraisals and appraisal systems. Journal of Applied Psychology, 66 ,248251.

Eisenberger, R., Fasolo, P., & Davis-LaMastro, V. (1990). Perceived organizational supportand employee diligence, commitment, and innovation. Journal of Applied Psychol-ogy,75, 5159.

Fisher, C. D., & Gitelson, R. (1983). A meta-analysis of the correlates of role conict andambiguity.Journal of Applied Psychology,68, 320333.

Frese, M., & Zapf, D. (1994). Action as the core of work psychology: A German approach.In H. C. Triandis, M. D. Dunnette, & L. M. Hough (Eds.), Handbook of industrial/organizational psychology(2nd ed., Vol. 4, pp. 271340). Palo Alto, CA: ConsultingPsychologists Press.

Harrison, S. J. (1996, January). Quality policing and challenges for leadership. PoliceChief,LXIII, 2632.

Heckhausen, H. (1991).Motivation and action.Berlin, Germany: Springer.Hipp,J. R.,Bauer, D. J.,Curran,P.J., & Bollen,K. A. (2004). Crimesof opportunityor crimes

of emotion? Testing two explanations of seasonal change in crime.Social Forces,82,13331372.

Hunter, J. E., Schmidt, F. L., & Jackson, G. B. (1982). Meta-analysis: Cumulating researchndings across studies.Beverly Hills, CA: Sage.

Ilgen, D. R., Fisher, C. D., & Taylor, M. S. (1979). Consequences of individual feedback onbehavior in organizations.Journal of Applied Psychology,64, 349371.

Jackson, S. E., & Schuler, R. S. (1985).A meta-analysis and conceptual critique of research

on role ambiguity and role conict in work settings. Organizational Behavior andHuman Decision Processes,36, 1678.Kahn,R. L. (1977). Organizational effectiveness: An overview. In P. S. Goodman & J.M.

Pennings (Eds.),New perspectives in organizational effectiveness (pp. 235248). SanFrancisco: Jossey-Bass.

Kanfer, R. (1990). Motivation theory and industrial/organizational psychology. In M. D.Dunnette & L. Hough (Eds.), Handbook of industrial and organizational psychology:Theory in industrial and organizational psychology(Vol. 1, pp. 75170). Palo Alto, CA:Consulting Psychologists Press.

Kanfer, R. (1992). Work motivation: New directions in theory and research. In C. L.Cooper & I.T. Robertson (Eds.),International review of industrial and organizational

psychology(Vol. 7, pp. 153). London: John Wiley and Sons.Kanfer, R., & Ackerman, P. L. (1989). Motivation and cognitive abilities: An integrative/

aptitude-treatment interaction approach to skill acquisition [Monograph].Journalof Applied Psychology,74, 657690.

Kanfer, R., Ackerman, P. L., Murtha, T. C., Dugdale, B., & Nelson, L. (1994). Goal setting,conditions of practice, and task performance: A resource allocation perspective.

Journal of Applied Psychology,79, 826835.Kaplan, R. S., & Norton, D. P. (1993, September-October). Putting the balanced scorecard

to work.Harvard Business Review, 134147.Kaplan, R. S., & Norton, D. P. (1996). Translating strategy into action: The balanced

scorecard.Boston: Harvard Business School Press.Kouzes, J. M., & Posner, B. Z. (1995). The leadership challenge: How to keep getting

extraordinary things done in organizations. San Francisco: Jossey-Bass.Latham, G. P., Mitchell, T. R., & Dossett, D. L. (1978). Importance of participative goal

setting and anticipated rewards on goal difculty and job performance.Journal ofApplied Psychology,63, 163171.

Latham, G. P., & Pinder, C.C. (2005).Work motivation theoryand researchat thedawnofthe twenty-rst century.Annual Review of Psychology,56, 485516.

Lind, E. A., Kanfer, R., & Earley, P. C. (1990). Voice, control, and procedural justice:Instrumental and noninstrumental concerns in fairness judgments. Journal ofPersonality and Social Psychology,59, 952959.

Locke, E. A., Kirkpatrick, S. A., Wheeler, J., Schneider, J., Niles, K., Goldstein, H., et al.(1991).The essence of leadership. New York: Lexington Books.

Locke, E. A., & Latham, G. P. (2002). Building a practically useful theory of goal settingand task motivation: A 35-year odyssey. American Psychologist,57, 705717.

Locke, E. A., & Schweiger, D. M. (1979). Participation in decision-making: One morelook.Research in Organizational Behavior,1, 265340.

London, M. (2003). Job feedback: Giving, seeking, and using feedback for performanceimprovement(2nd ed.). Mahwah, NJ: Lawrence Erlbaum.

Mahoney, F. X., & Thor, C. G. (1994). The TQM trilogy. New York: AMACOM.Mitchell, T. R., & Daniels, D. (2003). Motivation. In W. C. Borman, D. R. Ilgen, & R. J.

Klimoski (Eds.), Handbook of psychology: Industrial and organizational psychology(Vol. 12, pp. 225254). Hoboken, NJ: John Wiley and Sons.

Muckler, F. A. (1982). Evaluating productivity. In M. D. Dunnette & E.A. Fleishman (Eds.),Human performance and productivity: Human capability assessment(Vol.1, pp. 1347).Hillsdale, NJ: Erlbaum.

