1) Is it an effective use of police resources to provide a rapid response to all citizen calls for service? Why or why not? In answering the question make sure you also concern yourself with what it is about the nature of citizen calls, citizen activitie

Citation: 2 Criminology & Pub. Pol'y 97 2002-2003

Content downloaded/printed from HeinOnline (http://heinonline.org) Sat Jun 27 05:53:56 2015

-- Your use of this HeinOnline PDF indicates your acceptance of HeinOnline's Terms and Conditions of the license agreement available at http://heinonline.org/HOL/License

-- The search text of this PDF is generated from uncorrected OCR text.

-- To obtain permission to use this article beyond the scope of your HeinOnline license, please use:

https://www.copyright.com/ccc/basicSearch.do? &operation=go&searchType=0 &lastSearch=simple&all=on&titleOrStdNo=1538-6473

MANAGING CITIZEN CALLS TO THE POLICE: THE IMPACT OF BALTIMORE'S 3-1-1 CALL SYSTEM*

LORRAINE MAZEROLLE Griffith University

DENNIS ROGAN Statistical Analysis for Law Enforcement Strategies

JAMES FRANK University of Cincinnati

CHRISTINE FAMEGA California State University-San Bernardino

JOHN E. ECK University of Cincinnati

Research Summary: Our paper explores the impact of implementing a nonemergency 3-1 -

1 call system in Baltimore, Maryland. We found a large (34.2%) reduc- tion in 9-1-1 calls following the introduction of the 3-1-1 nonemergency call system. Many, but not all, of these calls simply migrated over to the 3-1-1 call system. Overall, we identified a 7.7% reduction in recorded citizen calls to the police post 3-1-1 intervention. This recorded reduc- tion in citizen calls was confounded by an increase in high priority calls to the 9-1-1 system (275%), a large overall reduction in low priority calls (54.3%), and an estimated increase (perhaps 8%) in unrecorded calls to the police. We also note a small increase in response times to high priority 9-1-1 calls following the implementation of the 3-1-1 call system and virtually no change in the amount of officer time available for community policing or problem-oriented policing activities.

* This study was sponsored by the National Institute of Justice through an award (Grant 98-IJ-CS-0067) to the University of Cincinnati with funds transferred from the Office of Community Oriented Policing Services. Findings, and the conclusions of the research reported here are those of the authors and do not necessarily reflect the official position or policies of the U.S. Department Justice. The authors acknowledge support from the Baltimore Police Department and the many research assistants who worked on the various components of this research project. Address all correspondence to Lorraine Mazerolle, School of Criminology and Criminal Justice, Griffith University (Mt Gravatt Campus), Brisbane, Australia 4111.

NUMBER 1 2002VOLUME 2 PP 97-124

MAZEROLLE ET AL.

Policy Implications: Our findings suggest that nonemergency call systems, such as 3-1-1,

can greatly facilitate police efforts to better handle citizen calls for police service. However, the intrinsic value of nonemergency call sys- tems is tightly woven with a police department's willingness to change dispatch policies (especially for those calls received via the 3-1-1 sys- tem), reallocate patrol resources, and adopt organizational reforms to support alternative methods (apart from dispatch) for handling non- emergency calls for service.

KEYWORDS: Police Workloads, Non-emergency Calls, Emergency Calls, Community Policing, Police Management, Police Organizations

"Calling the cops" using the emergency 9-1-1 number is what Bayley (1998) describes as the cornerstone of policing in modern democratic countries (see also Sparrow et al., 1990). Any citizen from any city, suburb, or town across the United States can mobilize police resources by simply picking up the phone and placing a direct call to the police. To a citizen of the United States this may seem a trivial entitlement; yet to millions of people from less democratized countries, the ability of a private citizen to call, expect, and receive police services by simply dialing 9-1-1 is seen as an outstanding privilege.'

The national emergency number, 9-1-1, was sold originally to the U.S. public as a method for getting police, fire, and medical personnel to emer- gencies fast, thereby improving services to people in need of help. By the early 1980s, however, numerous deficits with the 9-1-1 call system became clear: An overwhelming number of calls to 9-1-1 requested nonemergency services (national estimates range from 40% to 80% percent); 9-1-1 was not a magic pill that could reduce crime or increase arrests (Sherman, 1997; Spelman and Brown, 1981); citizens expressed frustration with police handling of 9-1-1 calls (Spelman and Brown, 1981); and the police consist- ently complained about the demands placed on them by the 9-1-1 call sys- tem (Bayley, 1998). Indeed, Malcolm Sparrow and his Harvard colleagues (1990) identified the late 1980s as a period when "thoughtful police execu- tives [began] to doubt the wisdom of policing's marriage to 9-1-1" (P. 105). Indeed, one participant at Harvard's Executive Session on Community Policing in 1985 declared to Sparrow and his colleagues that "we have cre- ated a monster" (Sparrow et al., 1990:105).

1. Although the 9-1-1 system is designed to handle emergency medical, fire, and police calls, the overwhelming number of callers request police services. For example, in Baltimore, Maryland about 70% of the 1.7 million 9-1-1 calls for service per year are directed to the police.

3-1-1 CALL SYSTEM

In July 1996, President Clinton called for a national community policing number to help alleviate the abundance of nonemergency calls flooding the 9-1-1 emergency system (July 23, 1996, Sacramento, California). The U.S. Department of Justice, Office of the Community-Oriented Policing Services responded by submitting a request to the Federal Communica- tions Commission (FCC) that a 3 digit phone number be reserved for non- emergency use. By 1997, the FCC had approved this request. At this time, many police departments across the United States of America were in the process of reviewing or implementing technological approaches, as opposed to management approaches, to relieve the overburdened emer- gency 9-1-1 systems. Some cities (e.g., Dallas) opted for a 3-1-1 call system that integrated city call centers to better handle nonemergency calls for a wide range of city service problems (including nonemergency police mat- ters). Some police agencies (e.g., Buffalo) chose to simply increase the advertising of a departmental nonemergency phone number. Baltimore, by contrast, implemented a police managed and operated 3-1-1 call system that changed the call routing and management system of citizen calls for police service.

Our paper explores the impact of implementing a nonemergency 3-1-1 call system in Baltimore, Maryland. In particular, we explore whether the technological call handling system (ie 3-1-1) reduces the number of emer- gency 9-1-1 calls for police service. We also examine whether the 3-1-1 call system helped the police to better manage citizen calls for service, free-up officer time, and improve response times for high priority emergency calls. We begin the paper with a review of past research on police handling of citizen calls for service. We then identify the characteristics of the City of Baltimore and Baltimore's calls for service system, and we describe our research methodologies. Next, we present our research findings and then conclude with a discussion of the strengths and weaknesses of the 3-1-1 call system and the policy implications of our research.

PAST RESEARCH

Emergency 9-1-1 call systems comprise the single most important tech- nological innovation that has shaped and defined police practices over the last three decades. The 9-1-1 emergency call system is a call handling sys- tems that was originally sold as a technological solution for reducing response times and increasing police effectiveness. The proliferation of 9- 1-1 call systems revolutionized the manner in which police handled citizen demands for police service and underpinned demand management during the reform era of policing (see Kelling and Moore, 1988).

In the early days, the 9-1-1 emergency call system was hailed as a huge success (Sparrow et al., 1990). By the late 1970s, however, police officials

MAZEROLLE ET AL.

had become concerned with the volume of calls their officers were han- dling. This concern was not universal, but neither was it isolated. In a pair of influential reports, which were funded by the Law Enforcement Assis- tance Administration, 2 analysts identified problems posed by the increas- ing number of calls per officer coming in over police phone lines (Gay et al., 1977). These analysts proposed a variety of management strategies for coping with these problems, including patrol shift scheduling by call vol- ume rather than equal staffing around the clock. The wisdom at this time was that increases in 9-1-1 calls could be better handled by reallocating existing resources to more effectively manage officer workloads.

Subsequent studies in Kansas City (MO) (Kansas City Police Depart- ment, 1977), Peoria (IL), Rochester (NY), Jacksonville (FL), and San Diego (CA) (Spelman and Brown, 1981) demonstrated that two human elements impeded the technology of 9-1-1. First, most crimes are discov- ered long after the offender has left the scene of the crime. Second, even when offenders have contact with victims, victims typically take several minutes to decide to call the police once the offender leaves. This time lapse, during which time offenders try to escape, was found to be far more important than the seconds saved by having 9-1-1 available (Spelman and Brown, 1981). In short, scholars started to show that 9-1-1 was useful in too few cases for it to have a substantial impact on public safety from crime.

