Calling the Police via 911 Versus a Non-Emergency Number: Variation Among Telephone Reporting Methods and their Implications for Public Safety

Abstract

Much police activity occurs at the request of the public. Most requests from the public are received by the police via telephone. I assessed and compared the characteristics of ~255,000 police calls for service generated via calls to 911 and non-emergency numbers. Dispatched calls for service generated via 911 calls were, on average, higher priority and dispatched faster and to more officers who arrived quicker and spent longer on-scene than dispatched calls for service generated via non-emergency calls. Nonetheless, many emergent events were reported via non-emergency numbers. I highlight the implications of telephone reporting methods for public safety, interrogate the concept of an “emergency” as it applies to policing, and describe the importance of non-emergency call handling for policing operations.

Keywords

911 call-takers calls for service dispatchers policing public safety

Introduction

The 911 system receives approximately 240 million calls annually in the USA (National Emergency Number Association, 2023). Whereas the sight of smoke may generate a request for an urgent fire response, a feeling of chest pain may generate a request for an urgent medical response, and a crime in-progress may generate a request for an urgent police response. Regardless of the emergency service requested, the 911 system is responsible for effectively and efficiently connecting people in need of immediate help with the emergency responders who can provide such help.

Formally introduced in the USA in 1968, the 911¹ system provides the foundation for which the emergency response system now stands (Neusteter et al., 2019). As the first, first responders, dispatchers—which includes personnel working in the specialized roles of call-takers (who engage with the public) and radio-operators (who engage with emergency responders in the field)—receive, handle, and then dispatch calls for help (Simpson, 2021). For example, call-takers must determine if an emergency is occurring and the nature of such emergency. Call-takers must also collect information about the emergency, including the location and persons involved. Radio-operators must initiate the response of the relevant emergency responders, if required, and maintain the safety of such responders during that process.

Not all callers explicitly request an emergency service, though, when calling 911. Many 911 calls are accidental. Some callers request multiple emergency services or a service not offered. Operating procedures generally dictate that all 911 calls must be handled by someone, and, hence, in the absence of an explicit request for an emergency service offered, these calls are typically handled by a police call-taker. From this perspective, police call-takers are responsible for handling a wide swath of 911 calls: both those where callers request the police and those where callers cannot articulate a clear need for any other emergency service. In many respects, this umbrella approach—which employs the police as the “default” emergency responder—reflects the generalist role that contemporary society now assigns to the police.

Not all callers are also experiencing an emergency when they call 911. In the case of policing, a caller’s need for the police may be non-urgent. This could be because the event that they are reporting is not in-progress, does not involve risk/danger, and/or only requires documentation for insurance purposes. For these non-emergent events, the police expect that the public will contact them via non-emergency numbers: by reporting non-emergent events via non-emergency numbers, police agencies are able to keep the 911 lines available for the reporting of emergent events. However, emergent events may still make their way through to the police via non-emergency numbers, similar to how non-emergent events often make their way through to the police via 911 (Gilsinan, 1989).

Given the different functions of these telephone reporting methods, calls received via 911 may be handled differently and answered quicker than calls received via non-emergency numbers. In light of such differences, it is imperative that contemporary scholarship rigorously examine the characteristics of calls for service generated via calls to 911 versus non-emergency numbers. This is especially relevant in today’s society, where the public’s access to cellular devices and associated ability to look-up non-emergency numbers (which are generally unique to each police agency and comprised of more than three digits) are greater now than ever before. As part of the present research, I empirically explore and discuss the nuance among telephone reporting methods for the police and their implications for public safety.

Background

As a public service, most police work is initiated by public requests for the police (Simpson & Bell, 2022). Most of these requests for police service are reported to the police via telephone. Although not all telephone requests are fulfilled by the police, in large part because they do not regard issues that the police can or are willing to address, requests that are determined by call-takers to warrant a police response are documented in the form of electronic calls for service. These calls for service—created by call-takers in their computer-aided dispatch (CAD) system—form the crux of operational police work.

Not only are calls for service used operationally by dispatchers to manage public requests for the police and the associated response by officers, but they are also used by researchers to study police activity. Previous research in the area of policing, specifically, and criminology and criminal justice, more broadly, has used calls for service data to examine public demand for the police (e.g., Ashby, 2020; Brantingham & Uchida, 2021; Kurtz et al., 1998; Lum et al., 2022a; Telep et al., 2014). In one recent example, Lum et al. (2022a) used calls for service data to assess the volume and types of incidents reported to the police across nine police agencies. In another example, Ashby (2020) used calls for service data to explore changes in call activity across 10 police agencies during the early months of the COVID-19 pandemic. In addition to using calls for service data to test research questions, other studies have examined the characteristics of calls for service themselves as part of their research questions (e.g., Klinger & Bridges, 1997; Nesbary, 1998; Pearce & Simpson, 2022; Simpson & Orosco, 2021).

