Development and Challenges of Community Policing

Exploring how an area’s crime-to-cop ratios impact patrol officer productivity

Luke Bonkiewicz Lincoln Police Department, Lincoln, Nebraska, USA

Abstract Purpose – The purpose of this paper is to examine how the combined crime rate and staffing levels of a patrol area affect patrol officers’ productivity. Specifically, the author identified and analyzed two macro-level correlates of patrol officer productivity: reported violent crimes per officer and reported property crimes per officer (a beat’s “crime-to-cop” ratios). Design/methodology/approach – Using hierarchical linear modeling, the author estimated the effects of a patrol area’s violent crimes per officer ratio and property crimes per officer ratio on the annual number of traffic citations, warrants, misdemeanor arrests, and felony arrests generated by patrol officers (n ¼ 302). The author also examined the effect of these crime-to-cop ratios on a more advanced productivity metric. Findings – The results suggest that a patrol area’s rate of property crimes per officer is associated with a moderate decrease in an officer’s annual number of traffic citations, warrant arrests, and misdemeanor arrests; a patrol area’s rate of violent crimes per officer is also associated with a moderate decrease in an officer’s annual number of traffic citations; and a patrol area’s rate of violent crimes per officer is associated with a moderate increase in an officer’s annual number of warrant and misdemeanor arrests. Notably, the crime-to-cop ratios are not correlated with a more sophisticated patrol productivity metric. Research limitations/implications – The author analyzed data from a mid-sized US police department that uses a generalists policing style. It is unknown if these results translate to smaller or larger police departments, as well as those agencies practicing a specialized policing style. Practical implications – The findings suggest that police scholars should not only recognize how the crime-to-cop ratios of a patrol area might impact patrol officer productivity, but also incorporate more sophisticated metrics of patrol officer activity in future studies. These findings likewise signal to police practitioners that an area’s crime-to-cop ratios should be considered when allocating officers and other resources across patrol areas. Originality/value – To the authors knowledge, this is the first study to identify and examine the link between a patrol area’s crime-to-cop ratios and patrol officer productivity. Keywords Patrol, Productivity, Crime-to-cop ratio, VORC Paper type Research paper

Introduction Perhaps no civil servant undertakes the variety of tasks as a municipal police officer. As not only law enforcement agents, but also community caretakers and first-responders to every calamity great and small, patrol officers handle a broad array of calls for service (CFS) during their daily tour of duty. The regular duties of a patrol officer include reactive tasks, such as handling CFS, as well as proactive tasks, such as conducting traffic and scofflaw enforcement, searching for suspected

Policing: An International Journal of Police Strategies & Management

Vol. 39 No. 1, 2016 pp. 19-35

© Emerald Group Publishing Limited 1363-951X

DOI 10.1108/PIJPSM-05-2015-0064

Received 21 May 2015 Revised 22 July 2015 13 September 2015

Accepted 16 September 2015

The current issue and full text archive of this journal is available on Emerald Insight at: www.emeraldinsight.com/1363-951X.htm

The author would like to thank Denver Police Department Crime Analyst Melissa Ohlhorst and Pennsylvania State University Professor Barry Ruback, as well as the three anonymous reviewers, for their insightful comments. The author also salute Lincoln Police Sergeant Jake Dilsaver for his logistical support of this project.

19

Patrol officer productivity

perpetrators, and patrolling for suspicious individuals or situations (e.g. drug deals, graffiti, vandalisms, or individuals “casing” a potential target).

Historically, both police researchers and practitioners have grappled with the lack of a valid and reliable method for evaluating patrol officer activity, as well as limited knowledge about how external factors affect officers’ productivity (Crank, 1990; Johnson, 2011; Shane, 2010). Although studies have demonstrated that differences in productivity between patrol officers may result from biographical factors and daily work experiences (Shane, 2011), operational variables (Amaranto et al., 2003), and organizational stressors (Brown and Campbell, 1990), scholars have devoted considerably less attention to how the characteristics of an area might impact police productivity. In other words, although individual-level variables impact a patrol officer’s productivity, how might the characteristics of a patrol area affect a police officer’s productivity?

This paper identifies and analyzes two macro-level correlates of patrol officer productivity: violent crimes per officer and property crimes per officer. We first review the extant literature on patrol officer productivity, discussing how an area’s violent and property crime per cop ratios might affect a patrol officer’s productivity. Next, we analyze productivity data from a mid-sized US police department to examine how these two variables impact the number of traffic citations, warrants, and misdemeanor and felony arrests generated by officers annually. We also examine the relationship between these two variables and a more sophisticated productivity metric, Value Over Replacement Cop (VORC). Finally, we discuss why police scholars and practitioners should pay closer attention to the practical, macro-level correlates of productivity, as well as incorporate more advanced patrol productivity metrics.

Literature review Productivity scholars have generally divided the concept of worker productivity into two spheres (Hatry, 1976). Effectiveness measures a worker or organization’s quality of service, while efficiency gauges how well a worker or organization produces a particular output using the least amount of resources, such as manpower, time, and money. Most police research has examined productivity in terms of efficiency, due mainly to the data available for analysis, such as CFS and arrests.

Police scholars have struggled with how to properly measure patrol officer productivity. There appear to be two main approaches. The first approach attempts to identify and quantify specific types of patrol productivity outputs. Some examples include evaluating productivity using an officer’s total arrests (Crank, 1990), clearance rates (Garicano and Heaton, 2010), number of traffic citations ( Johnson, 2011), and citizen complaints (Lersch, 2002). However, researchers have noted that evaluating productivity using these kinds of raw outputs is “conceptually simple,” shaped primarily by data availability (Chappell et al., 2006).

The second approach to analyzing patrol officer productivity has incorporated multi-dimensional indicators of police performance. For example, Shane (2011) conceptualized productivity using a z-score summary of multiple outputs, including arrests, traffic citations, field contacts, official reports, sick hours, and administrative complaints. Similarly, Van Meter’s (2001) Performance-Based Management, or “zero- based approach” uses three planks (non-scheduled absenteeism, cost of preventable error, and productive use of time) to evaluate patrol officers.