In 1977, the LEAA funded the Police Executive Research Forum (PERF) to examine alternative ways for the police to handle citizen calls for service. The resulting PERF report described how nonemergency calls could be shifted away from requiring an immediate patrol response. The alternatives PERF considered were delaying dispatches until officers were free to respond, taking reports over the phone, asking that callers mail in reports, or asking callers to come to a police station to file reports (Farmer, 1981). Based on this and other research, the National Institute of Justice (NIJ) developed and field tested the nationwide Managing Patrol Operations programs, consisting of regional seminars, manuals, and other materials (Cawley and Miron, 1977).

In addition to model programs, the NIJ also sponsored evaluations of call handling strategies throughout the 1980s. The Wilmington Police Department was the site for two evaluations. The first examined the utility of splitting the patrol service into two groups-one to handle calls and the other to proactively suppress crime (Tien et al., 1977). The second experi- ment looked at the impact of various call-management strategies designed

2. The Law Enforcement and Assistance Administration (LEAA) was the fore- runner of the National Institute of Justice, the Bureau of Justice Statistics, the Bureau of Justice Assistance, and the Office of Justice Programs.

3-1-1 CALL SYSTEM

to free up officer time for working on crime problems (Cahn and Tien, 1981). Another set of field trials that assessed alternative call handling strategies was conducted in Garden Grove (CA) and Toledo (OH) (McEwen et al., 1986). A third line of inquiry, more popular in Canada than in the United States, was to explore ways to wean the public away from placing nonemergency calls to the emergency call system. In Edmonton (Alberta), for example, a major public information campaign was launched to have citizens report minor thefts, noninjury accidents, and other problems directly to local police substations established throughout the city (Hawkins, 1996). Collectively, these studies established that the public was accepting of delays in responding to calls and phone reporting of nonemergencies, if police call takers clearly described how the call would be handled and did not imply officers would soon arrive.

The national emergency number, 9-1-1, remained in the background of these studies and public information campaigns. At first, there were many urban and suburban police agencies that did not have 9-1-1. However, as 9-1-1 became increasingly universal, the growing problem of call satura- tion became synonymous with the proliferation of 9-1-1.

Beginning in the mid-1980s, as policing increasingly undertook reform programs to implement community and problem-oriented policing strate- gies, officials found themselves confronting a common complaint from their officers: "We are too busy handling 9-1-1 calls to address the problems that give rise to these calls." This was particularly the case on busy evenings. Some of the officer's concerns may have been more percep- tual than real, as audits of time availability usually uncovered more discre- tionary time than officers claimed (Eck and Spelman, 1987; Skolnick and Bayley, 1986). One thing had become clear, however: Even if officers had time, the seemingly random nature of calls gave officers a sense of chaos and the perception that they could not accomplish preventive work. As community and problem-oriented policing require officers to engage in self-directed activities, these perceptions had to be addressed.

It became accepted wisdom within policing that to undertake commu- nity and problem-oriented policing, police managers would have to address the volume of calls. At least two widely read published books made this point: The New Blue Line (Skolnick and Bayley, 1986) and Beyond 9-1-1 (Sparrow et al., 1990). By 1996, the problem had become so well known within policing that it spilled out into the popular press with a cover story in U.S. News and World Report on the "tyranny of 9-1-1" (Witkin and Guttman, 1996).

In summary, nonemergency calls to the police had been a major prob- lem for local police for over 20 years. Police had attempted three strategies to address the problem. The first was to reallocate internal resources to

MAZEROLLE ET AL.

equalize officer workloads and free up time for proactive work. The sec- ond approach was to divert calls placed to the police so they did not imme- diately go to officers and so that some other calls could be handled without a patrol response. The third approach was to wean the public from using the telephone to report nonemergency concerns. Taking them in reverse order, these call management strategies sought to (1) keep calls from coming in, (2) separate calls by their need for quick response and assign them to appropriate services, and (3) adjust patrol resources to han- dle more calls with the resources available.

Police departments collectively spend billions of dollars each year man- aging and upgrading their Computer Aided Dispatch (CAD) systems. Much time, effort, and resources has been channelled into developing enhanced 9-1-1 systems that have made call routing procedures even more efficient than in the past. Moreover, most police agencies spend considera- ble energies appropriating resources to improve CAD-vehicle communica- tions and enhance their ability to better manage and respond to citizen calls for police service. Most jurisdictions have also grappled with the problem of overburdened emergency call systems and implemented a vari- ety of management approaches for handling the burden.

During the mid-1990s, many police agencies throughout the United States began looking for technological options, as opposed to management approaches, to solve their overburdened 9-1-1 call system problems. One family of technological solutions was nonemergency call systems that routed nonemergency calls away from the 9-1-1 system. The goals of non- emergency call systems typically were to relieve 9-1-1 call systems of non- emergency calls, to improve the management of all citizen calls to the police (both emergency and nonemergency), to improve response times for high priority calls, to increase citizen satisfaction with the police han- dling of calls for service, and to free-up patrol officer time to provide more opportunities for problem-oriented and community-policing activities.

In early 1998, the National Institute of Justice identified four cities that represented a cross section of jurisdictions leading the charge to imple- ment technological solutions for dealing with nonemergency requests for police service (Baltimore, Maryland; Buffalo, New York; Dallas, Texas; and Phoenix, Arizona). These cities agreed to participate in a comprehen- sive assessment of their nonemergency number systems. The University of Cincinnati was subsequently awarded a competitive grant to explore the processes these four cities adopted for implementing alternative methods for dealing with nonemergency citizen calls for police service. The evalua- tion also sought to determine the impact of implementing alternative methods for handling nonemergency citizen calls for police service on the quality and quantity of policing. The remainder of this paper draws on the results from the Baltimore site to assess the impact of implementing a 3-1-

3-1-1 CALL SYSTEM

1 nonemergency call system on police practice. 3

BALTIMORE AS THE RESEARCH SITE

At the time of our study, The City of Baltimore has a population of 716,446 people covering 86 square miles. The police department comprises nearly 3,000 sworn officers that handle about 1.4 million calls for police service each year. The existing 3-1-1 system was implemented on October 1, 1996 by the Baltimore City Police Department and AT&T funded in part by a 1996 grant by the Office of Community Oriented Policing. AT&T was replaced by Bell Atlantic as the service provider on December 17, 1998.4 This change resulted from cost considerations to the city, but left the system virtually unchanged.

Baltimore's 3-1-1 nonemergency call system comprises nine telephone lines and uses a Nortel DMS 100 Intelligent Call Processing distribution switch. The City estimates the 3-1-1 system cost approximately $1.3 million to implement, including the cost of public education campaigns, hardware, software, and training. Incoming 3-1-1 calls are handled by limited duty sworn officers who undergo a one-day orientation, three-day training pro- gram, and two weeks on-the-job training.

At the time the evaluation was conducted, 3-1-1 calls entered the system as nonemergency calls and thus were not afforded the security that 9-1-1 calls received. These 3-1-1 calls were not assigned ANI/ALI5 and arrived "blind" or without a caller-identification number. Upon answering 3-1-1 calls, call takers identified themselves as nonemergency call operators and asked how they could be of service. Depending on the caller's problem, a number of options were designated to resolve the call. First, in a case where the caller was reporting a life-threatening emergency or a crime in progress, the 3-1-1 call takers (like their 9-1-1 counterparts) sent the call directly to dispatch. In the case of 3-1-1, however, the call taker immedi- ately attempted to ascertain the telephone number, name, and location of the calling party while simultaneously transferring the call through a single button transfer to the appropriate dispatcher. The information gleaned by the call taker was entered into the CAD system, it received a CAD num- ber, and the CAD information was transferred directly to the dispatcher. CAD automatically validated the address.

3. Our examination of the Dallas, Phoenix, and Buffalo nonemergency call sys- tems revealed unique system shortcomings and challenges that subsequently led to our decision to reduce our impact analysis efforts in these three cities (for a full description and discussion of these issues, please refer to Mazerolle et al., 2001).