Police calls for service and their use in research are thus not “new” to the field. However, what is arguably “new” is the growing interest in studying the dispatch system and its implications for policing outcomes. All dispatched events attended by the police are brought to the attention of officers by dispatchers: specialty trained personnel who are responsible for managing the telephone and radio system that enables help to be provided where help is required. Indeed, before a police officer is even made aware of a telephone-generated request for service, that request has already been received, assessed, informed, and interpreted by dispatchers—including those acting in the capacities of call-takers and radio-operators (e.g., Antunes & Scott, 1981; Garcia, 2015; Garcia & Parmer, 1999; Gillooly, 2020, 2022; Gilsinan, 1989; Herbst & Walker, 2001; Riou et al., 2017; Simpson, 2021; Simpson & Orosco, 2021; Whalen & Zimmerman, 1990).

This recent surge in dispatch-related research has been instrumental for informing the scholarly understanding of policing (for a commentary regarding dispatchers and their historical exclusion from mainstream criminological research, see Simpson, 2021). For example, recent research has found that call-takers’ interpretive and gatekeeping work can impact who receives police service and the outcomes of such service (Gillooly, 2020; Lum et al., 2020b). Recent research has also found that call-takers’ priority assessments can impact responding officers’ perceptions (Gillooly, 2022) as can radio-operators’ priming of officers during dispatch transmissions (Taylor, 2020). Finally, recent research has found that when call-takers are procedurally unjust during their interactions with callers, it can negatively impact callers’ engagement with call-takers and officers (Flippin et al., 2019). This latter finding exhibits tremendous relevance for conceptualizing what have traditionally been officer-focused phenomena, like procedural justice, among other actors within policing, such as call-takers (for a commentary, see Goodier & Lum, 2023).

From this collective body of research, one can reasonably assume that the work of dispatchers can bleed into the thoughts and behaviors of both callers and officers as well as shape the outcomes of public-police interactions. Although traditional police subculture may insinuate outgroup status to dispatchers, dispatchers are integral to the policing team (Orosco & Gaub, 2023; Simpson, 2021). Moreover, it is possible that the implications of the work of dispatchers may systematically vary by how the public calls the police. Much existing research of this genre has either only examined calls for service generated via 911 calls or grouped all calls for service regardless of telephone reporting method together. This analytic approach may be appropriate for testing some research questions, but it is possible that it may mask importance nuance in the handling of public requests for police service that could pose issues for testing other research questions.

For example, if emergent events are reported via non-emergency numbers, then only examining calls for service generated via 911 calls would exclude these events that were not reported via 911. Relying only on calls for service generated via 911 calls could also skew the proportion of emergent events believed to have occurred within a police agency’s jurisdiction, especially considering that many calls to 911 do not regard emergencies. Similarly, if systematic variation exists among callers who decide to report emergent events via 911 versus non-emergency numbers, such as a caller’s sociodemographic characteristics and/or attitudes toward the police, then the sample of emergent events analyzed in a study of 911-generated calls for service may not be representative of all emergent events reported to the police agency. It is not currently clear what might lead two different people to report the same event via 911 versus a non-emergency number (for a discussion of the use of non-emergency numbers, see Bain et al., 2016; Mazerolle et al., 2001).

It is also possible that variation exists among the practices of call-takers handling 911 versus non-emergency calls that could impact the quality of calls for service records, and, thereby, the amount of measurement error present in the data (Simpson & Orosco, 2021). For example, on the one hand, if call-takers handling non-emergency calls must engage in more interpretive work to complete their appraisal and gatekeeping functions because callers are not reporting an obvious emergency, then that could negatively impact the precision of the call-taker’s work. Indeed, more complex events could be misunderstood by call-takers and/or classified multiple different ways. On the other hand, if call-takers handling non-emergency calls have more time to ask questions and sort through information because there is not an urgent need for police attendance, then that could potentially enhance their precision. Call-taking can be a very challenging process, and these challenges may implicate differently in different telephone reporting methods. If all calls for service are examined together without consideration of the telephone reporting method, then one may blind themselves to potentially meaningful variation that could otherwise lurk within the data.

For all of these reasons, it is important that the characteristics of police calls for service generated via different telephone reporting methods be assessed in the current policing context. Assessing such characteristics allows for the identification and interpretation of differences among telephone reporting methods, if differences exist. Doing so also enables any differences to be theorized and accounted for in both research and practice moving forward.

Overview of the Present Research

A widely accepted expectation implies that 911 is to be used for reporting emergent events and non-emergency numbers are to be used for reporting non-emergent events. By way of construction, this expectation is intuitive: whereas 911 is a short, easy-to-remember, and universal number, non-emergency numbers are generally 10-digits and unique to each police agency (unless accessed via 311 or an equivalent scheme). Police agencies—including their dispatch centers—generally prioritize answering 911 calls over non-emergency calls under the assumption that 911 calls are more urgent and in need of a faster response. Public expectation also often stipulates that 911 calls be answered quickly, including at the expense of other inbound telephone lines, such as non-emergency lines.