Murphy, K. R., & Cleveland, J. N. (1995). Understanding performance appraisal: Social,organizational, and goal-based perspectives.Thousand Oaks, CA: Sage.

Nanus, B. (1992). Visionary leadership: Creating a compelling sense of direction for yourorganization.San Francisco: Jossey-Bass.

Naylor, J. C., Pritchard, R. D., & Ilgen, D. R. (1980). A theory of behavior in organizations.New York: Academic Press.

Pritchard, R. D. (1990). Measuring and improving organizational productivity: A practicalguide.New York: Praeger.

Pritchard, R. D. (Ed.). (1995).Productivity measurement and improvement: Organizational

case studies.New York: Praeger.Pritchard, R. D., & Ashwood, E. L. (2008). Managing motivation: A manager's guide todiagnosing and improving motivation.New York: LEA/Psychology Press.

Pritchard, R. D., Harrell, M., DiazGranados, D., & Guzman, M. J. (2008). The ProductivityMeasurement and Enhancement System: A meta-analysis. Journal of AppliedPsychology,93, 540567.

Pritchard, R. D., Holling, H., Lammers, F., & Clark, B. D. (Eds.). (2002).Improvingorganizational performance with the Productivity Measurement and EnhancementSystem: An international collaboration. Huntington, NY: Nova Science.

Pritchard, R. D., Jones, S. D., Roth, P. L., Stuebing, K. K., & Ekeberg, S. E. (1989). Theevaluation of an integrated approach to measuring organizational productivity.Personnel Psychology,42, 69115.

Pritchard, R. D., Youngcourt, S. S., Philo, J. R., McMonagle, D. C., & David, J. H. (2007).Priority information in performance feedback.Human Performance,20, 6183.

Rotton, J. (1983). Affective and cognitive consequences of malodorous pollution. Basicand Applied Social Psychology,4, 171191.

Rotton, J., & Frey, J. (1985). Air pollution, weather, and violent crimes: Concomitanttime-series analysis of archival data.Journal of Personality and Social Psychology,49,12071220.



13/13

Rutter, M., Giller, H., & Hagell, A. (1998).Antisocial behavior by young people.Cambridge,UK: Cambridge University Press.

Salas, E., Kosarzycki, M. P., Tannenbaum, S. I., & Carnegie, D. (2004). Principles andadvice for understanding and promoting effective teamwork in organizations. In R.Burke & C. Cooper (Eds.), Leading in turbulent times: Managing in the new world ofwork(pp. 95120). Malden, MA: Blackwell.

Salas, E., Rosen, M. A., Burke, C. S., Goodwin, G. F., & Fiore, S. (2006). The making of adream team: When expert teams do best. In K. A. Ericsson, N. Charness, P. J.Feltovich, & R. R. Hoffman (Eds.), The Cambridge handbook of expertise and expert

performance(pp. 439453). New York: Cambridge University Press.Smither, J. W., London, M., & Reilly, R. R. (2005). Does performance improve following

multisource feedback? A theoretical model, meta-analysis, and review of empiricalndings.Personnel Psychology,58, 3366.Spector, P. E. (1986). Perceived control by employees: A meta-analysis of studies

concerning autonomy and participation at work. Human Relations,39, 10051016.Taylor, M. S., Fisher, D., & Ilgen, D. (1984). Individual reactions to performance feedback

in organizations: Control theory perspective. In K. Rowland & G. Ferris (Eds.), Re-search in personnel and human resource management(pp. 81124). Greenwich, CT:

JAI Press.Tubr,T. C.,& Collins, J. M. (2000). Jackson andSchuler (1985) revisited: A meta-analysis

of the relationships between role ambiguity, role conict, and job performance.Journal of Management,26, 155169.

Vroom, V. H. (1964).Work and motivation.New York: Wiley.Wall, T. D., Corbett, J. M., Martin, R., Clegg, C. W., & Jackson, P. R. (1990). Advanced

manufacturing technology, work design, and performance: A change study.Journalof Applied Psychology,75, 691697.

Wall, T. D., & Lischeron, J. H. (1977). Worker participation: A critique of the literature andsome fresh evidence.Maidenhead, England: McGraw-Hill.

Wang, F. (2005). Job access and homicide patterns in Chicago: An analysis at multiplegeographic levels based on scale-space theory.Journal of Quantitative Criminology,

21, 195217.West, M. A. (1990). The social psychology of innovation in groups. In M. A. West & J.L.

Farr (Eds.), Innovation and creativity at work: Psychological and organizational

strategies(pp. 436). Chichester, England: Wiley.West, M. A. (1994). Effective teamwork. Leicester, England: BPS Books.West, M. A. (2007, May).Flourishing people teams and organizations: The challenge for

work and organizational psychology. Keynote address presented at the meeting ofthe European Congress of Work and Organizational Psychology, Stockholm.

West, M. A., & Anderson, N. R. (1996). Innovation in top management teams. Journal ofApplied Psychology,81, 680693.

West, M. A., & Farr, J. L. (1989). Innovation at work: Psychological perspectives. SocialBehaviour,4, 1530.


Improving Performance in a Swedish Police Traffic Unit Results of an Intervention

Documents

Transcript of Improving Performance in a Swedish Police Traffic Unit Results of an Intervention