4. Bell Atlantic has subsequently merged with GTE and now trades as "Verizon." 5. Automatic Number Identification (ANI) and Automatic Location Identifica-

tion (ALl).

MAZEROLLE ET AL.

Second, if the calling party desired to fill out a police report for a crime that did not require dispatch, the call taker completed the report over the phone through the CAD system. The report was given a CAD number, and the information became a permanent record in CAD. The report was also available for review and analysis through the CAD system to the Dis- trict in which the offense occurred. Third, if the calling party reported a more general neighborhood problem, a description of the problem was entered into a Lotus Notes database and e-mailed or faxed to a Neighbor- hood Service Center (NSC). A paper trail of accountability was estab- lished among the district, the sector commander, the NSC, and the citizen's report (see Mazerolle et al., 2001). Fourth, for calls that did not require police, fire, or ambulance response, the call takers provided the caller with the number of the appropriate city agency or service (in Balti- more, these calls are called type "79" calls). These "type 79" calls were generally not' recorded in the CAD system.

The Baltimore nonemergency call-taking system is now somewhat dif- ferent than the original system implemented in 1996. The biggest change in the Baltimore system has been the implementation of a voice recogni- tion system. Today callers to the 3-1-1 call system in Baltimore select "1" for a police emergency, "2" for a police concern that is not an emergency, or "3" if they want to contact a non-police city service. This voice recogni- tion system is a technological enhancement to the 3-1-1 call system that seeks to reduce further the number of calls made to and handled by the police.

DATA AND METHODS

The primary goals of nonemergency call systems are to relieve 9-1-1 call systems of nonemergency calls, to improve the management of all citizen calls to the police (both emergency and nonemergency), to improve response times for high priority calls, to increase citizen satisfaction with the police handling of calls for service, and to free-up patrol officer time to provide more opportunities for problem-oriented and community-policing activities. Our research methods sought to assess the efficiency and effec- tiveness of the Baltimore 3-1-1 call handling system in achieving these goals.

Our main source of data was gathered from Baltimore's CAD system (October 1, 1994 through December 31, 1999) and included both 3-1-1 and 9-1-1 citizen calls for service. We also interviewed primary stakeholders; we conducted a survey of patrol officers, line supervisors, and sector man- agers; we conducted a telephone survey of callers to both the 3-1-1 and 9-

3-1-1 CALL SYSTEM

1-1 call systems; and we conducted a series of systematic observations dur- ing ride-alongs with patrol officers over a two-week study period (see Mazerolle et al., 2001 for a detailed discussion of our data and methods).

We examined a total of 3.4 million 9-1-1 calls for the pre-intervention period (October 1, 1994 through October 1, 1996) and the post-interven- tion period (October 2, 1996 through October 1, 1998) as well as 540,519 3- 1-1 calls entering the CAD system in the post-intervention period. We also examined the CAD data for what were labeled as "type 79" calls for ser- vice and were deemed non-police matters. To further explore the presence (or absence) of CAD records for these "type 79" calls, we collected data over a one-month period to document all calls (i.e., all type 79 calls) whether they were recorded in CAD or not. These type 79 calls were gen- erally calls that were not police, fire, or ambulance matters, and call takers referred the callers to the appropriate city agency without making any record of the call to the police.

We interviewed all Sector Managers (n = 29), asking them about their perceptions of 3-1-1, what types of data they review to decide what problems they have in their Sector (when and how much), how they iden- tify patrol officer discretionary time, how they typically use patrol officer discretionary time, how they interface with the Neighborhood Service Center Sergeant, where they have directed their patrol staff over the last week, and where they intended to assign their personnel in the forthcom- ing week. To tap police officers' perceptions of the 3-1-1 system, we devel- oped a self-administered police officer survey. A total of 386 respondents completed the survey for a response rate equalling 20%.

The sample for the citizen caller survey was drawn from the population of all 9-1-1 and 3-1-1 calls made to the Baltimore City Police Department between May 28, 1999 and June 28, 1999. This 32-day period represented a time frame in which a more extensive recording of 3-1-1 calls was made by call takers. 6 Our final sample consisted of 330 cases with 125 9-1-1 callers, 125 3-1-1 callers, and 80 3-1-1 LAN callers. Each case was randomly drawn from the population of 147,160 cases identified during our 32-day study period and then randomly assigned to a sequence and call order number.

Our observational study sought to assess the role and influence of 3-1-1 at the street-level, albeit in a limited way. In total, 96% (N = 241) of the scheduled observations were completed in accordance with the ride sched- ule (see Mazerolle et al., 2001).

6. During this period, all calls (including type 79 calls) received by the 3-1-1 sys- tem were documented providing a complete record of all calls. This represents a more extensive recording of 3-1-1 calls because informational calls requesting directions or other governmental services are generally not recorded within this system.

MAZEROLLE ET AL.

RESULTS

THE IMPACT OF 3-1-1 ON 9-1-1 CALLS

Our first step in analyzing the Baltimore CAD data was to explore the differences in 9-1-1 calls before the introduction of the 3-1-1 system com- pared with after the 3-1-1 intervention. We define the pre-intervention period as being from October 1, 1994 through October 1, 1996 (two years) and the post-intervention period as being from October 2, 1996 through October 1, 1998 (two years). We examine the differences in the absolute number of 9-1-1 calls, pre- to post-intervention. Table 1 reports the num- ber and percent change for 9-1-1 calls received by the Baltimore Police Department by time period (pre- and post-intervention) and by the prior- ity that the call was allocated.

TABLE 1. NUMBER AND PERCENT CHANGE FOR 9-1-1 CALLS RECEIVED BY TIME PERIOD (PRE- AND POST-INTERVENTION) BY CALL PRIORITY

Pre-intervention' Post-intervention 2 Percent change

Priority 1 417,728 470,263 +12.6 Priority 2 902,565 633,706 -29.8 Priority 3 415,133 177,967 -57.1 Priority 4 201,043 66,169 -67.1 Priority 5 111,500 375 -99.7 Total 2,047,969 1,348,480 -34.2

Pre-intervention period includes 730 days from October 1, 1994 through October 1, 1996, excluding February 29, 1996 (leap year). 2 Post-intervention period includes 730 days from October 2, 1996 through October 1, 1998.

As this table shows, there was a dramatic decline of about one-third (34.2%) in the total number of 9-1-1 citizen calls for police service received by the police following the introduction of the nonemergency call system in October 1996 (see also Maryland Sun, 1998; New York Times, 1997). As the police department had hoped and expected, the most dra- matic decline came from priority 5 (low priority) calls: from 111,500 calls during the pre-intervention period down to just 375 calls in the post-inter- vention period (99.7%). Apart from the priority 1 9-1-1 calls that exper- ienced a 12.6% increase (see below for more discussion on this issue), there were significant declines in the number of calls across all priority levels, and the declines got larger as the priority levels decreased.

Our time series analysis of these CAD data confirm these pre- and post- comparisons and show large and statistically significant declines in 9-1-1 calls for police service following the introduction of the 3-1-1 system (see

3-1-1 CALL SYSTEM

Mazerolle et al., 2001). In real terms, the Baltimore Police Department received about 5,000 fewer 9-1-1 calls per week, representing about a 25% decline in 9-1-1 calls for police service that could be directly attributed to the introduction of the 3-1-1call system.

THE IMPACT OF 3-1-1 ON ALL CALLS TO THE POLICE

The impact of the 3-1-1 system on the quantity of 9-1-1 calls received by the police clearly demonstrates that the system achieved one of its primary goals: to reduce the burden of calls on the 9-1-1 system. However, our analysis of the CAD data sought to explore whether, and to what extent, the calls previously made to the 9-1-1 simply migrated to the 3-1-1 call system. Table 2 depicts the number of calls broken down by 3-1-1 and 9-1- 1 call type pre and post the 3-1-1 intervention.