Claims regarding the use of different telephone reporting methods, however, remain empirically underexamined, especially in the policing context. As part of the present research, I analyze electronic records from a municipal police agency in the Western USA to assess three important research questions regarding telephone reporting methods. First, do substantive differences exist between police calls for service that are generated via calls to 911 versus non-emergency numbers? Second, if differences do exist, what do these differences regard and are they consistent with the police’s expectations about the use of 911 versus non-emergency numbers? Finally, what percentage of some arguably emergent events, such as kidnappings, and non-emergent events, such as found property, are reported via 911 versus non-emergency numbers?

Data and Methods

Setting

The present research draws upon electronic calls for service records (hereinafter referred to as “call records”) from a municipal police agency in the Western USA (hereinafter referred to as “this police agency”). This police agency employs more than 700 sworn officers and polices an urban city with more than 400,000 residents. Like other police agencies in the USA, this police agency has experienced staffing challenges, including in its dispatch center. The municipality policed by this police agency exhibits high rates of crime—with property and violent crime rates that exceed state and national averages. More than 10% of city residents live in poverty.

Procedure

This police agency generated approximately 386,000 call records via its CAD system in 2021. Call records, however, reflect only a sample of all telephone calls received by the police—suggesting that this police agency likely received many more telephone calls that did not result in the creation of formal call records. Within their call record dataset, this police agency indicates if the call record was generated via a telephone call to 911, a telephone call not to 911 (i.e., what I term “non-emergency numbers”), by dispatch at the request of an operational unit, or by the operational unit using their in-car computer. Given my interest in telephone-generated calls for service, I first removed the approximately 75,000 call records that were generated at the request of police officers (e.g., for self-initiated officer activities, such as traffic stops and emergency traffic, and other on-view activity).

As an added layer of verification, I also removed the approximately 56,000 call records that were technically listed as telephone-generated, but were not appropriate to include in my analyses. This was because they (1) would not likely have been generated by a public request for police service (e.g., “LoJack Hit”), (2) were classified as specific dispositions which by definition must be indicated by officers (e.g., “Pursuit”), or (3) were classified as call-types that were simply too ambiguous to effectively assess for inclusion (e.g., “Private Eye”). Finally, I removed the several hundred call records that were missing call-type information because I could not effectively analyze them. This resulted in a final sample of approximately 255,000 telephone-generated call records for analysis.²

In addition to cleaning the information provided for each call record,³ I also used the available information to generate new indicator variables for operational details of interest. For example, using dispatch information, I identified call records where at least one operational unit was dispatched to the event. Using that information, I also calculated the total number of officers dispatched to each event⁴—noting that being dispatched is not necessarily synonymous with attending the scene. With respect to time, I calculated several duration variables using temporal information, including the length of time between call record creation and the first unit being dispatched, the average length of time for units to arrive on-scene once dispatched, and the average length of time spent on-scene at the event.

As I revisit in my limitations section, analyzing call records poses challenges. For example, call records may be misclassified by call-takers (Pearce & Simpson, 2022; Ratcliffe, 2021; Simpson & Orosco, 2021), temporal measures can exhibit variability (Langton et al., 2023), officers may not always provide response updates (Simpson & Bell, 2022), and radio-operators may not always electronically scribe all relevant information (Simpson, 2021). However, call records currently offer the best available means to assess my research questions, especially at this large-scale. I therefore examined call records as part of my analyses with consideration of their limitations and the potential implications of such limitations for inference.

Analytic Strategy

I conducted several analyses to test my research questions. First, I conducted descriptive analyses to provide snapshots of call records generated via calls to 911 versus non-emergency numbers. Next, I conducted inferential analyses to assess differences in variables across telephone reporting methods. Finally, I assessed the frequency by which some arguably emergent and non-emergent events were generated via calls to 911 versus non-emergency numbers. In addition to discussing the results from the statistical tests, I also contextualize the potential practical significance of my findings as applicable.

Results

I analyzed approximately 255,000 telephone-generated police call records from 2021. Approximately 49% of call records were generated via calls to 911 and 51% of call records were generated via calls to non-emergency numbers. Of all call records, approximately 52% were dispatched to at least one operational unit. Significantly more call records that were generated via calls to non-emergency numbers (63%) were dispatched to units than call records that were generated via calls to 911 (41%) (z = 107.30, p < .001).

Snapshots of Call Records

911 Calls

Among call records that were generated via calls to 911, approximately 41% were dispatched to at least one operational unit. The remaining call records were not dispatched to an operational unit. In many instances, this was because the call record was cancelled or a duplicate record. In other instances, call records were not dispatched because they were referred to and/or regarded other agencies. Given that non-dispatched call records present limited substantive variables to compare as a function of telephone reporting method, I discuss findings related to dispatched call records in the subsequent paragraphs.

Among dispatched call records that were generated via calls to 911, the average priority level was 1.62 (the priority classification scheme for this police agency is reverse-coded, such that lower numerical values indicate higher priority levels). Approximately 42% of these call records were classified as the highest priority level and 3% were classified as the lowest priority levels—noting that the priority level of calls made to 911 that did not result in the creation of a formal call record would have been very low (call-takers may have redirected callers for some of these calls to non-emergency numbers). The mean number of officers dispatched to each call record was three and the median number of officers was two. An average of one double-crewed unit (i.e., a unit comprised of two officers; for a discussion of the implications of single- versus double-crewed patrol units, see Simpson & Grossman, 2024) was dispatched to these call records.