Notably, most existing police productivity research has ignored the concept of productive time, or the amount of time officers have for self-initiated activities after

20

PIJPSM 39,1

handling CFS, investigating crimes, completing reports and follow-ups, and assisting other officers. Using productive time to evaluate officers is critical because two officers could achieve the same output (in terms of raw arrests or reports), but have done so with considerably different amounts of productive time due to CFS, follow-ups, etc. Recent research has proposed the use of VORC (Bonkiewicz, 2015), a metric that incorporates more than a dozen patrol activities, weights those activities, evaluates performance in terms of productive time, and compares an officer’s performance against a fixed, hypothetical estimate (i.e. a replacement officer).

In addition to conceptualizing patrol officer productivity, researchers have also studied potential antecedents of productivity, including not only biographical variables but also factors involving workplace environment. First, certain biographical characteristics might impact productivity (Hawkins, 2001). For instance, officers who have children might experience sleep deprivation and greater fatigue as a result of parenting responsibilities, thereby reducing their workplace energy and motivation, and in turn, their productivity (Shane, 2010). Marital status may also impact productivity. Although the relationship between marriage and worker productivity is well debated in the literature (Nakosteen and Zimmer, 2001; Becker, 1985), it seems reasonable to believe that married officers might be more productive than unmarried officers. Married officers have someone in whom to confide, a partner who hears about the challenges and rigors of the job, which may lead to stress reduction and improve workplace motivation. An officer’s age is another consideration. Older officers might be less productive than younger officers, perhaps due to burnout as well as the physical demands of law enforcement (Goodman, 1990). Officers’ perceptions of their job may also affect their proactivity, and in turn, productivity. Brehm and Gates (1993) found that officers who disliked certain aspects of law enforcement (e.g. taking people to jail) impacted whether officers engaged in proactive activities or shirked their duties.

Additionally, there are also operational features (the “day to day” rigors) of law enforcement which can decrease productivity and increase misconduct, such as the perceived danger of the job (Amaranto et al., 2003), work schedules and shift work (Vila et al., 2002), and fatigue (Vila and Kenney, 2002). Notably, in studying the relationship between daily work experience and performance, Shane (2011) found that workload was the “most important predictor of performance” (p. 13).

We believe this last finding merits further study. First, supervisors evaluate patrol officers, at least in part, using productivity outputs. If there are factors (i.e. heavy workload) beyond an officer’s control that are affecting his or her productivity, both line-level supervisors and command staff members should acknowledge and account for these variables during performance reviews. Second, if macro-level variables create different workloads, thereby affecting productivity outputs, then police researchers should include these measures in future studies on police activity. We suggest that there are at least two macro-level variables important to future police productivity research.

Crime-to-cop ratios and patrol officer productivity Criminological research has long recognized that crime rates are not uniform across neighborhoods and cities (Sherman et al., 1989; Weisburd et al., 1992). Instead of “policing people,” law enforcement agencies frequently identify “hot spots” and devote additional resources to reducing crime and disorder in specific problem areas. With this perspective in mind, it might seem natural to examine the relationship between the

21

Patrol officer productivity

crime rate of a patrol area and an officer’s productivity outputs (or vice versa). For example, a high crime area might net an officer more arrests (and therefore, more productivity outputs) than a low crime area. On the other hand, a high crime area might also occupy a great deal of an officer’s productive time, prohibiting the officer from engaging in proactive activities. However, crime rate does not fully capture how a particular area might affect an officer’s productivity.

Instead, we believe macro-level variables related to crime should be estimated in terms of the number of patrol officers in a patrol area, specifically, violent crimes per officer and property crimes per officer. For instance, some officers may patrol a lower crime area but end up responding to a greater number of violent crimes than officers in a high crime area simply because they have fewer officers working the street. This “crime per cop” ratio can dramatically affect officers’ productivity.

For example, when an officer investigates a crime, he or she might interview the victim, suspect, and witnesses, as well as take photographs, collect evidence, and make an arrest. There may be additional follow-up the next day or week, such as checking pawnshops for stolen property or tracking down and interviewing additional witnesses. All these activities could decrease an officer’s time for self-initiated activities, and in turn, an officer’s overall productivity. A high crime per cop ratio, therefore, could decrease productivity for officers, even if these officers work in an area with a low crime rate. Furthermore, an area’s crime per cop ratio could also affect officers’ productive time even if they are not the primary investigating officer because secondary officers are often needed during serious incidents, such as robberies or burglaries.

The “load shedding” hypothesis by Maxfield et al. (1980; see also Maxfield, 1982) also has important implications for any study examining patrol officer productivity. Maxfield et al. suggested that while variations in workload do not affect patrol officers’ decisions to record serious crimes, workload does affect officers’ decisions to record less serious crimes; it may also affect self-initiated activities as well (Herbert, 1998).

Finally, we believe that macro-level variables affecting productivity should be divided between violent and property crimes to account for the time and resources needed to investigate each incident type. When officers respond to assaults, they often contact suspects and victims on scene, meaning that officers are able to interview all involved individuals, collect evidence, and make an arrest, sometimes in less than an hour. However, when officers respond to burglaries or larcenies, they may be required to dust for fingerprints, swab for DNA, canvass a neighborhood for witnesses, check pawnshops, and engage in other investigative activities that require hours and even days if the case requires follow-up.

With these issues in mind, our study addresses the following research questions:

RQ1. What is the relationship between a patrol area’s rate of violent crimes per officer and patrol officer productivity outputs (i.e. traffic citations, as well as warrant, misdemeanor, and felony arrests)?

RQ2. What is the relationship between a patrol area’s rate of property crimes per officer and patrol officer productivity outputs (again, traffic citations, and warrant, misdemeanor, and felony arrests)?

RQ3. What is the relationship between the rate of property and violent crimes per officer and a patrol officer’s productivity when we introduce a more sophisticated method for analyzing productivity, a metric based on productive time and not total time on duty?

22

PIJPSM 39,1

Before discussing our data and methods, there are two important issues meriting brief discussion. First, we argue that the crime-to-cop ratio may affect the number of traffic stops, warrants, misdemeanors, and felonies that a patrol officer makes. It can also be argued that the number of arrests might impact the crime-to-cop ratio. Our intent is not to comprehensively discuss the relationship between the number of officers/officer arrests and an area’s crime rate, but it bears noting that the research is inconclusive on this point.

