ON TIME ON TIME ON TIME BUSINESS MANAGEMENT A+ WORK, ON TIME, NO PLAGARIZING; ON TIME

DOI: 10.1111/1745-9133.12460

R E S E A R C H A R T I C L E

BEYOND FALSE POSITIVES

Beyond false positives

A typology of police shooting errors

Paul L. Taylor

University at Albany, State University of New

York

Correspondence Paul L. Taylor, School of Criminal Justice,

University at Albany, State University of New

York, Draper Hall Room 213, 135 Western

Avenue, Albany, NY 1222.

Email: [email protected]

Research Summary: Daniel Kahneman (2011) wrote,

“There are distinctive patterns in the errors people make.

Systemic errors are known as biases, and they recur pre-

dictability in particular circumstances. … The availability

of diagnostic labels for [these] biases make [them] easier to

anticipate, recognize, and understand.” In this article, we

examine the systemic nature of human error in the context

of officer-involved shootings—one of the most visible and

controversial aspects of the U.S. criminal justice system—

and we seek to provide a common language for discussing,

recognizing, and understanding these tragic outcomes.

Policy Implications: The resulting typology offers a

framework for a systems-based approach to researching

and investigating police shooting errors that, in turn, could

provide a powerful vehicle for reform, improved officer

decision-making, and ultimately better outcomes.

K E Y W O R D S deadly force, decision-making, human error, police

“To study error is to study the function of the system in which practitioners are embedded.”

—Woods, Dekker, Cook, Johannesen, & Sarter, 2010, p. 25

Daniel Kahneman (2011) wrote, “There are distinctive patterns in the errors people make. Systemic

errors are known as biases, and they recur predictably in particular circumstances. … The availability

of diagnostic labels for [these] biases make [them] easier to anticipate, recognize, and understand”

Criminology & Public Policy. 2019;18:807–822. wileyonlinelibrary.com/journal/capp © 2019 American Society of Criminology 807

808 TAYLOR

(pp. 3–4). The social and behavioral sciences are underpinned by the assumption that people with sim-

ilar experiences and training respond to similar situations in similar ways. We understand that behavior

tends to be systematically connected to the features of peoples’ tools, tasks, previous experiences, train-

ing, and environments (Dekker, 2014; Klein, 2011; Lipsky, 2010). In a similar fashion, the research

findings on human error have consistently demonstrated that situations, behaviors, and decision pro-

cesses that result in error tend to result in repeated errors across time and people (Kahneman, 2011;

Reason, 1990; Woods et al., 2010). This has allowed for the systematic study of error to be used as a

vehicle for understanding workplace decision-making, professional reform, and improved outcomes in

a number of high-risk occupational fields including medicine (Institute of Medicine, 2000), aviation

(Dekker, 2016; Reason, 2008), transportation (Green, 2017), and the military (Snook, 2002). James

Doyle (2010) and a growing group of others (e.g., Hollway, 2014; Shane, 2013) have called for a sim-

ilar lens to be applied to various aspects of the U.S. criminal justice system including police use of

deadly force (e.g., Pickering & Klinger, 2016; Sherman, 2018).

The purpose of this article is to present a typology of police shooting errors. The goal is to provide

both researchers and practitioners with a new framework—similar to Kahneman’s (2011) “diagnostic

labels”—to help facilitate more meaningfully research, understanding, better prediction, and perhaps

even the reduction of some of the most controversial police shootings.

1 HUMAN ERROR

James Reason (1990) defined human error as “a generic term to encompass all those occasions in which

a planned sequence of mental or physical activities fails to achieve their intended outcome, and when

these failures cannot be attributed to the intervention of some chance agency” (p. 9). The key words in

this definition are “intended” and “outcome,” and they are important to keep in mind if we are to apply

this definition of human error to officer-involved shootings. It would not be appropriate, for example,

to apply the term “error” to all police shootings involving unarmed people. If, for instance, an officer

was struggling with a person the officer knew or believed to be unarmed and the officer intentionally

shot the person anyway, this would not meet the definition of an error. In this case, the officer intended

to shoot an unarmed person and the outcome matched the intention. If, on the other hand, the officer

believed the person was armed and the officer intentionally shot the person when he reached into his

pocket and it subsequently turned out he was unarmed, this would meet James Reason’s definition of

human error. The officer in this case intended to shoot an armed person but, instead, shot an unarmed

person. The result is an unintended outcome. In a similar fashion, if an officer intentionally shot an

unarmed person in an act of criminal malice, it would not meet the definition of human error. The

officer intended to shoot an unarmed person illegally, and the action would be considered a “violation”

rather than an “error” by the standards outlined in the literature on human error (Reason, 2008).