Temporal measures also present insight into the police response to dispatched call records. Duration variables, however, are subject to outliers which can skew findings. For analytic purposes, I thus bound all of my duration analyses to call records where the specified duration was less than 6 hr. Once bound, the average length of time from call record creation to the first unit being dispatched was 57 min. Once dispatched, units averaged 10 min enroute to the scene—with the fastest unit averaging 8 min—and 40 min on-scene.

Non-Emergency Calls

Among call records that were generated via calls to non-emergency numbers, approximately 63% were dispatched to at least one operational unit. The remaining call records were not dispatched to an operational unit. Although many call records of this genre were not dispatched because they were cancelled or duplicate records, like in the 911 context, other call records were not dispatched for more substantive reasons. For example, many call records were concluded to broadcast or call management strategies. For the same reason provided for 911 calls, I discuss findings related to dispatched call records in the subsequent paragraphs.

Among dispatched call records that were generated via calls to non-emergency numbers, the average priority level was 2.02. Approximately 21% of these call records were classified as the highest priority level and 21% were classified as the lowest priority levels. The mean and median number of officers dispatched to each call record was two. An average of one double-crewed unit was dispatched to these call records.

For temporal measures, my analyses were again bound to call records where the specified duration was less than 6 hr. Once bound, the average length of time from call record creation to the first unit being dispatched was 78 min. Once dispatched, units averaged 11 min enroute to the scene—with the fastest unit averaging 10 min—and 37 min on-scene.

911 Versus Non-Emergency Calls

In addition to providing snapshots of call records, I also test for statistical differences in the characteristics of dispatched call records that were generated via calls to 911 versus non-emergency numbers. For the sake of simplicity, I present the findings from my comparisons in the order presented in the aforementioned sections. I also present a summary of my findings in Table 1.

Table 1.

T-tests Comparing the Characteristics of Dispatched Call Records That Were Generated Via Calls to 911 Versus Non-Emergency Numbers.

Variable	911	Non-emergency	Difference	t value
Priority	1.62	2.02	−0.40	−108.00*
# Officers dispatched	2.90	2.27	0.63	49.25*
# Double-crewed units dispatched	0.75	0.57	0.18	39.35*
Duration
First unit dispatched	56.63	78.24	−21.61	−38.75*
Enroute	10.08	10.89	−0.81	−12.25*
Fastest unit on-scene	8.08	9.58	−1.50	−25.12*
On-scene	40.27	37.14	3.13	12.91*

p < .001.

As shown in Table 1, the differences between dispatched call records that were generated via 911 versus non-emergency numbers are statistically significant for all variables. For example, dispatched call records that were generated via 911 calls were classified as higher priority than dispatched call records that were generated via non-emergency calls (t(132,970) = −108.00, p < .001). Dispatched call records that were generated via 911 calls tended to be dispatched to more officers (t(132,970) = 49.25, p < .001) and more double-crewed units (t(132,970) = 39.35, p < .001). The length of time for the first unit to be dispatched was shorter for dispatched call records that were generated via 911 calls (t(106,375) = −38.75, p < .001) as was the duration for the fastest unit to arrive on-scene (t(113,301) = −25.12, p < .001) and the enroute duration for all units to arrive on-scene (t(113,305) = −12.25, p < .001). In contrast, the length of time spent on-scene was longer for dispatched call records that were generated via 911 calls (t(113,337) = 12.91, p < .001).

I acknowledge, though, that the large sample size may have contributed to statistical significance, and so I also assessed the potential practical significance of these differences. Here I found some interesting variation as well. For example, the difference in priority levels between telephone reporting methods was 0.40. Considering that nearly all call records were classified within a three-point spread of the priority classification scheme, this difference is meaningful. The difference in the length of time for the first unit to be dispatched was also meaningful: 22 min is more than half of the average amount of time that units spent on-scene at dispatched call records.

The differences for the other variables were less practically interesting. For example, the differences in the number of personnel dispatched—both all officers (0.63 officers) and double-crewed units (0.18 units)—were quite small. This is likely because of policies which stipulate that a minimum number of officers be dispatched for certain events regardless of the reporting method. The differences in the length of time spent enroute (2 min for the first unit and 1 min for all units) and on-scene (3 min) were also quite small considering the means for these duration variables.

Are Emergencies Always Reported Via 911?

In this final section of results, I analyze the frequency by which some arguably emergent and non-emergent events were reported via 911 versus non-emergency numbers. Selecting call-types for this kind of analysis was very challenging. As I revisit later, what is defined as an “emergency” is in large part subjective based on the caller and context. However, selecting at least some call-types was required to conduct my analyses, and thus I drew upon my collective knowledge—derived via existing research and operational experience—to identify what I believed to be are arguably emergent and non-emergent call-types.