TABLE 2. THE NUMBER OF 9-1-1 AND 3-1-1 CALLS BY TIME PERIOD

Pre-Intervention Post Intervention Priority Pre 9-1-1 Only Post 9-1-1 Only Post 3-1-1 Only Post 3-1-1 + 9-1-1 1 417,728 470,263 62,534 532,797 2 902,565 633,706 184,931 818,637 3 415,133 177,967 138,722 316,689 4 201,043 66,169 103,878 170,047 5 111,500 375 50,454 50,829 Total 2,047,969 1,348,480 540,519 1,888,999

Table 2 shows a reduction in the total number of calls received following the introduction of 3-1-1 (from 2,047,969 calls before to 1,888,999 calls after) representing a 7.7% absolute decline in calls. Table 2 also reveals a large, absolute reduction in the total number of priority five calls received (from 111,500 to 50,929 representing a 54.3% reduction in priority five calls). Clearly our results show that a large proportion (about one in three) of priority 2, 3, 4, and 5 calls that used to be placed to 9-1-1 simply migrated over to 3-1-1.

Of major interest is the absolute increase in priority 1 calls following the introduction of 3-1-1. As Table 2 shows, there was a 27.5% increase (from 417,728 priority 1 calls before to 532,797 priority 1 calls after) in the total number of priority 1 calls received by the police department via 9-1-1 and 3-1-1 following the introduction of the 3-1-1 system. More than any other category of call, the priority 1 calls are a major drain on police resources, particularly when the response time and complexity of the call is taken into account. Hence, we examined more closely the apparent increase in

MAZEROLLE ET AL.

the priority 1 calls. Our data suggest (see below) that the increase in prior- ity 1 calls was most likely driven by an increase in reporting of several specific categories of serious crimes (particularly rape, robbery, and bur- glary). Moreover, our analysis of the weekly averages of priority 1 calls reveals that there was a trend increase in priority 1 calls that began several months prior to the introduction of 3-1-1 (see Mazerolle et al., 2001). We conclude that much of the observed increase in priority 1 calls was proba- bly spuriously related to the implementation of the 3-1-1 nonemergency call system.

At the other end of the call priority spectrum, our analysis of the CAD and "type 79" data reveals some important nuances of the impact of the 3- 1-1 call handling system on the pattern of citizen calls to the police for what the police believe to be non-police matters (i.e., nonspecific provi- sion of information, information requests of the police, reporting of other city agency matters, etc.). As mentioned earlier, large proportions of the "type 79" calls are not recorded in the CAD system. In order to analyze these cases, we systematically recorded all "type 79" calls for a one-month study period. Our analysis of these "type 79" calls shows that the police handle an average of about 538 "type 79" calls per day, of which less than one-third get recorded into CAD. That is, the police call takers receive the call; provide information, advice, or a referral to the caller; and then sim- ply hang up. Our estimates of police handling of "type 79" calls suggests that this category of calls increased by about 8% following the introduc- tion of the 3-1-1 call system, offsetting to some degree the reduction in priority five calls received and recorded in the CAD system. We suggest that although citizens called the police less following the introduction of the 3-1-1 call system, we suspect that 3-1-1 call operators were more will- ing than were their 9-1-1 call taker counterparts to treat non-police calls coming into the 3-1-1 system as "type 79" calls and not record them into the CAD system. Baltimore's new voice recognition system and telephone link directly to the city agency call center more effectively siphons these "type 79" calls away from the police. Indeed, police call takers no longer even need to handle the call and re-direct the caller, perhaps reducing the workload of the 3-1-1 call takers by as much as 500 calls per day.

THE IMPACT OF 3-1-1 ON DISPATCH PRACTICES

One of the principle factors that dictate patrol officer workloads is the nature of dispatch policies. The Baltimore Police Department made a cru- cial policy change at the same time that they adopted the 3-1-1 call han- dling system in that they decided to cease dispatching low priority (priority 5) calls for service. This dispatch policy change is a confounding, yet cru- cial factor that needs to be examined within the context of our impact analysis of the 3-1-1 system in Baltimore. Table 3 depicts the dispatched

3-1-1 CALL SYSTEM

proportions of 9-1-1 and 3-1-1 calls by time period (pre- and post-interven- tion) and by priority level.

TABLE 3. PERCENT OF 9-1-1, 3-1-1, AND TOTAL CALLS DISPATCHED BY TIME PERIOD (PRE- AND POST-INTERVENTION) AND BY CALL PRIORITY

Percent of Calls Dispatched Priority 9-1-1 - Pre 9-1-1 - Post 3-1-1 - Post 9-1-1 + 3-1-1 - Post 1 99.4 99.6 98.6 99.5 2 99.4 99.6 97.6 99.2 3 97.2 99.3 94.1 97.0 4 99.2 99.5 98.1 98.3 5 10.3 0.5 0.6 0.6 Total 81.1 79.7 77.8 78.9

Table 3 reveals some important findings. As expected, Table 3 shows that the Baltimore Police Department virtually stopped dispatching all pri- ority 5 calls following the implementation of the 3-1-1 call system regard- less of whether they were received by the 9-1-1 system or the 3-1-1 system 7 . The similar dispatch patterns observed for both the 3-1-1 and 9-1- 1 calls is because the police department utilized just one dispatch policy that governed both 3-1-1 and 9-1-1 calls alike. Table 3 also shows that dur- ing the two years before the introduction of 3-1-1, 81.1% of all 9-1-1 calls were dispatched: over 97% of the priority 1 through 4 calls were dis- patched and 10.3% of priority 5 calls were dispatched. During the two years after the introduction of 3-1-1, 78.9% of all calls were dispatched. This represents a marginal 2.7% decline in the total number of calls dis- patched to the patrol division comparing before to after implementation of the 3-1-1 call system.

THE IMPACT OF 3-1-1 ON THE TYPES OF CALLS HANDLED

One way to explore the qualitative changes in calls before to after the implementation of the 3-1-1 call system is to examine changes in the types of calls placed to the 9-1-1 and 3-1-1 call systems (see Mazerolle et al., 2001). Our analysis reveals that the introduction of the 3-1-1 system funda- mentally changed the patterns of citizen reporting of some crimes and dis- order incidents to the police. For example, before the introduction of 3-1- 1, the police received an average of nearly 700 calls per week (N = 677) for

7. We note, however, that recent changes to call handling practices in Baltimore suggest that more low priority calls are now being dispatched than in the two-year post- intervention period that we examined for our evaluation.

MAZEROLLE ET AL.

family disturbance problems via the 9-1-1 system. After the introduction of the 3-1-1 system, the police received nearly 200 fewer calls via the 9-1-1 system for family disturbance complaints. This represents a 27% decrease in 9-1-1 calls regarding family disturbances. Overall, the police received 90 fewer calls per week about family disturbances (9-1-1 + 3-1-1) after the introduction of 3-1-1. Citizen reporting of juvenile disturbances, parking, suspicious persons, auto accidents, and destruction of property followed similar declines as those demonstrated in reporting family disturbances.

Also intriguing was the change in citizen calling patterns regarding loud noise complaints: Before the introduction of 3-1-1, the Baltimore Police Department received about 266 calls for service per week about loud noises via 9-1-1. After the introduction of 3-1-1, the police only received about 34 calls per week about loud noises via the 9-1-1 system, represent- ing a 87% decline in the number of loud noise complaint calls to 9-1-1. Interestingly, however, the total number of loud noise complaints per week increased (from 266 before to 281 after) when we examine the sum of loud noise complaints to both the 9-1-1 and 3-1-1 systems. The majority (88%) of these loud noise calls were received by the 3-1-1 system. Citizen reporting of narcotics, motor vehicle theft, gambling, larceny, and aggra- vated assault followed similar patterns to those demonstrated in the reporting of loud noises.

Overall, following the introduction of the 3-1-1 nonemergency call sys- tem, there was a reduction in the total number of citizen calls from about 19,560 calls per week down to about 17,644 calls per week. However, it appears that the 3-1-1 system "adopted" about 30% of the calls that had previously been routed via the 9-1-1 system. As one would expect, some categories of complaints migrated in large numbers from the 9-1-1 system (e.g., larceny, parking, loud noise, destruction of property, gambling, and suspicious persons). In some cases, however, the introduction of the 3-1-1 system coincided with an absolute increase in citizen complaints for some categories of crime and disorder (e.g., loud noise complaints).