Some evidence suggests that the relationship between arrests and crime rates is non-linear, that is, crime rates decrease as arrests increase but eventually hit a “floor” and reach equilibrium (Kane, 2006). Second, when studies examine the arrest-crime rate relationship, they often find that the relationship only exists for specific crimes in their data sets, such as robbery and motor vehicle thefts (Corman and Mocan, 2005), robbery but not assaults (Kane, 2006), and homicide and auto thefts only (Cloninger and Sartorius, 1979), and so forth. Overall, the literature on the arrests-crime rate relationship is mixed – some research finds an effect between arrests and crime rates, some does not (Huizinga and Henry, 2008).

The second issue concerns whether a law enforcement agency operates using a generalist or specialist policing style. We analyze data from a generalist municipal police department, meaning that patrol officers investigate low and mid-level crimes, but investigators and detectives are responsible for handling serious crimes, cases involving lengthy investigations (e.g. narcotics trafficking or white-collar crimes), and incidents in which patrol officers may need assistance (e.g. an incident involving numerous victims and witnesses, as well as cases involving extensive follow-up or knowledge on a specific subject, such as gangs or narcotics).

By contrast, when patrol officers in a specialist-style police department respond to any incident potentially requiring follow-up, they might only contact the victim and either hold the scene until an investigator arrives or forward the information to a detective for follow-up. It is possible, indeed likely, that macro-level variables might affect generalist patrol officers differently than specialized officers. Due to data limitations, however, our study focusses on generalist police officers.

Data and methods We gathered data on patrol officers and their patrol areas from several sources. First, we collected patrol officer productivity data from the Lincoln (Nebraska) Police Department’s 2013-2014 productivity reports. Using Computer-Aided Dispatch (CAD) and a centralized productivity database, LPD measures numerous patrol officer activities. Every month (and year), LPD generates productivity reports which display each patrol officer’s activities and outputs (see Table I for an example of a squad’s annual productivity report). Each patrol activity and output merits further explanation.

CFS and officer assist time When dispatchers send patrol officers on a call, they use CAD to track the minutes the officer spends on the call. If the call is a priority incident, dispatchers send both a primary and secondary officer. Other unassigned officers might also respond to the CFS, meaning that multiple officers could be secondary officers. The annual productivity record tracks the amount of time an officer spent as a primary responder and a secondary responder.

23

Patrol officer productivity

O ff ic er

no .

P ro se cu ti on

ra te

(% )

C F S

m in ut es

F /U P

m in ut es

O ff ic er

as si st

m in ut es

In ci de nt

re po rt s

A C I

re po rt s

Se le ct iv es

W ar ni ng

s O ff ic ia ls

W ar ra nt s

D U Is

M is de m ea no rs

F el on ie s

1 88

22 ,8 12

5, 20 8

6, 13 2

21 6

34 8

39 6

39 6

26 4

60 24

72 7

2 95

30 ,0 72

5, 42 4

4, 11 6

20 4

69 6

22 8

31 2

43 2

84 48

12 0

8 3

72 32 ,4 96

6, 31 2

6, 60 0

14 4

56 4

44 4

72 0

54 0

84 36

24 12

4 90

33 ,0 00

5, 23 2

4, 98 0

21 6

19 2

48 0

76 8

24 0

14 4

12 72

12 5

91 41 ,7 24

4, 82 4

6, 04 8

31 2

63 6

13 2

50 4

32 4

10 8

7 24

24 6

83 23 ,9 16

7, 05 6

6, 19 2

14 4

75 6

52 8

99 6

33 6

13 2

36 72

12 7

95 24 ,4 68

6, 84 0

5, 89 2

15 6

62 4

32 4

45 6

48 0

72 36

12 0

24 8

90 39 ,9 72

5, 19 6

4, 92 0

24 0

50 4

38 4

49 2

26 4

36 2

72 24

9 79

25 ,5 00

4, 06 8

5, 10 0

33 6

39 6

19 2

68 4

20 4

24 36

12 0

10 10

87 21 ,9 84

5, 41 2

4, 52 4

34 8

55 2

48 0

79 2

31 2

10 8

48 84

36

Table I. Example of the Lincoln Police Department’s annual productivity report

24

PIJPSM 39,1

Report writing time Standard operating procedures dictate that LPD patrol officers complete specific types of reports for incidents. An incident report (IR) details biographical information of the individuals involved in the incident and summarizes the incident. An Additional Case Information (ACI) is a more detailed report that documents additional investigative steps or the actions of assisting officers. Although the time spent on an IR or ACI could vary depending on the call for service, the average IR and ACI take roughly twenty minutes to complete[1].

The Lincoln Police Department also tracks self-initiated activities. Most productivity results from the following proactive activities.

Selectives The Lincoln Police Department tracks specific officer patrol activity by coding its neighborhoods, schools, parks, and business districts. When an officer patrols an at-risk business district or a high-crime neighborhood, for example, the officer notifies dispatch using that area’s specific code. Girded by the principles of “hot spots” policing research, officers use selectives to identify potential candidates for POP projects and heighten the perception of risk among criminals (Braga and Bond, 2008).

Traffic warnings and official citations Every Lincoln patrol officer drives a police cruiser and conducts traffic enforcement, issuing traffic warnings and official citations. The LPD records department tracks patrol officers’ warnings and citations.

Misdemeanors, felonies, and warrants Patrol officers also make misdemeanor and felony arrests during their productive time. Officers might detain a motorist and arrest him or her for driving under a suspended license or driving while intoxicated, or contact a pedestrian and make an arrest for a sex offender registration violation. Officers can also search their department’s electronic database for individuals with outstanding warrants in their patrol areas and attempt to arrest these offenders. As with warnings and citations, the LPD records department documents and tracks misdemeanor, felony, and warrant arrests. Moreover, the court system also monitors these types of outputs to process cases, thereby adding an additional layer of reporting accuracy and reliability.