In my approach to selecting call-types, I relied upon the call-type classification itself to assess both temporal proximity and the risk of injury/escalation. For example, I defined call records classified as “Forced Entry In-Progress” as emergent because—by classification—these events are in-progress, could escalate, and involve risk to any person(s) inside of the premise. I applied a similar logic for call records classified as “Fight,” which naturally involve the risk of injury and escalation and should only be classified using the call-type of “Fight” if the fight itself is in-progress (otherwise a call-type related to “Battery” may be more appropriate). On the other hand, I defined call records classified as “Found Property” as non-emergent because they do not present any risk of injury or escalation nor are they in-progress.

Nonetheless, I recognize the measurement error potentially embedded in call-type classifications, which could complicate this process even beyond disputes about the definition of an “emergency.” Callers may not be able to provide descriptions of events and/or may describe events incorrectly. Some events can be classified as multiple different call-types and so call-takers may not always classify them in the same way. Events can also change over time, which can render earlier classifications as less accurate than later classifications (I only had access to a single classification in these data). For example, not all murders are obviously murders at first instance: a victim may die sometime after the initial call to the police about the event that led to the death.

I also note that I report on all call records as opposed to only dispatched call records for this specific set of analyses. My logic was two-fold. First, as noted earlier, many events—even those reported via 911 calls—are not formally dispatched but are still in some way handled by the police (e.g., cleared to broadcast). Second, even if a call for service is not dispatched to an officer, the caller still perceived the event as worthy enough of police attention to call the police to report it, which—depending upon their method of calling the police—provides insight into how they interpreted the use of 911 versus non-emergency numbers.

Across all call-types, only 58% of telephone-generated call records classified as the highest priority level were generated via calls to 911—suggesting that the other 42% of high priority call records were generated via calls to non-emergency numbers. As shown in Table 2, some nuance exists among call records when parsed by call-type. For example, two-thirds of call records classified using the call-type of “Forced Entry In-Progress” (66%) were generated via calls to 911 as were about half of call records classified using the call-type of “Kidnapping” (50%). Slightly more than half of call records classified using the call-type of “Subject Armed with a Weapon” (61%) were generated via calls to 911. A similar percentage of call records classified using the call-type of “Fight” (59%)—which assumes violence but not necessarily a weapon—were generated via calls to 911. Only 42% of call records classified using the call-type of “Murder”—arguably the most serious crime-type—were generated via calls to 911. These findings suggest that the remaining call records of these arguably emergent call-types were generated via calls to non-emergency numbers.

Table 2.

The Frequency by Which Call Records With Arguably Emergent and Non-emergent Call-types Were Generated Via 911 Calls.

Call-type	Nature	911 (%)
“Forced Entry In-Progress”	Emergent	66
“Subject Armed with a Weapon”	Emergent	61
“Fight”	Emergent	59
“Kidnapping”	Emergent	50
“Murder”	Emergent	42
“Found Property”	Non-emergent	12
“Stolen License Plate”	Non-emergent	8
“Auto Blocking Driveway”	Non-emergent	6
“Loud Music”	Non-emergent	4
“Barking Dog”	Non-emergent	2

Nonetheless, the analyses indicate that most call records of at least some arguably non-emergent call-types were generated via calls to non-emergency numbers. For example, call records classified using the call-types of “Barking Dog” (2%), “Loud Music” (4%), “Auto Blocking Driveway” (6%), “Stolen License Plate” (8%), and “Found Property” (12%) were all generated via calls to 911 very infrequently. Given that there are no obvious or immediate risks to safety and/or property at these events, police agencies would likely agree that they are best suited for reporting via non-emergency numbers. With that being said, it is possible that callers initially tried to report these events by calling 911, but were instructed by call-takers to call-back on a non-emergency number, at which time the call record was formally generated in the CAD system.

Discussion

Much police activity occurs at the request of the public. Most requests from the public are received by the police via telephone—including via 911 and non-emergency numbers. Different telephone reporting methods, in principle, have different functions. Whereas 911 is intended to be used for reporting emergent events, non-emergency numbers are intended to be used for reporting non-emergent events. Because the volume of telephone calls for the police often exceeds the number of personnel available to answer calls, police agencies generally prioritize the answering of 911 calls over non-emergency calls. This system of prioritization rests on the expectation that the different telephone reporting methods are used by the public as intended by the police. However, such assumption has received limited empirical exploration, at least in the policing context. Drawing upon call records from a municipal police agency in the Western USA, I examined several questions related to the characteristics of call records generated via calls to 911 versus non-emergency numbers.

When comparing police call records generated via calls to 911 versus non-emergency numbers, I found some evidence to suggest that these telephone reporting methods are working as intended. Consistent with the expectation that emergent events are reported via 911 calls, dispatched call records generated via 911 calls were, on average, higher priority and dispatched faster and to more officers who arrived quicker and spent longer on-scene than dispatched call records generated via non-emergency calls. Nonetheless, not all emergent events were reported via 911. Many emergent events, including persons armed with weapons and in-progress acts, reached the attention of the police via non-emergency numbers. Given differences in the handling of 911 and non-emergency calls, this finding that many emergent events were reported via non-emergency numbers exhibits concern.