RESPONSE TIME ANALYSIS

The total numbers and types of calls to 3-1-1 and 9-1-1 reveal part of the story about the impact of the 3-1-1 call system. How the police handled calls (both 9-1-1 and 3-1-1) provides additional insight into the implemen- tation story of Baltimore's 3-1-1 call system. From the outset, we expected that the reduction in calls to the 9-1-1 call system would lead to faster response times, especially to the more serious, priority 1 calls for police service. Our analysis of the Baltimore CAD data included a detailed examination of the before to after comparisons of dispatch times, times to arrive on the scene, and times to clear/complete the call for priority 1 calls. In contrast to our expectations, we found that the high priority 9-1-1 calls

3-1-1 CALL SYSTEM

(priority 1 calls) were dispatched slightly slower in the post-intervention period compared with the pre-intervention period. We remind readers that following the introduction of the 3-1-1 call system, there was an increase in priority 1 calls. We suspect that this increase in priority 1 calls was the underlying factor that slowed police response to these calls. We note, how- ever, that the police were quicker at handling 9-1-1 priority I calls than they were at handling 3-1-1 priority 1 calls, perhaps reflecting the more serious types of priority 1 calls received by the 9-1-1 call center compared with the priority 1 calls received by the 3-1-1 call center (as one would expect).

Apart from slower dispatch times for priority 1 calls in the post-inter- vention period, our analysis reveals that the police handled most catego- ries of 9-1-1 calls (arrival and cleared) slightly faster after the implementation of 3-1-1 than before the introduction of 3-1-1. By contrast, the police were slower at handling 3-1-1 calls compared with 9-1-1 calls across all priority levels. In short, our analysis tends to suggest that there were subtle differences in the speed at which the police handled 3-1-1 and to 9-1-1 calls. We suspect that these subtle differences reflect more the distribution of seriousness of the call within each priority category than conscious officer decision making to respond slower to 3-1-1 than 9-1-1 received calls. That is, we know that officers rarely knew whether the call they were dispatched to was received on either 3-1-1 or 9-1-1 (see Mazer- olle et al., 2001). As such, it is likely that 3-1-1 calls were the least serious types of calls within each priority category of calls.

CITIZEN SATISFACTION WITH BALTIMORE'S CALL HANDLING SYSTEMS

One of the goals of the 3-1-1 call system in Baltimore was to increase citizen satisfaction of police handling of citizen calls for service. Our evalu- ation was not implemented until after the 3-1-1 system was operational; hence, we have no pre-intervention measures of citizen satisfaction with the police handling of calls for service. We did, however, conduct a small survey of 9-1-1 and 3-1-1 callers to ascertain their levels of satisfaction with the way the police handled their calls. Our results show that citizens had a favorable view of 3-1-1 services. Citizens generally agreed that 3-1-1 improved city services, improved police-community relations, they felt that the 3-1-1 call system should be used for nonemergency calls only, and they expected the new system to result in fewer nonemergency calls to 9-1- 1. Over 90% of respondents felt that call takers were both polite and help- ful and about three-quarters were either very or somewhat satisfied with the police and city's response to their call.

MAZEROLLE ET AL.

UPTIME AND DOWNTIME ANALYSIS

One of the goals of the 3-1-1 nonemergency call system was to increase the amount of uncommitted time for patrol officers. The expectation was that the 3-1-1 technology could be an important vehicle by which the police could implement problem-oriented and community-oriented polic- ing activities. In this section, we examine the issue of "down-time" for patrol officers.

Our patrol officer survey revealed that the majority of responding officers stated that they did not perceive 3-1-1 to have had an effect on either the quantity of calls to which they were dispatched on a typical shift or the nature of these calls. Indeed, the majority of patrol officers respond- ing to our survey perceived no change (on a typical shift) in the amount of discretionary time attributable to the implementation of 3-1-1.8

We also analyzed the CAD data to examine the number of patrol units available for call response in the pre-intervention to post-intervention period. We found that during the 730 days (two years) prior to the intro- duction of the 3-1-1 system, there was a daily average of 489.41 units responding to 9-1-1 calls for service. 9 In the 730 days (two years) following the implementation of the 3-1-1 system, we identified a daily average of 488.93 units responding to 3-1-1 or 9-1-1 calls. This represents a mere 0.1% decrease in the number of units responding to calls following the introduc- tion of the 3-1-1 system. Indeed, there was virtually no change in the num- ber of patrol units handling calls for service from before to after the introduction of the 3-1-1 call system. Further, our analysis of the patrol units responding to 9-1-1 and 3-1-1 calls shows that about 460 of the patrol units handled over 90% of all 3-1-1 and 9-1-1 dispatched calls for service both before and after the introduction of the 3-1-1 call system.

We also sought to calculate the total number of minutes that patrol units spent responding to calls for service ("uptime") before and after the implementation of the 3-1-1 call system. We found that patrol units spent 189.88 minutes per unit per shift responding to 9-1-1 calls prior to the

8. Our observational study suggests that officers rarely knew the origin of a call to which they were dispatched. That is, dispatchers did not inform patrol officers whether the call was a 9-1-1 call or a 3-1-1 call. As such, for the majority of calls, patrol officers were indifferent to whether the call was actually a 9-1-1 or 3-1-1 call.

9. We calculated the daily average number of units responding to calls for service by aggregating the number of unique "alpha" (or patrol) units per shift per day. If a unit did not respond to either a 9-1-1 or 3-1-1 call, they were not included in the analy- sis. If a unit "carried over" a call from one shift to another, we counted the unit to the originating shift. This was a crucial decision as we did not want to inflate the number of available units simply because a unit claimed several minutes of overtime running into a new shift. As such, we believe that our count of "units responding to calls" approxi- mates the real number of available units and is not inflated by nuances of overtime.

3-1-1 CALL SYSTEM

implementation of the 3-1-1 system, and they spent a total of 168.91 min- utes per unit per shift responding to all calls (3-1-1 + 9-1-1 calls combined) after the implementation of the 3-1-1 system. This represents an 11% decrease in the total time spent per unit responding to calls for service following the implementation of the 3-1-1 system. Of the 168.91 minutes spent dealing with calls post 3-1-1, patrol units spent two-thirds of this time responding to 9-1-1 calls (111 minutes) and one-third of this time responding to 3-1-1 calls. This represents a reduction of 41.5% in the num- ber of minutes per unit per shift responding to 9-1-1 calls from before to after the introduction of the 3-1-1 call system.

In sum, our analysis of the CAD data reveals that the police department maintained the pre-3-1-1 levels of patrol response units following the introduction of the 3-1-1 system, that the patrol units spent considerably less time responding to 9-1-1 calls, and overall they spent less time han- dling calls for service after the introduction of the 3-1-1 call system, even when the time spent on 3-1-1 calls was taken into account.

TIME AVAILABLE FOR PROBLEM-ORIENTED AND COMMUNITY POLICING ACTIVITIES

One way to ascertain whether the 3-1-1 call system created increased opportunities for patrol officers to engage in problem-oriented and com- munity-policing activities is to calculate true "blocks" of uptime and down- time available to patrol units during any one shift. By "blocks" of uptime, we mean blocks of calls that can be linked together by time to show that a patrol unit is "occupied" and thus not available for community-policing activities. By "blocks" of downtime, we mean substantial blocks of time (more than 30 minutes in duration) where patrol units are "uncommitted" to any type of recorded task.

The majority of the calls originating as 3-1-1 or 9-1-1 calls (if dispatched) were handled by what the Baltimore Police called "Alpha Units." About 488 alpha units were assigned to the nine police districts and they handled 98% of all 3-1-1 and 9-1-1 calls for service. As such, we examined the block sequencing patterns for these alpha patrol units and we used a 730- day period before the 3-1-1 implementation and a 730-day period follow- ing the implementation of the 3-1-1 system.10 We point out, from the out- set, that the average downtime was not calculated as the inverse of

10. N = 6,935,001 dispatched records were used in our analysis. The cases were sorted and ordered by unit responding and by time. Sequential cases were examined and categorized into "blocks." Simple "blocks" of uptime calls included, for example, three or four dispatched calls that started and finished in sequence with short time breaks (e.g., five minutes) between calls. These "run-on" calls were counted as one unique block of committed time (i.e. uptime). Calculation of uptime and downtime, however, was confounded by overlapping times and our need to estimate a shift start

MAZEROLLE ET AL.

committed time (or "uptime"). As such, one does not expect that increases in uptime will necessarily translate into a decrease in downtime. We also remind readers that short bursts of "downtime" (less than half an hour) were excluded from our analysis.