Field information reports We also measured the number of field information or “intelligence” reports. Officers complete these internal reports to document suspicious or dangerous individuals or circumstances, though these reports are not publicly available.

VORC In addition to the raw outputs of traffic citations, warrants, misdemeanors, and felonies, we also include an advanced metric for measuring a patrol officer’s productivity. VORC is a method for evaluating officers that assigns different weights to productivity outputs and then assesses officer productivity in terms of available time for self-initiated activities, or productive time (Bonkiewicz, 2015). We use a variation of this formula that multiplies an officer’s officials, misdemeanors, and felonies by his or her

25

Patrol officer productivity

prosecution rate. Our modified use of VORC can best be explained using the following equation and five-step process:

VORC ¼ 100 � Η þ Θ þ Kð Þ þ Ιð Þ þ Λð Þ þ Νð Þð Þ � Ρð Þð Þ Α− Β þ Γ þ Δ þ 20 Ε þ Ζð Þð Þ þ 10 Mð Þð Þð Þ

� �

−100 � 0:9ð Þ � Ξð Þ Ο Π

� � Ξ

� �

where Α is the total monthly on-duty minutes, Β the CFS minutes, Γ the follow-up time and meetings, Δ the officer assist time, Ε the no. of IRs, Ζ the no. of ACIs, M the no. of intelligence reports, Η the no. of selectives, Θ the no. of warnings times two, Κ the no. of warrants times three, Ι the no. of officials times three, Λ the no. of misdemeanor arrests times their respective weights, Ν the no. of felonies types their respective weights, Ρ an officer’s prosecution rate, Ξ an officer’s productive time (Α−(Β+Γ+Δ+20(Ε+Ζ))), Ο, the department average P-score, or average (Η+((Θ)+(Ι)+(Κ)+(Λ)+(Μ)+(Ν)×Ρ)), and Π the department average self-initiated time.

First, VORC calculates an officer’s total minutes spent on CFS, follow-ups, officer assists, and reports. This amount is an officer’s total work time, or service minutes. Second, the formula subtracts an officer’s service minutes from the total on-duty minutes[2]. This is the officer’s amount of productive time. Third, VORC calculates an officer’s productivity score (or P-score) using a weighted index. Departments can choose different weights for productivity outputs, but for the purpose of this study, we used our weights based on the statutes governing the Lincoln Police Department. Citations and arrests are weighted based on whether the offense was a Traffic Offense, a Class III, II, or I Misdemeanor, or a Class V, IV, III, II, or I Felony. Accordingly, a patrol selective is worth one point, a traffic warning two points, a traffic citation three points, a warrant arrest three points, a Class III misdemeanor arrest four points, a Class II misdemeanor arrest five points, a Class I misdemeanor arrest six points, a Class V Felony seven points, a Class IV Felony eight points, a Class III Felony nine points, and Class II and I Felonies (e.g. homicides) ten points[3]. The sum of the weighted productivity outputs is the officer’s P-score. The P-score is divided by amount of productive time and then multiplied by 100 (for ease of understanding). This value, the productivity quotient, represents the officer’s productivity given the available minutes for proactivity, such as searching for wanted individuals or traffic enforcement.

Fourth, VORC creates a hypothetical replacement officer against which to measure the patrol officer’s output. This replacement player might be imagined as a rookie patrol officer, or a recruit who has not completed the academy, performing at 90 percent the capacity of fully trained police officer[4]. Using LPD’s data, VORC creates a hypothetical replacement officer by dividing the departmental average P-score by the departmental average for productive time, and then multiplying it by the officer’s productive time to determine what an average officer would have generated given the officer’s amount of productive time. This equation is then multiplied by 0.9 (the capacity of a replacement officer), divided by the officer’s productive time, and then multiplied by 100 to generate the replacement officer’s productivity quotient. Finally, VORC subtracts the replacement officer’s score from the patrol officer’s score to calculate the officer’s value over the replacement officer. This number, or VORC, indicates the officer’s productivity per time for self-initiated activities compared to the replacement officer’s productivity. For example, a VORC of 1.2 means than officer would produce 0.012 more productivity points (some combination of warnings, citations, and arrests) per minute than a replacement officer.

26

PIJPSM 39,1

Biographical variables We gathered biographical information about the officers in the sample, including gender, years of LPD experience, marital/cohabiting status, and number of children. We also controlled for whether the officer worked first, second, or third shift because our data showed significant differences in the number of CFS each shift handled, as well as the number of reports each shift completed.

Geographic variables It is necessary to measure the characteristics of patrol areas to account for variation across patrol areas. For example, some areas may feature different crime rates or levels of social disorganization which command more department resources and police officers. We gathered data by first mapping out every squad’s patrol area and then obtaining tract-level data from the 2010 US Census. Our models included measures of established of correlates of crime, including population density (Hipp and Roussell, 2013), household income (Gould et al., 2002), and racial composition (Light and Harris, 2012). Specifically, we use each area’s population density, median household income, and percent white. We also included measures for each area’s violent and property crime rate. Finally, we gathered information about the number of violent crimes (murders, assaults, aggravated assaults, sexual assaults, and robberies), property crimes (larcenies, burglaries, vandalisms, frauds, forgeries, and motor vehicle thefts), and number of police officers in each patrol area, thereby creating variables for violent crime per officer and property crime per officer.

Since LPD’s patrol officers are grouped with squads, it is reasonable to believe that an officer’s productivity might be more similar to an officer within their squad compared to an officer in a different squad. To account for the potential correlations between the individual units, as well as the multi-level nature of the data, we used hierarchical linear modeling to analyze the data.

Our primary level variables include an officer’s gender, marital status, number of children, years of law enforcement experience, shift, CFS, annual number of traffic stops and citations, field information reports, and misdemeanor, felony, and warrant arrests. We also used the officers’ productivity information to calculate each patrol officer’s VORC.

Our secondary level variables include the violent crime per officer rate and property crime per officer rate for the patrol area of each squad, as well several control variables, including population density, household median income, percent white, and the patrol area’s violent and property crime rate. Table II displays the summary statistics for the first and second-level variables.

Results We hypothesized that:

H1. The violent and property crime-to-cop ratios of an area might affect the raw productivity outputs of patrol officers, as well as a more advanced productivity metric.

