Zoom into the map to explore the areas of interest to you.
Appreciating the complexity of the 911 system in the US is key to understanding how the 911 and its related initiatives work.
The boundaries of the Public Safety Answering Points (PSAPs) on this additional map demonstrate the complex nature of how 911 is structured in the US. Zoom in to see the PSAP boundaries, along with their related cities, counties, and states. This PSAP map has been simplified to easily render and manipulate. To see these boundaries in greater detail, click here. (This map has been developed by GeoPlatform and is open data.) PSAPs are continually changing. At the time that this map was developed, nationwide there were a total of 5,597 PSAPs.
As 911 initiatives in the field are growing, so too is this map. If you’re aware of an initiative that is missing, have questions, or would like to offer any corrections, please contact us at email@example.com with a link to information about the initiative. Future iterations of this map will contain additional search features and the ability to download the file to examine or analyze the data in full. Check the site regularly for updates.
Related Empirical Research
Adapted from Neusteter et al. (2020). Understanding Police Enforcement: A Multicity 911 Analysis.
Summary of findings
|Response time||Bennett||2018||CAD data from 40 police departments from 2015-2016 was used to study response time. Researchers found that CAD and location data can be used to generate models of optimal coverage. However, decreasing response time in one priority category increased response time in other categories.|
|Int’l/Police actions||Blackstone, Buck, Hakim et al.||2007||This Manchester, UK, study modeled traffic patterns in automated alarm calls for service and response and explored ways to optimize behavior. After determining what “normal” looked like, the study treated alarms as a “disruption” and developed a model for ideal numbers of alarms per officer per shift depending on officer experience and location, including the possibility of false alarms.|
|Int’l/response time/police actions||Blake & Coupe||2001||In Manchester, UK, two-officer patrols were more successful in apprehending in-progress burglary suspects than single-officer patrols under specific circumstances because of faster response times. The study sampled 441 911 cases between July and December 1996 from an anonymized police force serving 2.6 million people.|
|Int’l/response time||Blanes i Vidal & Kirchmaier||2017||This study of the 2008–2014 internal records of the Greater Manchester Police found that for certain crimes in the UK, response time has a statistically significant effect on clearance, with a 10% faster response time leading to 4.7% increase of likelihood of clearance, overall. The effects of a faster response time were stronger for theft, less for violent crime.|
|Caller behavior/IPV-DV||Bonomi, Holt, Martin et al.||2006||The study explored frequent caller behavior among 448 Seattle, Washington, women who had been involved in intimate partner violence (IPV) and found that women were more likely to contact police if they experienced severe physical or psychological IPV, had injuries, or lived with children.|
|Int’l/Alternatives to police||Bouveng,, Bengtsson, & Carlborg||2017||A descriptive study of the first year follow-up of the Psychiatric Response Team (PAM) in Stockholm, Sweden, which was started in spring 2015. The results are PAM was requested 1,580 times and had 1,254 cases attended to. 1,036 individuals of all ages were attended to, and 96 of them had contact more than once. 1/3 of all attended cases resulted in no further action after a psychiatric assessment and sometimes crisis intervention had been made on site.|
|Police actions/Int’l||Braga||2001||This study used 911 data from nine selected studies of Houston, Jersey City, Kansas City, Minneapolis, St. Louis, and Beenleigh (Queensland, Australia) to locate “hot spots” and found that focused police actions can prevent crime and disorder in crime hot spots without necessarily resulting in crime displacement.|
|Police actions||Charlier & Reichert||2021||This article examines the literature on the evolution of deflection in policing and the current framework around police use of deflection. The literature mentioned explored effective practices for police deflection and lessons learned from established programs. More research needs to be done on deflection programs.|
|Call volume/call type||Chohlas-Wood, Merali, Reed et al.||2015||Researchers mined 911 and 311 data from 2013–2014 in New York City and disaggregated calls by type to detect patterns in types such as “noise” and “crime.” Researchers suggested that other data like weather and humidity could be included to refine results and better detect patterns.|
|Response time||Cihan||2014||As a follow-up to the 2012 study by the same authors, this study compared 5,898 in-progress burglary calls for service in Houston and 7,746 in Dallas in 2006. Concentrated disadvantage, immigrant concentration, and residential stability were important predictors of the distribution of police response time patterns in Dallas and Houston, although not always in the same ways.|