Our analysis reveals that during the pre-intervention period, patrol units spent on average 28.02 minutes in committed blocks of time and they had about 109 minutes (over an hour and a half) available to them in any one block of down time. During the post-intervention period, alpha units spent an average of slightly longer (29.28) minutes during any one committed time slot and they had slightly more minutes (112.73) in any one block of downtime. As expected, there were differences across shifts in the average minutes that alpha units had in committed (uptime) and uncommitted (downtime) blocks of time.

To estimate how the blocks of committed and uncommitted time played out in any one patrol shift, we used the daily average of patrol units during the before (489.41) and after (488.93) to estimate how many "slots" patrol units had during an average shift. II We estimate that before implementa- tion of the 3-1-1 system, patrol units had 2.62 downtime slots available of

and end time. To deal with these confounding problems, we calculated the "true" shift start time and used this time as our shift "starting" point. We truncated blocks of com- mitted time that ran into the next shift. We examined each and every sequencing of calls and developed programs to handle all combinations and permutations of complicated call sequences. A simple example of a call sequence is thus: A call is received, a unit is on route and then diverted to another call that was received after the initiating call, the unit responds to the second call first, leaves the clearance time open, handles the first call, and then simultaneously clears the first and second call. We identified dozens of call sequences, some involving just two calls, but some involving up to four or five "run- on" calls. Together these calls would equal a block of committed time. All categories of call sequences were included to calculate the number of minutes in an uptime block.

Calculating downtime was even more complicated than calculating the number of uptime minutes. The primary complicating factor in the Baltimore CAD data was the difficulties we encountered in assigning a start and end time to a shift. For example, the time stamp in the CAD data was not always accurate (e.g., off-line periods did not always end in correct times being assigned to cases; indeed, the CAD system in Balti- more periodically resets the time stamp to account for inaccuracies in the "time" fields). The police department also uses a series of shift start and end times to account for busy time and day-of-week periods. Finally, and most obviously, patrol units rarely start their shift with a call dispatch that coincides with their shift starting time. With these nuances in mind, we decided to restrict our calculation of downtime: We include the data for the alpha units only in our analysis, and we restricted the analysis to count only those blocks of uncommitted time that were equal to or greater than half an hour. This half- hour criterion allowed us to eliminate all short bursts of "downtime" that we believed were useless blocks of time for patrol units to engage in any type of meaningful prob- lem-oriented policing or community policing.

11. To calculate the number of downtime slots per unit shift for the before period, we multiplied the daily average of available patrol units (489.41) with the number of days in the analysis (730 days) = 357,269.3. We divided the total number of downtime

3-1-1 CALL SYSTEM

about 109 minutes in duration during an average shift (about 4.8 hours). In the post-intervention period, patrol units had about 2.70 downtime slots available of about 112 minutes in duration (about five hours). This repre- sents a very marginal increase in the number and duration of downtime slots available following implementation of the 3-1-1 call system.

Conversely, we estimated the average number and duration of uptime slots before and after 3-1-1 implementation. We found that there was an average of 5.09 uptime slots of about 28.02 minutes duration before 3-1-1 implementation (about 2.4 hours) and an average of 4.79 uptime slots of about 29.28 minutes duration in the post-intervention period (about 2.3 hours). This represents a marginal decrease in committed time from pre- to post-implementation.

We also calculated the ratio of uptime to downtime time slots to assess whether there was any difference in the before to after time periods. We estimate that before the 3-1-1 intervention, 66% of the time slots that we identified in the data were classified as committed time and 34% of the identified blocks of time were classified as downtime. Following the imple- mentation of the 3-1-1 call system, a slightly smaller proportion of a patrol shift was dedicated to a block of "uptime" (64%) and a slightly larger proportion of an average patrol shift was available as "downtime" (36%). This means that in the post-intervention period, at least one in three iden- tified blocks of time are downtime slots that will last for nearly two hours. These results reveal that following the implementation of the 3-1-1 system, patrol officers had slightly more blocks of downtime that were slightly longer in duration, suggesting that patrol units in Baltimore had slightly more time available to them to engage in problem-oriented policing and community-policing activities following the implementation of the 3-1-1 system.

We also examined the patterns of uptime and downtime by day of week and by shift. As expected, the "night" shift (shift 1 from midnight to 8am) showed the greatest amount of fluctuation in downtime between weekday nights and weekend nights (about 120 minutes of downtime during week- day nights and about 102 minutes of downtime on the weekend nights). The morning shift (shift 2; 8am to 4pm) both before and after 3-1-1 imple- mentation revealed a tremendous degree of stability with an average of about 112 minutes of downtime per shift (slightly less per shift in the post

cases (937,674) by the total number of available patrol units (357,269.3) to identify 2.62 blocks of downtime per shift for the before period. We repeated the analysis for the after period using the daily average of patrol units after (488.93) and the total number of downtime cases after (N = 962,219). The average number of uptime slots was calcu- lated using the same method: 1,818,896 uptime cases divided by 357,269.3 patrol units gives 5.09 uptime slots before and 1,708,449 divided by 356,918.9 patrol units gives 4.79 uptime slots after.

MAZEROLLE ET AL.

3-1-1 period) recorded on both weekdays and weekends. The afternoon shift (shift 3; 4pm to midnight) showed slightly more downtime midweek (about 108 minutes before and 104 minutes after 3-1-1 implementation) and less downtime over the weekends (about 101 minutes before and about 97 minutes of downtime after).

Like the downtime patterns, there were similarities in the patterns of uptime before and after implementation of the 3-1-1 call system. There was, however, even more consistency and predictability in the amount of uptime across the days of the week. For example, night shifts had an aver- age of 27 minutes of uptime per shift (slightly less, surprisingly, on week- end "nights," i.e., the early hours of the night time shift). The afternoon shift (shift 2) clearly had the greatest amount of uptime (about 32 minutes both before and after 3-1-1 implementation), peaking earlier in the week before 3-1-1 implementation (Mondays and Tuesdays) and fairly consist- ently around 33 minutes on all weekdays in the post-3-1-1 period. Of inter- est is that the weekends show the lowest scores both for the number of uptime as well as for the number of downtime minutes. This apparent par- adox can be explained as follows: First, we remind readers that the uptime and downtime estimates are not mirror images of one another, but calcu- lated independently of one another and reflecting "blocks" of uptime and downtime, not absolute numbers of minutes uncommitted or committed; second, the low scores for uptime and downtime on the weekends is most likely the result of two things: (1) They have less blocks of downtime on weekends because they have a consistent flow of calls to respond to on those nights and (2) patrol officers spend less time responding to individ- ual calls on what they perceive to be "busy nights" (i.e., weekends) because they expect to have a consistent flow of calls (which they probably have).

Overall, our analysis of the uptime and downtime patterns by shift and by day-of-week reveal very little influence of the 3-1-1 call system on time slots available for patrol officers to engage in problem-oriented and com- munity-policing activities. Indeed, the only real pattern that emerges from the day-of-week comparison is that the 3-1-1 call system might have equal- ized calls for service across the days of the week, changing somewhat the pattern in which citizens report crime. That is, it appears that citizens tend to report crime to the 3-1-1 system in a steadier manner than what they used to report to the 9-1-1 system.