To examine these questions, we constructed five hierarchical linear models estimating the correlations between reported violent and property crimes per officer and a patrol officer’s annual traffic citations, warrant, misdemeanor, felony arrests, and VORC. Although not presented here, we estimated null models for all dependent variables and found that significant variation existed between squads when we included no variables included in the models. The χ2 for the LR test for all five null models was statistically significant ( po0.000 in all models).

27

Patrol officer productivity

Table III shows that a rise in reported violent crimes per officer increases an officer’s number of warrants and misdemeanors (b ¼ 0.29 and 0.76), while arise in reported property crimes per officer decreases an officer’s number of warrants and misdemeanors (b ¼ −0.15 and −0.52). We found no relationship between the crime-to-cop ratios and felony arrests. However, both the rates of violent and property crimes per officer decreased the number of traffic citations an officer writes (b ¼ −1.68 and −2.25).

Table III also shows that traffic stops and field information reports are positively associated with warrant and misdemeanor arrests. This supports what police practitioners already suspect: the cops with the most arrests are extremely proactive patrol officers who stop a large number of vehicles and build investigations through the use of field contacts and information reports. For instance, an officer can build felony narcotics cases by contacting suspects and informants in the field, documenting these efforts through field information reports, and eventually using this information to write search warrants and conduct drug interdiction.

Next, we investigated whether this relationship between the crime-to-cop ratios and productivity outputs exists when using an advanced productivity metric, specifically, an officer’s VORC.

Table IV shows that there is no correlation between the ratio of property and violent crimes per officer and VORC. Although this stands in contrast to our earlier results, we would not expect to find a relationship between these variables and VORC. Advanced metrics such as VORC evaluate officers based on the amount of time for self-initiated activities (productive time) and not raw productivity outputs like misdemeanor or felony arrests. Even if an officer took an extremely large number of property crimes, thereby decreasing their productive time, VORC would only evaluate that officer based

Range SD Mean

(a) Second-level variables (patrol area) Population density 2,022-6,956 1,687 4,436 Household median income ($) 26,281-76,020 16,535 49,756 % white 73-93% 0.07 0.83 Violent crime rate (per 100,000) 800-4,800 1,203 2,521 Property crime rate (per 100,000) 3,100-9,800 2,027 6,524 Property crimes per officer 53.2-109.7 16.63 81.67 Violent crime per officer 19.4-37.2 5.59 29.29

(b) First-level variables (officer) Gender (female ¼ 0; male ¼ 1) 0-1 0.24 0.94 Married/Cohabiting 0-1 0.18 0.79 No. of Children 0-5 1.70 0.90 Years experience 1-37 7.79 10.63 Calls for service 481-1,011 161.21 613.65 Traffic stops 16-901 143.76 206.24 Field information reports 0-227 21.53 25.30 Traffic citations 24-1,246 146.32 186.42 Misdemeanor arrests 19-403 59.65 106.63 Felony arrests 0-71 10.66 8.52 VORC −1.95-10.43 1.51 0.24 Shift 1st

104 2nd 96

3rd 102

Notes: (a) n ¼ 40; (b) n ¼ 302

Table II. Summary statistics for variables

28

PIJPSM 39,1

on what the officer produced with the small amount of productive time. This suggests that while a team’s crime-to-cop ratio could affect an officer’s raw productivity outputs, i.e. arrests, a team’s crime-to-cop ratio would not affect an officer’s VORC.

For example, during a given month, an officer may take an unusually large number of burglaries or larcenies, thereby leaving the officer with only ten hours for self-initiated activities. With this reduced amount of productive time, the officer may generate a below-average number of officials, warrants, misdemeanors, and felonies, which in turn may lead a supervisor to potentially downgrade the officer’s productivity. However, VORC would recognize and account for the high number of CFS minutes and analyze the officer’s productivity based on what the officer generated with only ten hours of productive time. It may be that the officer was actually very productive, but they simply did not have enough productive time to generate the same number of traffic citations and arrests as an officer who responded to far fewer property crimes.

Table IV also shows that traffic stops and field intelligence reports are positively correlated with VORC. These results are consistent with the principles of how VORC evaluates patrol officer productivity and provides additional support for the validity of this advanced metric. VORC accounts for numerous variables that affect

Traffic citations Warrant arrests Misdemeanor

arrests Felony arrests b SE b SE b SE b SE

(a) Second-level variables (patrol area) Population density 0.01 0.03 0.01 0.01 −0.01 0.01 0.00 0.00 Household median income ($) 0.02 0.01 −0.00 0.00 0.00 0.00 −0.00 0.00 % white 133.63 324.72 −89.21 56.64 −32.74 192.11 −44.47 32.55 Violent crime rate (per 100,000) 0.06 0.04 0.01 0.01 −0.01 0.02 0.01 0.02 Property crime rate (per 100,000) −0.03 0.05 −0.02 0.01 0.02 0.03 −0.01 0.00 Property crimes per officer −2.25* 0.99 −0.15*** 0.03 −0.52* 0.24 −0.03 0.03 Violent crime per officer −1.68* 0.68 0.29* 0.14 0.76* 0.31 0.22 0.12

(b) First-level variables (Officer) Male 46.73** 16.22 3.66* 1.47 16.13 10.51 0.91 1.58 Married/cohabiting −75.16* 32.54 2.01 2.78 −11.58 11.69 −0.95 1.67 No. of children 1.83 7.02 0.75 0.77 −2.62 2.17 −3.96 2.07 Years experience −10.65* 4.92 −0.12 0.13 −1.63*** 0.33 −0.06 0.05 Calls for service 0.24 0.13 0.02 0.02 0.14*** 0.01 −0.08 0.05 Traffic stops – – 0.03*** 0.00 0.09** 0.02 0.00 0.00 Field information reports 2.09*** 0.33 0.20*** 0.04 0.79*** 0.19 0.10*** 0.03 Second shift (reference group: first) 26.03 17.78 1.87 1.95 8.09 7.30 1.47 1.60 Third shift (reference group: first) 98.72*** 21.55 0.79 2.13 25.32** 9.52 3.93** 1.32 Intercept 325.91* 53.61 7.58* 3.39 41.47* 19.12 3.10 4.84 Sigma-squared 15,414.67 192.87 2,067.02 55.98 χ2 82.94*** 61.33* 123.48*** 106.63*** Notes: (a) n ¼ 40; (b) n ¼ 302. All variables are uncentered. *po0.05; **po0.01; ***po0.001