The scholars who study human error generally agree that systematic errors should be viewed as the

undesirable by-product of otherwise useful psychological processes rather than irrational or maladap-

tive tendencies (e.g., Dekker, 2014; Kahneman, 2011; Reason, 1990; Woods et al., 2010). In fact, Ernst

Mach (1975/1905) wrote, “Knowledge and error flow from the same mental source, only success [or

failure] can tell one from the other” (p. 80). James Reason (1990) referred to this as a type of “cognitive

balance sheet” in which correct action and systematic error can be viewed as “two sides of the same

coin” (p. 2).

We know that cognitive and attentional resources are finite. Humans are not able to attend to a

fraction of the stimuli in their environment at any given time, and the more of their attentional resources

they devote to an object or a task, the less they have to give to other objects or tasks (Chabris & Simons,

TAYLOR 809

2010). To overcome this limitation, people tend to rely on pattern recognition and automaticity in their

decision-making strategies (e.g., Kahneman, 2010; Klein, 2011). The benefit of developing cognitive

automaticity—System 1 in Kahneman’s (2011) vernacular—is that it utilizes implicit shortcuts (i.e.,

heuristics and biases) that allow for us to speed up our decision-making processes with a high degree

of reliability while freeing up our more deliberative attentional and cognitive resources—System 2—

for more complex and/or novel problems. Most of the time, this is beneficial. In fact, expertise can

be defined as the optimization of heuristic decision-making (e.g., Klein, 2011; Schmidt & Lee, 2014).

There is also a downside, however; the cost of developing automaticity in our decision-making and

its subsequent behavioral output is that we give up the ability to monitor the appropriateness of the

decision strategy and/or behavioral response we are employing at an implicit level, and this leaves us

susceptible to error—the debit in the cognitive balance sheet. In fact, James Reason (1990) wrote that,

“[A]utomaticity (the delegation of control to low-level [cognitive] specialists) makes slips, or actions-

not-as-planned, inevitable” (p. 2).

2 TYPES OF ERROR

Given the vast potential for error across human activities, it is somewhat surprising how limited error

is in both its abundance and its variation. Error tends to be much rarer than correct action and takes a

limited number of forms across a wide range of mental and behavioral activities. Thus, according to

Reason (1990), “[I]t is possible to identify comparable error forms in action, speech, perception, recall,

recognition, judgement, problem solving, decision making, concept formation and the like” (p. 2).

Rasmussen and Jensen (1974) developed, and Reason further refined (1990), a “skill-rule-knowledge”

framework for understanding the cognitive processes behind three levels of human performance and the

subsequent errors that can occur at each level. Reason (1990) defined these levels in the following ways:

Skill-based level

At the skill-based level, human performance is governed by stored patterns of preprogrammed

instructions represented as analogue structures in a time-space domain. Errors at this level are

related to intrinsic variability of force, space or time coordination.

Rule-based level

The rule-based level is applicable to tackling familiar problems in which solutions are governed

by stored rules (productions) of the type if (state) then (diagnosis) or if (state) then (remedial

actions). Here, errors are typically associated with the misclassification of situations leading to

the application of the wrong rule or with the incorrect recall of procedures.

Knowledge-based level

The knowledge-based level comes into play in novel situations for which actions must be planned

on-line, using conscious analytical processes and stored knowledge. Errors at this level arise from

resource limitations (‘bounded rationality’) and incomplete or incorrect knowledge.

With increasing expertise, the primary focus of control moves from the knowledge-based towards

the skilled-based levels. (p. 43).