|Response time||Cihan, Zhang & Hoover||2012||This study of Houston 911 data for 5,290 in-progress burglary calls for service in 2007 found correlation between response time and likelihood of arrest for in-progress burglaries. The study also examined neighborhood characteristics, finding police calls for service had faster response times in disadvantaged neighborhoods than in more affluent ones, as determined by census tract data.|
|Response time/Int’l||Coupe & Blake||2005||This study of two-officer patrols and 911 response sampled 406 911 cases between July and December 1996 from an anonymized police force serving 2.6 million people and found that quicker response times elevated the likelihood of an arrest being made in the UK, although the authors specifically declined to extend those results to U.S. policing.|
|Call volume||Cramer, Brown & Hu||2011||Researchers analyzed 2,000 calls made to 911 between January 1 and May 31, 1998, in the Portland, Oregon, metro area to determine where areas of concentration of call volume occurred and compared these hotspots to neighborhood data to see what factors influenced call volumes.|
|Call volume/911 call takers||Dankert, Driscoll & Torres||2015||Researchers analyzed CAD data from San Francisco from May 2011 to February 2015, and collected data on 475 calls by shadowing dispatchers to examine whether the city’s CAD system was adequate to deal with increases in call volume. They recommended small changes to increase efficiency as well as to increase transparency into “unknown type” calls.|
|Caller behavior/call volume/Race/Use of force||Desmond et al.||2016||This study analyzes how one of Milwaukee’s most publicized cases, the beating of Frank Jude, affected police-related 911 calls. The results were residents of Milwaukee’s neighborhoods, especially black neighborhoods, were far less likely to report crime after Jude’s beating was broadcast. The effect lasted for over a year and resulted in a net loss of ~22,200 calls for service. Other cases of police violence against unarmed black men also had a significant impact on crime reporting in Milwaukee.|
|Alternatives to police/police actions||El-Sabawi & Carroll||2020||This article explains and recommends the Model Behavioral Health Response Team Act (BHRT) – modeled after CAHOOTS – that can be tailored to meet the needs of local and state policymakers who want to defund and replace the police in responding to mental health, substance use, and housing crises. It is informed by empirical evidence, federal guidelines, and a case-study of political activity motivated by police use of excessive force in Greensboro, N.C.|
|Police actions/ call volume||Famega, Frank & Mazerolle||2015||Researchers studied Baltimore dispatch data gathered over 1,304 hours of observation in 1999 to see how it could be used to assess efficiency and manage patrol officers’ time. They concluded that analyzing call patterns and adjusting patrol routes to include predicted high call volume areas and times could increase efficiency.|
|911 call takers/alternatives to police||Gardett, Clawson, Scott et al.||2016||A literature review of 149 studies of emergency dispatch research. Common topics of study were first point of contact care, professional status and consistency/protocols/training, resource allocation, and best practices for dispatch. Additional emerging topics that are ripe for study include using CAD/algorithms to determine the best response to an emergency.|
|Caller behavior/Int’l||Ghasemi et al.||2020||In this exploratory study, the researchers applied machine learning (ML) algorithms to a data set with nearly 100,000 LS/CMI administrations to provincial corrections clientele in Ontario, Canada, and followed up in approximately 3 years. They found ML was a slightly more effective tool to assess criminogenic risk/need or recidivism for some types of individuals.|
|911 call takers||Gillooly||2020||This dissertation is squarely focused on the role of 911 in American policing. The author worked for two years as a 911 call-taker in Southeast Michigan, which allowed her an inside perspective of the challenges facing dispatchers. She recommends police leaders formally acknowledge the functions call-takers play in policing, codify the effective tactics of call-takers, and technological reforms to assist call-taker’s performance in key functions.|
|911 call takers/caller behavior||Gilsinan||1989||This study used 265 recorded 911 calls to examine the interpretive function of call-takers in event construction before the call is dispatched and found that the types of questions asked help callers determine their narrative (and description) of the event.|
|Police actions||Glazener et al.||2019||This study examined Prince George’s County, Maryland police department’s 65 patrol beats, over a 10‐ year period, 2006–2015. They found that the calls for service rate is the most important variable in explaining misdemeanor enforcement variation. misdemeanors make up the majority of all arrests, and misdemeanor enforcement rates vary across the jurisdiction.|