Before discussing the implications of my findings, I remind readers that defining an “emergency” is difficult, especially if using secondary data as done here. The concept of an “emergency” is inherently subjective within policing. Whereas one person might define a burglary as an emergency, another person might not. Whereas one person might define a theft of any value as an emergency, another person might only define a theft of high value (at a threshold which, too, is subjective) as an emergency. Whereas one person might define a person brandishing a firearm as an emergency, another person might only define the active use of that firearm against another person as an emergency. It is not possible to assess this level of perceptual nuance using summary information about call records alone.

Moreover, even if more nuanced data were to be provided for analysis, like crime reports in the case of crime events, assessing perceptions of the caller at the time they made the decision to first call the police may not be possible. What matters most for these judgments are the perceptual conditions under which the initial telephone call to the police was made—not the conditions when the officer(s) eventually arrived. This kind of analysis would also require that nuanced information about each event be formally documented by officers, which is unlikely for many events, especially those not related to crime. Finally, this kind of analysis would require tremendous time from researchers to manually review reports, which would not be practically feasible if attempting to assess hundreds of thousands of events, like as done here.

In light of these challenges, I assessed differences among multiple variables as part of my research. In principle, events reported via 911 should be classified as a higher priority than events reported via non-emergency numbers. Events reported via 911 should be dispatched faster and to more units than events reported via non-emergency numbers. Events reported via 911 should be classified as call-types that reflect greater levels of risk and/or harm than events reported via non-emergency numbers. Although some of the differences in the operational response to call records may have been driven by the call record’s priority classification alone—which is assigned after a call-taker answers the telephone call and decides to generate a formal call record in the CAD system—assessing differences among multiple variables helped to minimize the potential issues that could have resulted from relying on any one single variable for the analyses.

Implications

The present research exhibits several important implications. First, even with public education about when to call 911 versus non-emergency numbers,⁵ there will still likely be discrepancies in how these telephone reporting methods are used by the public. This is because of the inherent subjectivity in what each person defines as an “emergency.” Ensuring that people only call 911 to report emergent events and non-emergency numbers to report non-emergent events would require that the public reach universal consensus with the police about what exactly constitutes an emergency, which would be incredibly challenging for many reasons. As but one example, some callers to this police agency reported events even as serious as kidnappings via non-emergency numbers.

Second, considering this variability, scholars and practitioners should not assume that every call received via 911 is an emergency and every call received via non-emergency numbers is not an emergency. It is very plausible that emergent events that require immediate police attention may be reported via non-emergency numbers. This suggests that call-takers answering non-emergency calls for the police should either be trained and equipped to handle the emergent events that are reported via non-emergency numbers or—at minimum—taught to assess when to transfer a call to a 911-trained call-taker. This latter point is especially pertinent for police agencies which task different personnel with answering 911 and non-emergency calls. It is also important that call-takers handling non-emergency calls receive the necessary supports for their health and wellness—recognizing that they may experience negative impacts from managing emergent events that should have been reported via 911 (Pierce & Lilly, 2012). Future research should examine factors that lead people to report emergent events via 911 versus non-emergency numbers.

Third, institutional priorities regarding the handling of the telephone response system should be continuously assessed. As noted earlier, calls received via 911 are often prioritized for answer by call-takers over calls received via non-emergency numbers. This is typically an operational side-effect of the imbalance between call volume and personnel available to answer telephone calls. But, if emergent events are being reported via non-emergency numbers that are not being answered immediately, that could negatively affect public safety. Delayed answer by call-takers translates into delayed response by officers which translates into delayed help for callers. In some instances, these delays could mean that a suspect flees the scene of a crime before officers arrive. In other instances, these delays could mean that an event escalates, which could lead to greater damage to property and/or greater risk to life.

In the event that a caller hangs up before a call-taker answers their non-emergency call, there is also little to no likelihood that the caller would receive the follow-up that they otherwise would have received if they called 911 and hung-up. If they hung up for reasons related to the event that they were trying to report via a non-emergency number (e.g., escalation of violence and/or injury), that lack of follow-up could pose grave concern. It could also mean that a crime goes unreported to the police, hence contributing to the dark figure of crime (Skogan, 1977). It is important that dispatch centers receive adequate staffing to minimize call wait times across all telephone reporting methods.

Fourth, police reform efforts should be expanded to include non-emergency call handling. Many reform efforts, including diversion programs and co-responder models, have historically focused on 911 calls and the calls for service generated via 911 calls. As shown by this research, many non-emergency calls could benefit from these reforms as well. The telephone reporting method should not systematically impact the likelihood of diversion if the intent of the diversion program is to divert calls away from the police. Similar arguments could be made for co-responder models as well—especially those that center around mental health. For example, 48% of call records classified by this police agency as “Mentally Ill” and “Attempt Suicide” were generated via calls to non-emergency numbers. Finally, call handling procedures that have been the subject of reform attention, such as the incorporation of procedural justice principles, should encompass non-emergency call handling as well. Call-takers handling of callers reporting events via non-emergency numbers could impact callers’ eventual interactions with officers in consequential ways.