This conception of effortful processing giving way to cognitive shortcuts and implicit automaticity

with exposure to similar experiences and patterns is consistent with Kahneman and Tversky’s (1979)

810 TAYLOR

T A B L E 1 Characteristics of skill-based, rule-based, and knowledge-based errors

Dimension Skill-Based Slips and Lapses

Rule-Based Mistakes

Knowledge-Based Mistakes

Type of activity Routine actions Problem-solving activities

Focus of attention On something other

than the task at

hand

Directed at problem-related issues

Control mode Mainly by automatic processors (schemata) (stored

rules)

Limited, conscious

processing

Predictability of

error types

Mostly predictable “strong-but-wrong” errors

(actions) (rules)

Variable

Ration of error to

opportunity for

error

Although absolute numbers may be high, these

constitute a small proportion of the total number

of opportunities for error

Absolute numbers

small, but

opportunity ratio

high

Influence of

situational factors

Low to moderate; intrinsic factors (frequency of

prior use) likely to exert the dominant influence

Extrinsic factors

likely to dominate

Ease of detection Detection usually

fairly rapid and

effective

Difficult, and often only achieved through external

intervention

Relationship to

change

Knowledge of

change not

accessed at proper

time

When and how

anticipated change

will occur

unknown

Changes not

prepared for or

anticipated

Source. This table was taken directly from Reason (1990, p. 62).

dual-process theory. The skill- and rule-based levels neatly align with Kahneman’s conception of

System 1 processing, whereas the knowledge-based level corresponds with the effortful System 2

(Kahneman, 2011). Using Rasmussen’s performance levels, Reason (1990) defined three associated

error types: skill-based slips and lapses, rule-based mistakes, and knowledge-based mistakes (Table 1).

Where these errors show up depends primarily on familiarity with the task in question and on the

correlated type of cognitive processing engaged to solve the problem. The question is: How do these

errors manifest themselves in the context of police-involved shootings?

3 POLICE USE OF DEADLY FORCE

We know that police shootings are rare events (Alpert & Fridell, 1992; Fyfe, 1988; Geller & Scott,

1992; Sherman, 1980). Yet, they have become one of the most “visible and controversial” aspects of

the U.S. criminal justice system today (Klinger, Rosenfeld, Isom, & Deckard, 2015, p. 194). Despite

the controversy, most police shootings never rise to the level of public consciousness (Zimring, 2017).

Instead, the public discourse tends to center on just a handful of cases, and as a result, these cases have

a disproportionate impact on the public’s perception of police competence and legitimacy (Gau, 2014;

Pickering & Klinger, 2016; White, 2015). Four such police shootings will be used as examples for this

article:

1. On December 12, 2016, in Bakersfield, California, police officers responded to reports of a man

brandishing a firearm. Responding officers shot and killed Francisco Serna, a 73-year-old man

TAYLOR 811

suffering from the early stages of dementia, who rapidly produced a dark-colored object from his

pocket that was later determined to be a crucifix (Washington Post Staff, 2016).

2. On January 1, 2009, in Oakland, California, police officers contacted Oscar Grant on a train platform

in Fruitvale Station. A struggle ensued between Grant and the officers. During the struggle, Grant

ended up in a prone position with his hands pulled underneath his chest and officers on top of

him attempting to pull his hands out from underneath him. Unsuccessful, one officer told the other

officers to “get back” and announced he was going to “TASE” Grant. The officer proceeded to stand

up, draw his firearm, and shoot Oscar Grant one time in the back killing him (Martin, 2016). Later,

in testimony during his trial for murder, the officer adamantly maintained that he intended to use

an electronic control device on Oscar Grant and did not intend to shoot him (State of California v. Johannes Mehserle, 2010).

3. On December 26, 2015, in Chicago, Illinois, two police officers were dispatched to a domestic

disturbance in progress. When the officers entered the apartment complex and began climbing the

stairs to the apartment in question, they were confronted by a man running down the stairs with a

baseball bat raised over his head. The closest officer rapidly discharged his firearm seven times at

the advancing man. Five rounds struck and killed the officer’s intended target. One round struck

and killed Bettie Jones, an innocent bystander (Johnson, 2018).