|Police actions/Data||Gonzales, Henke & Hart||2005||This DOJ report discusses the type of data gathering necessary to ensure that certain policy changes are effective in separating emergency and nonemergency policing response in the context of Baltimore, Maryland’s 311 implementation.|
|Race/Use of force||Gray & Parker||2020||This study empirically examines state-level counts of police shootings of White and Black citizens from 2014–2016 to analyze the racial threat theory. The findings provide some support the theory’s explanation of patterns in police killings of Black Americans with the exception of economic threat.|
|Race/Use of force||Hoekstra & Sloan||2020||Using data on officers dispatched to over 2 million 911 calls in 2 cities, this study found white officers use force 60 percent more and gun force twice as often as black officers on average. While white and black officers use gun force at similar rates in white and racially mixed neighborhoods, white officers are 5 times as likely to use gun force in mostly black neighborhoods.|
|Call volume||Jasso, Fountain, Baru et al.||2007||Researchers developed and tested models with data from emergency calls made between September 1, 2004, and August 31, 2006, in San Francisco, then used them to predict high call volumes as a result of anomalous occurrences. “Predictions” were tested with anomalous occurrences already in the system.|
|Response time/Civilian satisfaction||Kansas City Police Department||1978||The Kansas City Police Department studied three years of internal records and found that overall response time was statistically unrelated to arrest probability. In addition, civilian satisfaction was more closely related to expectation and perception of response than to actual response time.|
|Data/Civilian satisfaction||Kelly||2003||A study of 50 U.S. locations including both police and fire department data found that the link between objectively measurable data and subjective measures of satisfaction is tenuous at best. The researcher experienced significant difficulty in developing a sampling strategy because of inconsistencies in identifiers across the locations studied.|
|Call taker/Int’l||Kent & Antaki||2019||This UK study of 514 emergency calls found that the call-taker’s first substantive question already carried a diagnosis of the merits of the caller’s case, and an implication of the call’s likely outcome.|
|Race/Police actions||Knox, Lowe, & Mummolo||2020||The authors show that if police racially discriminate when choosing whom to investigate, analyses using administrative records to estimate racial discrimination in police behavior are statistically biased, and many quantities of interest are unidentified. Their results suggest the evidence on this topic that relies on police administrative data may be largely uninformative or even misleading.|
|Data/Responsetime||Kuhn & Hoey||1987||During the implementation of E-911 in the U.S., researchers examined the ways the system could improve police response, including what data collection is possible and how the system can match demand with deployment.|
|IPV-DV/Response time/Race||Lee, Lee & Hoover||2017||This study focused on a narrow band of IPV calls in Houston, analyzing 10,000 cases from September 2010 to August 2013 to find factors that influence response time on a personal and neighborhood level. Researchers found that the race of the caller, whether a weapon was involved, and the day and time of incidents were all significantly correlated with response time—predictably, in the case of a weapon, which raised the priority code of the call. Latino callers experienced the fastest response times. At a neighborhood level, concentrated disadvantage, immigrant concentration, and residential instability were also significantly associated with faster response times.|
|Call takers/police actions||Lum et al.||2020||This study examines the role public safety communication specialists play in the criminal justice footprint using systematic observations of the Fairfax County, VA DPSC. Their results indicate that 911 call takers can play a significant role in initially shaping the criminal justice footprint and can help prevent its growth. Organizational constraints, liabilities, and rules, the expectations/beliefs of citizens and activities of police officers, can inhibit this gatekeeping function.|
|Police actions||Lum et al.||2020||This study analyzes CAD records and observes officers to better understand the degree to which police proactivity as practiced on an every-day basis reflects ideal strategies and implementation methods as recommended by the 2017 National Academies of Sciences Committee and Report on Proactive Policing. It finds there is a significant difference between interventions discussed in the research and the practice of everyday police proactivity. Notably, that proactive policing practices are limited in scope and generally implemented in less than optimal ways. Patrol commanders have minimal control over proactive practices generally.|