Limitations

The use of secondary police data—like the call records examined here—does not come without limitations. For example, call-types can exhibit measurement error (e.g., Simpson & Orosco, 2021). However, given my focus on how the public perceived the events that they reported to the police, this error may not be so concerning. Indeed, it is the caller’s perception that impacts their decision to call 911 versus a non-emergency number—not the police’s perception. Thus, even if the caller’s perception of an event is technically incorrect, what mattered for this research is that they perceived it as urgent enough to call 911 to report it and vice versa.

Temporal information can also present inferential challenges (e.g., Langton et al., 2023). In many instances, officers must proactively indicate when they are enroute, on-scene, and clear of events if not specifically queried by their radio-operator—as would frequently be the case for non-emergent events. During emergent events, excessive or priority radio traffic can limit a radio-operator’s ability to ask for response updates and/or for officers to share such updates over the radio channel. This can mean that response updates are either lagged or missed, leading to imprecise duration calculations.

There are some additional limitations worthy of attention as well. For example, I only examined call records from a single police agency. Each police agency may have its own unique staffing, operating, and dispatching procedures, including those that vary from the police agency studied here, and therefore my results may not generalize to all police agencies. Such variation could also impact the magnitude of differences observed between telephone reporting methods. In this vein, it is worth noting that the high volume of crime handled by this police agency likely contributed to the delays in dispatching many call records.

I also did not have data regarding the wait time for each telephone call to be answered by a call-taker, although estimates for this police agency suggest that approximately 45% of all 911 calls are answered within 15 s. This information would have been helpful for assessing exactly how long emergent events reported via non-emergency numbers held in a queue before even being answered—recognizing that a call record cannot be generated nor dispatched to officers until it is first answered by a call-taker. Nonetheless, I remain confident in my expectation that 911 calls would have been answered faster than non-emergency calls (at least in an aggregate and/or relative sense) for the reasons discussed throughout this article.

Conclusion

911 is central to the emergency response system. It is responsible for connecting people in need of urgent help with that help. In the context of policing, this includes events where there are immediate risks to life and/or property. Considering the function of 911, many stakeholders, including governments, emergency responders and the public writ large, expect that callers report emergent events via 911 and non-emergent events via non-emergency numbers. In fact, it is this belief that 911 should be reserved exclusively for reporting emergencies that has led to widespread education campaigns about the intended use of 911 versus non-emergency numbers.

Despite efforts to align telephone reporting methods with their intended functions, discrepancies still exist. Scholars and practitioners must thus acknowledge that not all emergent events are reported via 911 and not all calls received via non-emergency numbers are non-emergent. It is critically important that dispatch centers are sufficiently staffed to ensure that all incoming telephone calls are answered and assessed in a timely manner. In the absence of sufficient staffing, it may be operationally necessary to prioritize answering some telephone calls at the expense of others. As demonstrated by my findings, this approach—which has been the norm for some time for many police agencies—may incur significant costs given the public’s discrepant use of telephone reporting methods. Reformers should also more explicitly consider the role of non-emergency call handling in public safety. Neglecting to consider the unique dynamics of telephone reporting methods could mask important variation among such methods and cloud assessments of the efficacy, efficiency, and quality of police service. Future research should continue to unpack the nuance regarding telephone reporting methods and their implications for public safety.

Footnotes

Declaration of Conflicting Interests

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

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Rylan Simpson

Notes

Author Biography

Rylan Simpson, Ph.D. is an Assistant Professor in the School of Criminology at Simon Fraser University. His research interests include policing, perceptions of police, police organizations, theories of crime, and social psychology. He approaches his research using a variety of different methodologies, including experimental and quantitative analyses. He has recently published his work in PLoS ONE, Journal of Criminal Justice, Crime Science, Journal of Experimental Criminology, and Crime & Delinquency.

References

Antunes

Scott

E. J.

(1981). Calling the cops: Police telephone operators and citizen calls for service. Journal of Criminal Justice, 9(2), 165–180.

Ashby

M. P.

(2020). Changes in police calls for service during the early months of the 2020 coronavirus pandemic. Policing: A Journal of Policy and Practice, 14(4), 1054–1072.

Bain

Brooks

Golding

Ellis

Lewis

(2016). Calling the police: The use of non-emergency 101 in England and Wales. The Police Journal, 89(1), 55–69.

Brantingham

P. J.

Uchida

C. D.

(2021). Public cooperation and the police: Do calls-for-service increase after homicides? Journal of Criminal Justice, 73, 101785.

Flippin

Reisig

M. D.

Trinkner

(2019). The effect of procedural injustice during emergency 911 calls: A factorial vignette-based study. Journal of Experimental Criminology, 15(4), 651–660.

Garcia

A. C.

(2015). “Something really weird has happened”: Losing the “big picture” in emergency calls for service. Journal of Pragmatics, 84, 102–120.