DISCUSSION AND CONCLUSION

In general terms, nonemergency call systems aim to relieve 9-1-1 call systems of nonemergency calls, improve the management of all citizen calls to the police (both emergency and nonemergency), improve response times for high priority calls, increase citizen satisfaction with the police

3-1-1 CALL SYSTEM

handling of calls for service, and free-up patrol officer time to provide more opportunities for problem-oriented and community-policing activi- ties. Our analysis of the impact of the Baltimore 3-1-1 system reveals four main findings. First, the implementation of the 3-1-1 call system resulted in a large reduction in 9-1-1 calls (34.2%). Second, there was an absolute decline of 7.7% in the total number of calls (including 9-1-1 and 3-1-1 calls) recorded into the CAD system. This overall reduction in calls, how- ever, was confounded by a number of factors. There was a 27.5% increase in priority 1 calls, of which a substantial number of these calls were placed to the 3-1-1 call system. We speculate that at least some of these high pri- ority calls were placed to the 3-1-1 system because of the lack of ALI and ANI identifiers within the 3-1-1 system, thus, offering a degree of anonym- ity for a 3-1-1 caller that is not afforded to a 9-1-1 caller. Further con- founding the reduction in calls to the police was the large reduction in priority five calls. Although some of this reduction in priority 5 calls was offset by an estimated 8% increase in nonrecorded "type 79" calls, we can only offer some suggestions as to why there was such a large reduction in low-level, quality-of-life calls for police service following the introduction of the 3-1-1 system in Baltimore. One explanation is that there was a true reduction in quality-of-life problems and that the reduction in these calls was spuriously related to the implementation of the 3-1-1 system. Another explanation is that citizens redirected their calls from the police to non- police city agency call centers following the advertising campaign launched by the Baltimore Police Department when they introduced the 3-1-1 call system. Yet another explanation is that citizens became dissuaded to call the police regarding quality-of-life problems because the Baltimore Police Department adopted a no-dispatch policy for priority 5 calls.

Our analysis of the 3-1-1 call system in Baltimore was confounded by the accompanying policy change where priority 5 calls ceased being dis- patched. This no-dispatch policy for priority 5 calls was possibly a factor that dissuaded citizens to call the police about low-level neighborhood problems. Some scholars might argue that reducing the number of calls to the police is a step in the right direction in terms of improving police han- dling of emergency calls for service. Indeed, we note initiatives in Cana- dian policing where a core call management strategy has been to dissuade citizens from calling the police when they can call a local police station or report an incident in person (see Hawkins, 1996). Other scholars are more prone to argue the opposite: that one of the core principles of community and problem-oriented policing is encouraging citizen reporting of crime and neighborhood problems and enabling the police to depict a more accurate picture of the spatial distribution of crime and quality-of-life

MAZEROLLE ET AL.

problems. 12 Recent changes to the way the Baltimore police handle low priority calls tends to suggest that the Baltimore police department has reconsidered their no-dispatch policy for priority 5 calls. Their most recent dispatch policies tend to suggest that police departments should not unilat- erally dissuade citizens from reporting low level neighborhood problems and that the police need to be a little more responsive to citizens about how they have handled their call.

One way to improve citizen satisfaction with a no-dispatch policy to low priority calls is to systematically call back those callers who offer their return phone number and who want the police to get back to them about actions they have taken following their call to the police. We note, how- ever, that the citizens included in our survey of Baltimore callers to 311 and 911 did not reveal dissatisfaction with the police response to their 3-1- 1 call, even when the police did not dispatch a patrol unit to the call.

The third major finding of our study was that the absolute reduction in 9-1-1 calls, coupled with the overall reduction in calls to the Baltimore Police Department failed to reduce response times for priority 1 calls. Indeed, with the increase in priority 1 calls, our data revealed an increase in the time taken to dispatch and respond to priority 1 calls following the implementation of the 3-1-1 call system.

Finally, our research shows that implementation of the 3-1-1 call system resulted in marginal gains in uncommitted blocks of time experienced by "alpha" patrol units. The introduction of the 3-1-1 call system left patrol units with slightly more time slots and longer time slots available for prob- lem-solving and community-policing activities. But the marginal gains in uncommitted time observed in our data went unnoticed by the patrol officers responding to our survey.

Our analysis of the 3-1-1 call system in Baltimore leads us to ask two basic questions: First, what is the value of implementing a 3-1-1 call system and could a police department achieve the same results (i.e., reducing non- emergency calls to the 9-1-1 system) using other, perhaps less costly approaches to better handle their calls for police service? Second, how can a 3-1-1 system be better utilized as a technological tool to facilitate com- munity policing?

Our research suggests that the implementation of a 3-1-1 call system reduces the number of calls made to a police department. It is not clear, however, whether the observed reduction calls in Baltimore was a function of a successful marketing campaign to reduce calls to the police (especially the 9-1-1 system), the result of citizen dissatisfaction with the police

12. We point out that the 3-1-1 system does not collect ANI and ALl information, thus, hampering the police department's ability to map out crime and quality-of-life problems reported by citizens to the police via the 3-1-1 system.

3-1-1 CALL SYSTEM

department's no-dispatch policy for priority 5 calls, the result of citizens redirecting their calls to other city call centers, a true reduction in low- level crime problems, or an artifact of call recording practices. Further research is needed to disentangle these rival hypotheses.

Police departments need to think very carefully about whether they want an absolute reduction in calls. When police departments encourage citizens to call the police, they arguably have better information about the spatial distribution of crime and quality-of-life problems and thus a more accurate picture of the locations of ongoing problems. These data are cru- cial for effective scanning and analysis within the problem-oriented approach to reducing neighborhood problems. We note, however, that the 3-1-1 system adopted in Baltimore did not have ANI and ALI, thus, reducing the utility of these 3-1-1 call data for problem-solving purposes.

Our research suggests that 3-1-1 call systems have the potential to help police efforts to better manage citizen calls and more aptly determine appropriate police responses. Indeed, with the adoption of a 3-1-1 call sys- tem, citizens take on some of the responsibility for determining the type of police response required in police handling of their call. That is, by calling 3-1-1, citizens are stating to the police that their problem does not require immediate patrol response or perhaps any patrol response. In essence, this self-screening process embodied by 3-1-1 challenges the value of the uni- versal dispatch policy used by the Baltimore Police Department to handle citizen calls. Adoption of a 3-1-1 system should motivate police depart- ments to adopt dual dispatch policies that acknowledge and capitalize on citizen decisions to call either 9-1-1 or 3-1-1.

From a citizen perspective, 3-1-1 technology provides several distinct advantages over more low tech approaches (e.g., hotlines and easy-to- remember seven-digit phone numbers) for more effectively and efficiently sorting out calls for police service. First, 3-1-1 call systems are superior to decentralized phone number systems because the 3-1-1 number is a uni- versally easy to remember number (especially for out-of-town visitors), calls can be centrally tracked and analyzed, and the switch over to the 9-1- 1 system is seamless in emergency situations. A 3-1-1 call system also appears to be a superior call handling system than simply adding more switches, call takers and phone lines to the 9-1-1 call system because the 3- 1-1 system gives citizens an anonymous route to report crime incidents, it gives citizens some decision making power as to whether they want or require a patrol car to be dispatched, and it reinstates, to some degree, the purity of the 9-1-1 call system. The disadvantages, however, with 3-1-1 calls systems include the cost (about $1.3 million in Baltimore) and the potential lack of location identifiers attached to 3-1-1 calls that could be used to scan for neighborhood problems.

Our second question asks how police departments, if they were to

MAZEROLLE ET AL.

implement 3-1-1 technology, might better utilize the technology than what was evident in Baltimore. We recall that the Baltimore Police Department dispatched 3-1-1 and 9-1-1 calls in the same way, and they adopted a no- dispatch policy for priority 5 calls. The Baltimore Police department made no changes in the number of patrol units allocated to respond to calls. We suggest that the implementation of 3-1-1 technology, coupled with organi- zational reform and careful policy change, could greatly facilitate the adoption of community-policing and problem-oriented policing activities. Indeed, 3-1-1 technology, if adopted at the same time as a split-force model of policing (see Tien et al., 1977), could provide the technological infrastructure to better manage calls for service, more efficiently use scare patrol response, and more effectively engage in problem-oriented policing. This obviously would require police departments to adopt different dis- patch policies for calls made to 3-1-1 compared with calls made to the 9-1- 1 system.

The Baltimore Police Department got many things right: 3-1-1 is a uni- versal and easily recognizable number; citizens are, for the most part, using the 9-1-1 system less and using the 3-1-1 system to identify incidents when they do not expect an emergency response. Moreover, the policy decision to cease dispatching low priority calls is a precursor to utilizing the 3-1-1 system as a technological tool to further divide (or split) the patrol response into two parts: those officers that respond to 9-1-1 dis- patched calls and those officers that respond, in a problem-oriented polic- ing capacity, to the 3-1-1 calls.

There are a number of assumptions, however, that underlie the success- ful implementation of a 3-1-1 call system. First and foremost, police departments should make sure they adequately advertise their nonemer- gency call systems to increase the probability that citizens use 3-1-1 for nonemergencies and 9-1-1 for emergencies. In effect, the citizens are the ones deciding as to how they want the police to handle their call. If they are ill-informed and make poor calling decisions, then it reduces the tech- nological advantages of adopting a 3-1-1 call system. As such, marketing and effective citizen education on when and how to use the 311 system is as important to its effectiveness as the technology itself.