|Data/Responsetime||Maxfield||1982||This paper examines how information routinely collected by urban police departments may be used to monitor the performance of the patrol response function. Data from one anonymized large city is used to examine the problem of delay in responding to civilian requests for police service.|
|Data||McDevitt, Sheppard, & Douglas||2018||This paper provides an overview of big data analysis, its place in public policy, and an assessment of the general practical and ethical concerns related to its use. The Boston Police Department’s (BPD) 911 call data is provided as an example of how big data may be treated and used effectively.|
|Call type/call volume/Data||McEwen, Ahn, Pendleton et al.||2002||A study combining national surveys of 420 police departments and case studies in San Diego, the District of Columbia, and Aurora, Colorado, found that CAD systems collect rich basic data that can and should be used to support community policing, and that less than 20 percent of the civilian calls in a CAD system are for serious crime incidents. The rest are for incidents that affect callers’ quality of life to such an extent that they believe police intervention is necessary.” The major identified weaknesses in CAD data are insufficient list of call types (largest call volume is “other”) and dependence on caller assessment (e.g., is it burglary or robbery?).|
|Co-responder/Police actions||Morabito & Savage||2021||Using grant funding, the Boston Police Dept. implemented a 1-month dedicated Co-Responder car pilot program and collected data about all related encounters. This study evaluated the outcomes for this program. Results suggest much of the co-response work is proactive and that formal actions are infrequently used.|
|Co-response/mental-behavioral health/police actions||Morabito & Savage||2021||This study discusses the development and implementation of the Boston Police Dept.’s Co-Response model and dedicated car pilot. The authors examine how the Co-Responder team spent its time during the dedicated car pilot and the outcomes of encounters with community members, with a focus on involuntary evaluations and commitments and what predicted these particular outcomes. Results indicate that much of the Co-Response work is proactive and that formal dispositions, such as involuntary commitment and evaluation, are used sparingly.|
|Response time/Police actions||Moskos||2007||The study examined approximately 113,000 calls made in 2000 in Baltimore’s Eastern District to determine whether response time has a positive effect on odds of arrest or a deterrent effect on crime, and found that the effect in either case was minimal at best.|
|Police actions/Data||National Institute of Justice||1999||This compilation of discussion papers details the results of three meetings called “Measuring What Matters.” The meetings were based on these topics: “Measuring What Matters in Policing,” “How Higher Expectations for Police Departments Can Lead to a Decrease in Crime” and “A New Way of Thinking About Crime and Public Order.”|
|Data/Call takers/multiple themes||Nuesteter et al.||2019||This report summarizes the current state of 911 research, discusses the problems and potential of current 911 data collection practices, and recommends steps that law enforcement and emergency communications professionals can take to conserve resources and help ensure that the right response reaches the right caller at the right time.|
|Call type||O’Brien & Sampson||2015||Researchers re-examined “broken windows” policing as a paradigm and found that it doesn’t hold up to large-scale data analysis. A study of 200,000 calls for service from Boston showed that private conflict is a better predictor of crime than public disorder.|
|Crime concentration||O’Brien, Ciomek, & Tucker||2021||This study analyzes how high crime concentration on particular streets varies across neighborhoods in Boston. It evaluates the degree to which neighborhoods have characteristic levels of crime concentration and then tests two hypotheses: the compositional hypothesis and the social control hypothesis. It found that the degree to which crime concentrates is driven by neighborhood diversity and neighborhood level processes.|
|Data/police actions||Parks||1984||Survey data from Los Angeles, California, and Tuscaloosa, Alabama, shows that objective and subjective measures of police performance aren’t necessarily exclusive: conceptually linked objective and subjective measures return correlated results. More data is needed to confirm the results of this study.|
|Response time/Civilian satisfaction||Pate, Ferrara, Bowers et al.||1976||This Kansas City, Missouri, study of 1,106 response time surveys collected over a four-month period in the South Patrol District in 1973 showed that short response time is likely to be unrelated to positive results, but can be related to civilian satisfaction. Setting and meeting expectations was more important to satisfaction than actual response time.|