Garcia

A. C.

Parmer

P. A.

(1999). Misplaced mistrust: The collaborative construction of doubt in 911 emergency calls. Symbolic Interaction, 22(4), 297–324.

Gillooly

J. W.

(2020). How 911 callers and call-takers impact police encounters with the public: The case of the Henry Louis Gates Jr. arrest. Criminology & Public Policy, 19(3), 787–803.

Gillooly

J. W.

(2022). “Lights and sirens”: Variation in 911 call-taker risk appraisal and its effects on police officer perceptions at the scene. Journal of Policy Analysis and Management, 41(3), 762–786.

10.

Gilsinan

(1989). They is clowning tough: 911 and the social construction of reality. Criminology, 27(2), 329–344.

11.

Goodier

Lum

(2023). First point of contact: Can procedural justice be applied by emergency calltakers? Policing: A Journal of Policy and Practice, 17, paac102.

12.

Herbst

Walker

(2001). Language barriers in the delivery of police services: A study of police and Hispanic interactions in a Midwestern city. Journal of Criminal Justice, 29(4), 329–340.

13.

Klinger

D. A.

Bridges

G. S.

(1997). Measurement error in calls-for-service as an indicator of crime. Criminology, 35(4), 705–726.

14.

Kurtz

E. M.

Koons

B. A.

Taylor

R. B.

(1998). Land use, physical deterioration, resident-based control, and calls for service on urban streetblocks. Justice Quarterly, 15(1), 121–149.

15.

Langton

Verlaan

Ruiter

(2023). Operationalizing deployment time in police calls for service. Crime Science, 12, 20.

16.

Lum

Koper

C. S.

Stoltz

Goodier

Johnson

Prince

(2020b). Constrained gatekeepeers of the criminal justice footprint: A systematic social observation study of 9-1-1 calltakers and dispatchers. Justice Quarterly, 37(7), 1176–1998.

17.

Lum

Koper

C. S.

(2022a). Can we really defund the police? A nine-agency study of police response to calls for service. Police Quarterly, 25(3), 255–280.

18.

Mazerolle

Rogan

Frank

Famega

Eck

(2001). Managing citizen calls to the police: An assessment of non-emergency call systems. National Institute of Justice.

19.

National Emergency Number Association. (2023). 9-1-1 statistics. Retrieved December 29, 2023.

20.

Nesbary

D. K.

(1998). Handling emergency calls for service: Organizational production of crime statistics. Policing: An International Journal, 21(4), 576–599.

21.

Neusteter

S. R.

Mapolski

Khogali

O’Toole

(2019). The 911 call processing system: A review of the literature as it relates to policing. Vera Institute of Justice.

22.

Orosco

Gaub

(2023). “I am doing my part, you are doing your part”: The sworn-civilian divide in police dispatching. Policing: An International Journal, 46(1), 164–178.

23.

Pearce

Simpson

(2022). The role of police in conducting wellness checks: Insight from a study of police data. Police Practice and Research, 23(4), 400–413.

24.

Pierce

Lilly

M. M.

(2012). Duty-related trauma exposure in 911 telecommunicators: Considering the risk for posttraumatic stress. Journal of Traumatic Stress, 25(2), 211–215.

25.

Ratcliffe

J. H.

(2021). Policing and public health calls for service in Philadelphia. Crime Science, 10, 5.

26.

Riou

Ball

Williams

T. A.

Whiteside

O’Halloran

K. L.

Bray

Perkins

G. D.

Smith

Cameron

Fatovich

D. M.

Inoue

Bailey

Brink

Finn

(2017). “Tell me exactly what’s happened”: When linguistic choices affect the efficiency of emergency calls for cardiac arrest. Resuscitation, 117, 58–65.

27.

Simpson

(2021). Calling the police: Dispatchers as important interpreters and manufacturers of calls for service data. Policing: A Journal of Policy and Practice, 15(2), 1537–1545.

28.

Simpson

Bell

(2022). Unpacking the police patrol shift: Observations and complications of “electronically” riding along with police. Crime Science, 11, 15.

29.

Simpson

Grossman

(2024). Patrol officer activity by single- versus double-crewed status: The call-related output of one-officer and two-officer patrol units. Journal of Criminal Justice, 94, 102248.

30.

Simpson

Orosco

(2021). Re-assessing measurement error in police calls for service: Classifications of events by dispatchers and officers. PLoS One, 16(12), e0260365.

31.

Skogan

W. G.

(1977). Dimensions of the dark figure of unreported crime. Crime & Delinquency, 23(1), 41–50.

32.

Taylor

P. L.

(2020). Dispatch priming and the police decision to use deadly force. Police Quarterly, 23(3), 311–332.

33.

Telep

C. W.

Mitchell

R. J.

Weisburd

(2014). How much time should the police spend at crime hot spots? Answers from a police agency directed randomized field trial in Sacramento, California. Justice Quarterly, 31(5), 905–933.

34.

Whalen

M. R.

Zimmerman

D. H.

(1990). Describing trouble: Practical epistemology in citizen calls to the police. Language in Society, 19(4), 465–492.