Table III. Hierarchical linear models estimating

officers’ annual traffic citations,

warrants, misdemeanors, and

felonies

29

Patrol officer productivity

productive time (i.e. CFS minutes), meaning that the key factors which increase or decrease an officer’s VORC are largely related to self-initiated activities (i.e. traffic stops and information reports). VORC, therefore, accounts for the potential effects of macro-level variables and analyzes the activities of the officer during his or her productive time. Nonetheless, we recognize that because traffic stops and field information reports are correlated with VORC, an area’s crime-to-cop ratio might still affect an officer’s VORC indirectly by impacting the number of stops and information reports they generate.

Discussion There are several factors which affect a patrol officer’s productivity, including biographical characteristics, officers’ perceptions and attitudes, and organizational and operational features, among others. This paper identified two macro-level variables – property crimes per officer and violent crimes per officer – which affect the number of traffic citations, warrants, misdemeanors, and felonies generated annually by an officer. Using data from a mid-sized police department, we found that a patrol area’s rate of property crimes per officer was associated with a moderate decrease in an officer’s annual number of traffic citations, warrant arrests, and misdemeanor arrests, while a patrol area’s rate of violent crimes per officer was also associated with a moderate decrease in an officer’s annual number of traffic citations. We also found that a patrol area’s rate of violent crimes per officer is associated with a moderate increase in an officer’s annual number of warrant and misdemeanor arrests. However, when we used a more advanced productivity metric, we found no relationship between property and violent crimes per officer and an officer’s VORC.

b SE

(a) Second-level variables (patrol area) Population density −0.00 0.00 Household median income ($) 0.00 0.01 % white 3.18 3.45 Violent crime rate (per 100,000) −0.00 0.00 Property crime rate (per 100,000) 0.00 0.00 Property crimes per officer −0.02 0.04 Violent crime per officer 0.02 0.05

(b) First-level variables (officer) Male 0.27 0.15 Married/cohabiting −0.32 0.31 No. of children 0.05 0.03 Years experience 0.00 0.01 Calls for service 0.02 0.01 Traffic stops 0.01*** 0.00 Field information reports 0.02*** 0.00 Second shift (reference group: first) 0.23 0.12 Third shift (reference group: first) 0.55*** 0.14 Intercept 1.65*** 0.36 Sigma-squared 0.66 χ2 98.46***

Notes: (a) n ¼ 40; (b) n ¼ 302. All variables are uncentered; *po0.05; **po0.01; ***po0.001

Table IV. Hierarchical linear models estimating officers’ annual Value Over Replacement Cop (VORC)

30

PIJPSM 39,1

There are several factors which explain the differing effects of the crime-to-cop ratios on productivity outputs. Let us first address why the violent crime-to-cop ratio would increase a patrol officer’s warrants and misdemeanors while the property crime-to-cop ratio would decrease these outputs. Notably, half of all victims of violent crimes know the offender (Truman and Rand, 2010), which aids law enforcement in identifying and locating the perpetrator. Moreover, both suspect and victim are frequently on scene when police officers arrive, such as during domestic disturbances. Being able to identify and locate the perpetrator helps patrol officers make arrests, thereby increasing their misdemeanor and felony arrests in violent crimes. Additionally, when patrol officers arrive and contact individuals during an incident, it is standard procedure to check if the parties involved have outstanding warrants. This means that the greater number of violent crimes that officers respond to, the more chances they have to arrest individuals who might have outstanding warrants. It makes sense, then, that officers who respond to a greater number of violent crimes have more chances to make warrant and misdemeanor arrests.

Conversely, property crimes may supply the officer with fewer warrant and misdemeanor arrests because of low clearance rates. We examined the difference in clearance rates between property and violent crime using incident data from the Lincoln Police Department, and we found that the clearance rate for violent crimes was 41.8 percent compared to only 23.7 percent for property crimes. Moreover, property crimes often also require more investigative time than violent crimes. Property crime investigations often require considerable follow-up by officers, such as processing a scene, contacting financial institutions, and checking pawnshops for stolen property. In general, property crimes are more time-intensive than violent crimes, thereby decreasing productive time and opportunities for proactive activities, while simultaneously netting officers fewer arrests because of low clearance rates.

The impact of a team’s crime-to-cop ratio can also be seen in the relationship between property and violent crime per officer and the number of traffic citations. Table III shows that a rise in both reported violent and property crimes per officer decrease an officer’s number of traffic citations. Quite simply, if officers respond to a high number of property and violent crimes, then they will have fewer minutes to conduct traffic enforcement and other proactive enforcement.

As noted, it is possible that the correlation is bi-directional. If an officer makes a greater number of arrests, then a team may witness a decrease in crime because prolific criminals may have been arrested and placed in jail. For example, one might interpret Table III to mean that an increase in warrant and misdemeanor warrants led to a decrease in the number of property crimes per officer. However, under this hypothesis, we should also see that an increase in warrant and misdemeanor arrests leads to a decrease in the number of violent crimes per officer. Instead, we see the opposite hypothesized effect, i.e. more arrests leading to more reported violent crimes[5].

We believe our explanation has a stronger theoretical foundation: the ratio of property crimes per officer affects productivity outputs differently than the ratio of violent crimes per officer, for two reasons. First, violent crimes are different in nature than property crimes, especially with regard to clearance rates. Second, property crimes often require more investigative and follow-up time than violent crimes, thereby reducing an officer’s amount of productive time and proactive activity.