|Response time/Civilian satisfaction||Priest & Carter||1999||This study surveyed 338 people in Charlotte, North Carolina, most of whom were black, and found a strong relationship between respondents’ evaluations of police response time and their evaluations of overall police performance. Respondents’ evaluations of the service their neighborhood receives also influenced their evaluations of overall police performance. The authors noted that previous studies had significantly different results but sampled a populations consisting mostly of white people.|
|Data/Police actions||Sherman||1998||This project discusses the research potential of CompStat in developing evidence-based policing methods, including what evidence is necessary all the way through case outcome. At the time of the study, current data practices were to collect only time of response rather than quality of service or repeat call data.|
|Response time/Police actions||Spelman & Brown||1984||This U.S. Department of Justice study of four jurisdictions— Jacksonville, Florida; Peoria, Illinois; Rochester, New York; and San Diego, California—confirmed work by Kansas City Police Department that improved response time to crime calls does not significantly increase odds of arrest. The researchers hypothesized that this is because callers delay reporting until crime is over—even with access to instantaneous reporting via 911. The researchers found a slight correlation between the type of crime and whether police response time had a statistically significant impact on likelihood of arrest, noting that, in most cases, it did not.|
|Response time||Stevens, Webster & Stipak||1980||A study of York, Pennsylvania, data, sampling approximately 31,000 calls for service in 1976 found little if any correspondence between response time and likelihood that a crime will be “cleared.” Researchers noted that more study is needed with more variables such as call type: response time almost certainly makes a difference for some calls but not others, and this was lost when aggregating all calls into three categories in the overall sample.|
|Call volume/police actions||Stinson, Brewer & Liederbach||2014||Researchers analyzed a year of call for service type and location data from the Lorain, Ohio, CAD system to optimize police districting to better serve hot spots and balance workload.|
|Use of force||Taylor||2019||This study used a randomized controlled experiment that incorporated a police firearms simulator and 306 active law enforcement officers to examine the effects of dispatch priming on an officer’s decision to use deadly force. The results suggest officers rely heavily on dispatched information in making the decision to pull the trigger when confronted with an ambiguously armed subject in a simulated environment. When the dispatched information was erroneous, it contributed to a significant increase in shooting errors.|
|Mental-behavioral health||Tentner et al.||2019||Using CPD arrest and CFD BH-involved ambulance event data, the researchers identified at-risk individuals who accumulated at least 1 BH-involved ambulance and at least 1 arrest event between May 2016 and April 2017. Of the individuals and events in the emergency events data, 1842 at-risk individuals accounted for 2.2% of individuals, 5.6% of all events, and 16% of BH-involved CFD events with police involvement. A total of 330 high-use individuals accounted for 0.4% of individuals, 2% of events, and 4.7% of CFD events with police involvement. Top-100 high-use locations accounted for 9% of CFD events, and individuals of high-use location events are largely distinct from high-use individuals.|
|Multiple themes||The Center for Evidence-Based Crime Policy (CEBCP at George Mason University)||2020||Translational Criminology magazine that features international articles on criminal justice research, practice, and policy. Articles topics include perspectives from a deputy chief, data usage by police in São Paulo, Brazil, policing research in crisis and other relevant subjects.|
|Police actions||Todd & Chauhan||2020||This study examined crisis incidents (n = 19,648) over a three-year period from the Seattle Police Department (SPD). They found that arrest was the least frequent outcome followed by no action, referral to services, and emergency detention. Arrests occurred primarily in the context of person and property-related crimes and when officers perceived individuals as belligerent and disruptive. Emergency detention was significantly more likely when officers perceived a suicide risk than when they did not.|
|IPV-DV/Call takers||Townsend, Hunt, Kuck et al.||2005||This study of IPV call handling from intake to outcome shows how call-taker training can be part of an early intervention to shape IPV call procedure. The study found, among other things, that only half of the 368 departments surveyed required specialized training for call-takers and dispatchers regarding IPV.|
|Mental-behavioral health/IPV-DV/Int’l/call volume||Vaughan, Wuschke, Hewitt et al.||2018||This study mapped 20,000 mental health-related and 20,000 IPV calls for police service in Surrey, BC, and found that they have a distinct temporal pattern for both days of the week and hours of the day. Specifically, mental health calls for police service peak during the middle of the week and in the midafternoon, while IPV calls peak on Saturday and Sunday between 6:00 pm and 2:00 am.|