Second, the suggested dual policy model for handling calls for service relies on the skill base for effective problem-solving being organizationally allocated efficiently and effectively. Further, it assumes that police depart- ment personnel have effective problem-solving skills.

Third, effective implementation of 3-1-1 within a problem-oriented and community-policing model relies on the notion that 9-1-1 calls and 3-1-1 calls tend to derive roughly from similar places. If the spatial distribution of 9-1-1 and 3-1-1 calls were fundamentally different (and they seem not to

120

3-1-1 CALL SYSTEM

be; see Mazerolle et al, 2001), then our proposed model would bias prob- lem-oriented policing efforts toward reducing less intractable problems. However, because we know that, for the most part, the street blocks with 3-1-1 problems tend also to be the street blocks with 9-1-1 problems, then we can be reasonably sure that problem-solving will occur at both crime prone as well as disorderly places.

In sum, our research sought to uncover the impact of nonemergency number systems on the quality and quantity of policing. We used the Balti- more 3-1-1 system to assess the impact of nonemergency number systems on policing, and we relied principally on our results from Baltimore to generalize and speculate how an ideal-type system might look for handling and managing calls for police service, given the adoption of the technology that underpins the 3-1-1 nonemergency call system. Our findings suggest that nonemergency call systems, such as 3-1-1, can greatly facilitate police efforts to better handle citizen calls for police service. However, the intrin- sic value of nonemergency call systems is tightly woven with a police department's willingness to change dispatch policies (especially for those calls received via the 3-1-1 system), reallocate patrol resources, and adopt organizational reforms to support alternative methods (apart from dis- patch) for handling nonemergency calls for service. Our research has, per- haps, raised more questions than we have been able to answer. Most notably, our research was unable to disentangle the relative factors that appear to have discouraged citizens from calling the police about low-level neighborhood problems. An answer to this, and other remaining ques- tions, will greatly enhance our capacity to make informed policy decisions and justify the expenditures that 3-1-1 technology requires.

REFERENCES

American Bar Association 1973 The Urban Police Function. Chicago.

Bayley, David H. 1994 Police for the Future. New York: Oxford University Press.

Bayley, David H. 1998 What Works in Policing. New York: Oxford University Press.

Cahn, Michael F. and James M. Tien 1981 An Alternative Approach in Police Response: Wilmington Management

of Demand Program. Washington, D.C.: United States Government Printing Office.

Cawley, Donald F. and H. Jerome Miron 1977 Managing Patrol Operations Manual. Washington, D.C.: United States

Government Printing Office.

MAZEROLLE ET AL.

Eck, John E. and William Spelman 1987 Solving Problems: Problem-Oriented Policing In Newport News. Wash-

ington, D.C.: Police Executive Research Forum.

Farmer, Michael F. 1981 Differential Police Patrol Strategies. Washington, D.C.: Police Executive

Research Forum.

Gay, William G., Thomas H. Schell, and Stephen Schack 1977 Prescriptive Package: Improving Patrol Productivity Volume I Routine

Patrol. Washington, D.C.: Office of Technology Transfer, Law Enforce- ment Assistance Administration.

Hawkins, Carl W. Jr. 1996 Ready, Fire, Aim: A Look at Community Policing in Edmonton, Alberta,

Canada. In Themes and Variations in Community Policing: Case Studies in Community Policing. Washington, D.C.: Police Executive Research Forum.

Hermann, Peter 1998 Non-emergency line reduces calls to 9-1-1 Maryland Sun (October 2):1B.

Janofsky, Michael 1997 Baltimore Cites Success With Alternative to 9-1-1. New York Times

(October 2):A7.

Kansas City, Missouri Police Department 1977 Response Time Analysis. Kansas City, MO.

Kelling, George and Mark Moore 1988 The Evolving Strategy of Policing, Perspectives on Policing, November

1988; No. 4. National Institute of Justice (U.S. Department of Justice) and the Program in Criminal Justice Policy and Management, John F. Kennedy School of Government, Harvard University.

Mazerolle, Lorraine, Dennis Rogan, James Frank, Christine Famega, and John Eck 2001 Managing Citizen Calls to the Police: An Assessment of Non-Emergency

Call Systems. Final Report to the National Institute of Justice.

McEwen, Tom, Edward F. Connors, and Marcia I. Cohen 1986 Evaluation of the Differential Police Response Field Test. Washington,

D.C.: United States Government Printing Office.

Sherman, Lawrence 1997 Policing for Crime Prevention. In Lawrence Sherman, Denise Gottfred-

son, Doris MacKenzie, John Eck, Peter Reuter, and Shawn Bushway (eds.), Preventing Crime: What Works, What Doesn't and What's Promising-A Report to the Attorney General of the United States. Washington, D.C.: U.S. Department of Justice, Office of Justice Programs, NCJ 165366.

Skolnick, Jerome and David Bayley 1986 The New Blue Line. New York: The Free Press.

Sparrow, Malcolm K., Mark Moore, and David M. Kennedy 1990 Beyond 9-1-1: A New Era for Policing. New York: Basic Books.

3-1-1 CALL SYSTEM

Spelman, William and Dale K. Brown 1981 Calling the Police: Citizen Reporting of Serious Crime. Washington, D.C.:

Police Executive Research Forum.

Tien, James M., James W. Simon, and Richard C. Larson 1977 An Evaluation Report of an Alternative Approach to Police Patrol. The

Wilmington Split Force Experiment. Cambridge, Mass.: Public Systems Evaluation, Inc.

Witkin, Gordon and Monika Guttman 1996 This is 9-1-1 . . . please hold. U.S. News and World Report. (June):30-34,

36, 38.

Lorraine Green Mazerolle is a lecturer in the Department of Criminology and Crimi- nal Justice at Griffith University (Australia) and formerly an Associate Professor at the University of Cincinnati (United States). She received her Ph.D. from Rutgers Univer- sity, New Jersey, in 1993. During her ten-year tenure in the United States, she received numerous federal grants on topics such as problem-oriented policing, police technolo- gies (e.g., crime mapping, gunshot detection systems, 3-1-1 call systems), civil remedies, street-level drug enforcement, and policing public housing sites. She is the author of Policing Places with Drug Problems (Sage Publications) and a co-editor, with Jan Roehl, of Civil Remedies and Crime Prevention (Criminal Justice Press). She has writ- ten many scholarly articles on policing, drug law enforcement, displacement of crime., and crime prevention.

Dennis Rogan is the President of Statistical Analysis for Law Enforcement Strategies (SALES) and formerly a research fellow at the Crime Control Institute. He received his Ph.D. from the University of Maryland. He has worked on many policing experiments, including the Minneapolis Hot Spots Experiment, the Kansas City Crack House Raid Experiment, and the Gun Market Experiment in Kansas City. His research interests include drug enforcement, domestic violence, and the analysis of criminal career pat- terns of persons, places, and areas.

James Frank is an Associate Professor in the Division of Criminal Justice at the Uni- versity of Cincinnati. He holds a Ph.D. in Criminal Justice from Michigan State Univer- sity and a Law Degree. He has worked on many policing projects, most notably, police observational studies with the Cincinnati Police Department. He writes in the area of police officer behavior, police organizational reforms, and police technologies.

Christine Famega is an Assistant Professor at California State University-San Bernar- dino. She is a Ph.D. student in Criminal Justice at the University of Cincinnati and was formerly the Field Research Coordinator of the evaluation of nonemergency call sys- tems across four jurisdictions. Her research interests are in police behavior and the organizational context of proactive and reactive policing.

John Eck is a Professor in the Division of Criminal Justice of the University of Cin- cinnati. He received his Ph.D. from the University of Maryland in 1995 and his Masters in Public Policy from the University of Michigan in 1977. Dr. Eck has worked for the Police Executive Research Forum in Washington DC from 1977 through 1994 and for the Washington/Baltimore High Intensity Drug Trafficking Area from 1996 through 1998. He has written extensively on policing, crime patterns, and drug markets.

124 MAZEROLLE ET AL.