Table III also provides a key piece of evidence to support this assertion. Table III shows that the number of reported violent crimes per officer increases an officer’s number of warrant and misdemeanor arrests. Yet the relationship is the opposite for

31

Patrol officer productivity

traffic enforcement; the violent crime-to-cop ratio actually decreases traffic citations. This change is due to the fact that while violent crime investigations often yield warrant and misdemeanor arrests, they do not produce traffic citations. Furthermore, responding to a greater number of violent crimes decreases the amount of available time to conduct traffic enforcement, in turn reducing the number of traffic citations. In other words, the violent crime-to-cop ratio affects traffic stops the same way that the property crime-to-cop ratio affects all productivity outputs: by reducing an officer’s productive time and opportunities for proactivity. In fact, we found a moderate correlation between a patrol area’s crime-to-cop ratio and the amount of productive time enjoyed by officers who worked that area. The correlation between an area’s violent crime-to-cop ratio and productive time was −0.21, while the correlation between an area’s property crime-to-cop ratio and productive time was −0.33. We believe this evidence indicates that officers who work in areas with a high crime-to-cop ratio have less time for self-initiated activities, which drives down productivity.

We suggest that the number of police officers and crime rates of an area need to be analyzed in conjunction with each other. Prior theories of police productivity have viewed and examined these variables separately, when in reality, we should view these variables in tandem. Combining these variables into crime-to-cop ratios provides a more nuanced – and accurate – understanding of patrol activities and their relationship to productivity.

Although we found strong evidence that an officer’s patrol area affects his or her productivity (in terms of raw outputs), we also found that a few officer-level variables affected productivity as well, including traffic stops, intelligence reports, and working third shift. Our results appear to substantiate what many police practitioners already suspect: officers who are proactive in contacting citizens (traffic stops) and building investigations (intelligence reports) generate more warrant and misdemeanor arrests. In addition, while some newer police officers are forced to work third shift because they do not have sufficient seniority to bid for better days off and shift times, many officers work nights because they enjoy the types of calls (e.g. in-progress felonies), meaning they may be more motivated to stop vehicles, make field contacts, and in turn, engage in activities that generate arrests (Brehm and Gates, 1993).

These results have important implications for police researchers and administrators. First, our results suggest a need for a deeper understanding of how the staffing levels, crime rates, and the other characteristics of a patrol area intersect. Second, if departments evaluate their patrol officers using raw productivity outputs, such as misdemeanor or felony arrests, administrators must recognize that factors beyond the control of patrol officers might affect monthly or annual productivity. Our research suggests that a patrol area’s crime-to-cop ratio can substantially affect a patrol officer’s productivity, regardless of the actual crime rate. If the characteristics of a patrol area affect raw productivity outputs, police departments should employ more advanced productivity analytics (i.e. VORC) that account for activities which reduce productive time and evaluate an officer’s productivity in terms of available minutes for proactivity.

Nonetheless, evaluating both patrol officers and police departments using traffic citations, warrants, misdemeanors, and felonies is useful because this type of data are readily understood by citizens and civic officials. A neighborhood watch group, for instance, might care more about the number of arrests their officers make and less about those officers’ productivity scores. Our data suggests that if citizens want patrol officers to write more traffic citations and make more arrests, they should lobby for more police officers on the street. It is true that reducing an area’s crime rate impacts

32

PIJPSM 39,1

the crime-to-cop ratio, but our results indicate that putting police officers on the street may be the most direct way to affect the crime-to-cop ratio.

Our study is not without limitations. We only examined a police department which uses the generalist style of policing, and it is unknown if our results are applicable to police departments which use a specialist style of policing. Theoretically, a department’s policing styles can affect factors that increase or decrease a patrol officer’s productive time. It is possible that the number of property crimes and violent crimes per officer do not significantly affect a specialist patrol officer’s productivity outputs vs a generalist officer’s outputs. Future studies should investigate whether our results can be generalized to all police departments.

We have only identified two macro-level variables which affect patrol officer productivity, and we recognize there may be other variables. In addition, we recommend that future studies incorporate a wider array of variables and test competing models to determine whether micro or macro-level variables exhibit stronger effects on productivity. In the end, managing patrol officers and improving productivity requires a greater understanding of both patrol officers and the areas they work.

Notes 1. Officers type all reports on computers. We obtained this average by observing a sample of

Lincoln patrol officers completing IRs and ACIs, and we recognize that this average may differ from other police departments. Moreover, if departments require officers to complete their reports before becoming available for another call for service, this measurement is unnecessary.

2. On-duty time does not include training time, vacation or holiday days off, or sick leave. In other words, on-duty time is the amount of time a patrol officer spent working the street.

3. Our weighting system is only one example of the flexibility that VORC offers departments; each agency can prioritize and assign corresponding weights to their preferred productivity outputs.

4. Scholars and policy makers may disagree with estimating a replacement officer’s capacity at 90 percent. Our goal is not to definitively estimate a hypothetical officer’s productivity, but merely to establish a reasonable fixed point against which to compare all patrol officers.

5. We constructed regression models using citations, warrant arrests, misdemeanor arrests, and felony arrests from the preceding month to predict both violent and property crime rates for the months in our data set. We found no statistically significant relationships between citations, warrant arrests, misdemeanor arrests, and felony arrests and the violent and property crime rates of the beats. However, after disaggregating the violent and property crime rates into specific crimes, we did find that misdemeanor and felony arrests significantly predicted a decrease in robbery rates, which is consistent with the literature. Additionally, we interviewed department personnel and learned that the Lincoln Police Department had staffed a team designed specifically to apprehend robbery suspects during this timeframe, which seems to explain the arrest-robbery rate relationship.

References

Amaranto, E., Steinberg, J., Castellano, C. and Mitchell, R. (2003), “Police stress interventions”, Brief Treatment and Crisis Intervention, Vol. 3 No. 1, pp. 47-53.

Becker, G. (1985), “Human capital, effort, and the sexual division of labor”, The Journal of Labor Economics, Vol. 3 No. 1, pp. 33-58.

Bonkiewicz, L. (2015), “Bobbies and baseball players: evaluating patrol officer productivity using sabermetrics”, Police Quarterly, Vol. 18 No. 1, pp. 55-78.

33

Patrol officer productivity

Braga, A. and Bond, J. (2008), “Policing crime and disorder hot spots: a randomized controlled trial”, Criminology, Vol. 46 No. 3, pp. 577-607.