|Police actions/alternatives to police||Vermeer, Woods, & Jackson||2020||This article details the commonality between what police and the ‘defund the police’ movement want: for police to be last resort in solving community problems. The authors explain the current role of police in society and give examples of related successful and unsuccessful interventions and policies. The authors argue that if police are to be defunded, they can’t be expected to handle all of the same problems.|
|Alternatives to police/Diversion||Wahbi, Johnson, & Beletsky||2020||This report discusses Integrated Service Facilities, 24/7 facilities that bundle assistance for substance use, mental and behavioral health, housing, and other health, legal, and social needs. It characterizes them, describes their purpose and details public support for their implementation.|
|Co-response/call type/call takers||Watson et al.||2021||This paper examines how factors in the environment and response process affect how CIT calls are resolved. Factors analyzed include CIT response, call location and upstream decisions to pre-identify calls as mental health-related. Its findings indicate that CIT response, dispatch coding and the place calls original contribute to shaping outcomes.|
|Use of force/police actions||Weisburst||2020||This study shows that arrests critically depend on which officer responds to a 911 call, with a 1 standard deviation increase in officer arrest propensity corresponding to a 30% increase in arrest likelihood. High arrest propensity officers use force more often, make more low-level arrests, are more likely to arrest non-white civilians, and are less likely to make felony arrests that result in conviction. There is suggestive evidence that officers with low levels of experience are driving these results.|
|Use of force/Race||Weisburst||2019||The author investigates police bias in use-of-force incidents by analyzing over 130,000 arrests over 3 years in Dallas, Texas. The results showed that black civilians are disproportionately likely to be involved in a use-of-force incident during the course of an arrest, but this disparity stems directly from differences in the likelihood of arrest; the black share of use-of-force incidents are nearly identical to the black share of total arrests, at ~50 percent. Conditional on arrest, use-of-force rates do not differ systematically by race. There’s limited evidence of taste-based racial bias in use of force, conditional on arrest.|
|Mental-behavioral health/police actions||Wood, Watson, & Barber||2020||This paper provides insight into officer perceptions and experiences of the mental health-related calls they respond to involving youth, adults and families. Deficiencies in the resources needed to address the unmet needs of people and their families frustrate officers’ desires to make a difference and effect long-term outcomes.|
|Police actions||Wu, Koper, & Lum||2021||Police calls for service and automated vehicle location data from a large suburban jurisdiction were used to create comprehensive measures of police proactivity. The study found that daily police proactivity was highly stable at micro places, although police did intensify their activities briefly in response to recent changes in crime. Increases in proactive patrol generated immediate increases in crime reporting, followed by fleeting residual deterrent effects that were weaker and less robust.|
This article co-authored by Rebecca Neusteter describes the origins of the 911 system as a reaction to the civil rights protests of the 1960s.
This is a “starter kit” that discusses methods for creating change including social innovation, nonprofits and system leadership.
This is the fall 2020 issue of Translational Criminology magazine and notably contains a framework of emergency response approaches to vulnerable populations, harm-reduction in policing and the role of research partnerships for reforming policing in Scotland.
This paper proposes an evidence-based social support program (ACT) to replace police.
This is a report by the DOJ about determining key metric indicators in policing/emergency response.
This report details Los Angeles County’s diversion plans, specifically its goals around creating social service frameworks to keep people out of prisons and living in their communities.
This report details the activities and achievements of the Valley Communication Center (serving South King County residents, workers and public safety agencies). It covers metrics including but not limited to quality assurance, employee retention and average call occupation time.
Help us Transform911.
The Health Lab strives to improve public health, its impacts, and how it is discussed. If you identify an area of our work that you believe misses a critical perspective or employs language that needs improvement, please contact us at firstname.lastname@example.org. We welcome your feedback.