4. On November 20, 2014, in Brooklyn, New York, two police officers entered an unlit stairwell of a

New York City Housing Authority high-rise apartment complex. One officer, still on probationary

status, had his firearm drawn, which as it came out in the later trial of the officer, was an unofficial

practice for officers walking up the unlit stairwell of the building. Akai Gurley entered the stairwell

14 steps below the officers. The officer fired his weapon in what was claimed to be an accidental

discharge. The bullet ricocheted off a wall and struck Gurley in the chest, killing him (Wilson,

2014).

Each of these shootings made national and international headlines; officers in two shootings were

successfully prosecuted and sentenced—a rare outcome when it comes to officer-involved shootings

(Stinson, 2015)—and one shooting was made into a popular movie. To say that these tragedies have

had a disproportionate impact on the public discourse surrounding police use of deadly force would be

an understatement. In addition, the outcome of each of these shootings can be classified as human error

using James Reason’s (1990) definition. The shooting officer’s intent in each case did not align with

the ultimate outcome. In the case of Francisco Serna, the officers intended to shoot an armed threat

but ended up shooting an unarmed man with a crucifix. In the case of Oscar Grant, the officer’s stated

intention was to use his TASER, but he ended up using his firearm instead. In the case of Bettie Jones,

the officer intended to shoot an attacker with a bat, but he did not intend to shoot Bettie Jones as well.

In the case of Akai Gurley, the officer did not even intend to fire his weapon, let alone shoot another

human being. Although each of these cases falls under the general category of police use of deadly

force, the causal mechanisms behind each are much different, both between the cases and from other

instances in which the police intend to use deadly force and the ultimate outcome matches the intention

of their action. This distinction is important because, according to Woods et al. (2010), where we find

error in a professional setting, we find an opportunity to improve practice and reduce it. In the case of

officer-involved shootings, this could mean a reduction in some of the most visible and controversial

outcomes (Pickering & Klinger, 2016; Sherman, 2018).

812 TAYLOR

4 CURRENT APPROACHES TO UNDERSTANDING POLICE USE OF DEADLY FORCE

Broadly speaking, there have been two approaches to investigating and trying to understand police use

of deadly force: the academic approach and the practitioner approach. Equifinality, the principle that

a given end state can be reached by many potential means (e.g., Von Bertalanffy, 1973), is a problem

for both approaches. On the academic side, the tendency has been to ignore equifinality and aggregate

indiscriminately all available cases from a defined population or sample—typically individual cities

and departments (e.g., Fyfe, 1979; Klinger et al., 2015); a sample of cities and departments from a

national level database (e.g., Ross, 2015; Sherman & Langworthy, 1979); or all available cases from

a national level database (e.g., Kania & Mackey, 1977; Nix, Campbell, Byers, & Alpert, 2017)—on

either the outcome variable of the police discharging a firearm (e.g., Fachner, Carter, & Collaborative

Reform Initiative, 2015; Fyfe, 1978) or, even more problematic (e.g., Geller & Scott, 1992; Klinger,

2012), the death of the person against whom force was used (e.g., Nix et al., 2017). This would be akin

to aggregating all fatal aviation accidents and trying to determine the variable or variables that cause

planes to crash. Given the complexity of these events and the extensive variation in their causal paths,

model misspecification would be all but guaranteed and the factors behind any given incident or types of incidents would almost certainly be lost with such a coarse methodological approach. While the

superficial correlations that are uncovered may be of some value, the systemic causal influences for

particular types of incidents within the larger sample are likely to remain opaque. In addition, the vari-

ables of greatest interest to academics studying police use of deadly force have been sociodemographic.

Which, according to Shane and Swenson (2018), “[D]o very little, if anything, to help understand the

dynamics of a police shooting, or to help reduce use of force encounters, which should be the overriding

goal of use of force research” (p. 56).