Brehm, J. and Gates, S. (1993), “Donut shops and speed traps: evaluating models of supervision on police behavior”, American Journal of Political Science, pp. 555-581.

Brown, J. and Campbell, E. (1990), “Sources of occupational stress in the police”, Work and Stress, Vol. 4 No. 4, pp. 305-318.

Chappell, A., MacDonald, J. and Manz, P. (2006), “The organizational determinants of police arrest decisions”, Crime and Delinquency, Vol. 52 No. 2, pp. 287-306.

Cloninger, D. and Sartorius, L. (1979), “Crime rates, clearance rates and enforcement effort: the case of Houston, Texas”, American Journal of Economics and Sociology, Vol. 38 No. 4, pp. 389-402.

Corman, H. and Mocan, N. (2005), “Carrots, sticks, and broken windows”, Journal of Law and Economics, Vol. 48 No. 1, pp. 235-266.

Crank, J.P. (1990), “The influence of environmental and organizational factors on police style in urban and rural environments”, Journal of Research in Crime and Delinquency, Vol. 27 No. 2, pp. 166-189.

Garicano, L. and Heaton, P. (2010), Information Technology, Organization, and Productivity in the Public Sector: Evidence from Police Departments, RAND Corporation, Santa Monica, CA.

Goodman, A. (1990), “A model for police officer burnout”, Journal of Business and Psychology, Vol. 5 No. 1, pp. 85-99.

Gould, E., Weinberg, B. and Mustard, D. (2002), “Crime rates and local labor market opportunities in the United States: 1979-1997”, Review of Economics and Statistics, Vol. 84 No. 1, pp. 45-61.

Hatry, H. (1976), “Approaches to productivity measurement and program evaluation”, Public Productivity Review, Vol. 1 No. 3, pp. 21-28.

Hawkins, H. (2001), “Police officer burnout: a partial replication of Maslach’s burnout inventory”, Police Quarterly, Vol. 4 No. 3, pp. 343-360.

Herbert, S. (1998), “Police subculture reconsidered”, Criminology, Vol. 36 No. 2, pp. 343-370.

Hipp, J. and Roussell, A. (2013), “Micro-and macro-environment population and the consequences for crime rates”, Social Forces, Vol. 92 No. 2, pp. 563-595.

Huizinga, D. and Henry, K.L. (2008), “The effect of arrest and justice system sanctions on subsequent behavior: findings from longitudinal and other studies”, The Long View of Crime: A Synthesis of Longitudinal Research, Springer, New York, NY, pp. 220-254.

Johnson, R.R. (2011), “Officer attitudes and management influences on police work productivity”, American Journal of Criminal Justice, Vol. 36 No. 4, pp. 293-306.

Kane, R. (2006), “On the limits of social control: structural deterrence and the policing of ‘suppressible’ crimes”, Justice Quarterly, Vol. 23 No. 2, pp. 186-213.

Lersch, K.M. (2002), “Are citizen complaints just another measure of officer productivity? An analysis of citizen complaints and officer activity measures”, Police Practice and Research, Vol. 3 No. 2, pp. 135-147.

Light, M. and Harris, C. (2012), “Race, space, and violence: exploring spatial dependence in structural covariates of white and black violent crime in US counties”, Journal of Quantitative Criminology, Vol. 28 No. 4, pp. 559-586.

Maxfield, M.G. (1982), “Service time, dispatch time, and demand for police services: helping more by serving less”, Public Administration Review, pp. 252-263.

Maxfield, M., Lewis, D. and Szoc, R. (1980), “Producing official crimes: verified crime reports as measures of police output”, Social Science Quarterly, pp. 221-236.

34

PIJPSM 39,1

Nakosteen, R. and Zimmer, M. (2001), “Spouse selection and earnings: evidence of marital sorting”, Economic Inquiry, Vol. 39 No. 2, pp. 201-2013.

Shane, J. (2010), “Organizational stressors and police performance”, Journal of Criminal Justice, Vol. 38 No. 4, pp. 807-818.

Shane, J. (2011), “Daily work experiences and police performance”, Police Practice and Research, Vol. 13 No. 3, pp. 1-19.

Sherman, L., Gartin, P. and Buerger, M. (1989), “Hot spots of predatory crime: routine activities and the criminology of place”, Criminology, Vol. 27 No. 1, pp. 27-56.

Truman, J. and Rand, M. (2010), “Bureau of Justice Statistics Bulletin, National Crime Victimization Survey, Criminal Victimization, 2009”, US Department of Justice Office of Justice Programs, Washington, DC.

Van Meter, D. (2001), Evaluating Dysfunctional Police Performance: A Zero-Based Approach, Charles C Thomas Publisher, Springfield, IL.

Vila, B. and Kenney, D. (2002), “Tired cops: the prevalence and potential consequences of police fatigue”, National Institute of Justice.

Vila, B., Morrison, G.B. and Kenney, D.J. (2002), “Improving shift schedule and work-hour policies and practices to increase police officer performance, health, and safety”, Police Quarterly, Vol. 5 No. 1, pp. 4-24.

Weisburd, D., Maher, L., Sherman, L., Buerger, M., Cohn, E. and Petrisino, A. (1992), “Contrasting crime general and crime specific theory: the case of hot spots of crime”, Advances in Criminological Theory, Vol. 4 No. 1, pp. 45-69.

Further reading Kubrin, C., Messner, S., Deane, G., McGeever, K. and Stucky, T. (2010), “Proactive policing and

robbery rates across US cities”, Criminology, Vol. 48 No. 1, pp. 57-97. Vollaard, B. and Koning, P. (2009), “The effect of police on crime, disorder and victim precaution.

Evidence from a Dutch victimization survey”, International Review of Law and Economics, Vol. 29 No. 4, pp. 336-348.

About the author Luke Bonkiewicz is a Patrol Officer with the Lincoln Police Department. He has previously published research on police productivity and law enforcement response in disasters. Luke Bonkiewicz can be contacted at: [email protected]

For instructions on how to order reprints of this article, please visit our website: www.emeraldgrouppublishing.com/licensing/reprints.htm Or contact us for further details: [email protected]

35

Patrol officer productivity

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.