On the other hand, law enforcement practitioners tend to treat police shootings as if equifinality is

absolute. In many of the classic studies of the police, scholars noted a culture of professional deference

to the involved officer(s) discretionary decision-making and subsequent actions taken during a use-of-

force encounter; scholars have ascribed this, in large part, to the inherent danger and the unpredictable

nature of human violence (e.g., Bittner, 1990; Muir, 1977; Skolnick & Fyfe, 1993; Van Maanen, 1973;

Wilson, 1978). To compound matters, the primary, and often only, modes of professional inquiry into

police shootings are focused almost exclusively on the criminal and administrative culpability of the

involved officer(s) (IACP, 2012; Klinger, 2004). To use fatal aviation accidents again as an illustrative

example, this would be like focusing a plane crash investigation solely on the administrative and crim-

inal culpability of the involved pilot(s). Applying such a narrow, blame-oriented lens, in the case of

both fatal aviation accidents and officer-involved shooting investigations would all but guarantee that

most causal patterns and systemic influences are missed.

Typologies are specifically designed to remedy equifinality and could be used to address the issues

found in both the prevalent academic and practitioner approaches to investigating and understanding

police shootings (Bailey, 1994; Collier, LaPorte, & Seawright, 2012). Discriminately aggregating like

cases opens them up to the power of quantitative analysis and allows for more meaningful examinations

of causal relationships. In a similar fashion, grouping cases on processes rather than on outcomes allows

for the identification of systemic flaws and weaknesses in current practice.

TAYLOR 813

T A B L E 2 Scharf and Binder’s (1983) typology of police shooting outcomes

Shots Fired Decision Yes No Correct A person who is armed and dangerous and

an immediate threat to life is prevented

from harming the officer or another

person by a use of fatal force.

A person who appears to be armed and

dangerous is not shot by a police

officer, and there are no unfortunate

consequences.

Incorrect A person presumed dangerous but, in fact,

not actually armed or dangerous is

killed by a police officer.

A police officer or citizen is killed

because a police officer fails to shoot.

Source. This table was taken directly from Scharf and Binder (1983, p. 23).

5 SCHARF AND BINDER’S TYPOLOGY OF POLICE SHOOTING OUTCOMES

Recognizing the issues with treating all police shootings as equivalent processes based on the result,

Scharf and Binder (1983) developed a typology of outcomes for potential police deadly force encoun-

ters, categorized on whether the police fired a shot and whether the decision was “correct” (Table 2).

Two of the four “logical” outcomes Scharf and Binder (1983, p. 23) identified were categorized as

errors:

False-Positive Errors: A person presumed dangerous but, in fact, not armed or dangerous is killed by a police officer.

False-Negative Errors: A police officer or citizen is killed because a police officer fails to shoot.

Although this was an important first step, a superficial examination of the four empirical cases,

described earlier in this article, demonstrates that the categories in Scharf and Binder’s typology are

not exhaustive, one of the two fundamental requirements for any typology (Bailey, 1994). In fact, only

one of the four cases, the false-positive shooting of Francisco Serna, fits into their typology at all.

6 EXTENDED TYPOLOGY OF POLICE SHOOTING ERRORS

To address this issue, an extended typology of police shooting errors has been developed that incor-

porates all instances in which the police pull the trigger of their firearm and the result is an error as

Reason (1990) defined it (Table 3).

The typology is categorized by whether an officer intends to discharge a firearm and whether the

object struck by the fired bullet was the intended target of the officer’s behavior at the time of the

shooting. Again, this typology only encompasses events in which the police discharge a firearm and

the outcome meets the definition of human error. Each bullet fired is treated as a separate event.

Therefore, a police shooting involving multiple bullets being fired by a single or multiple officers

could and often would result in a mixture of errors and/or in correct actions, as was exemplified by

the case of Bettie Jones in which the police intended to shoot an attacker but did not intend to shoot

Jones, an innocent bystander, as well.

There are several reasons for limiting the typology to police shooting errors rather than trying to

incorporate all police shootings, as Scharf and Binder (1983) did, or including errors that result from the

police not shooting. First, apart from police shootings that are intentionally criminal in nature, police

814 TAYLOR

T A B L E 3 New typology of police shooting errors

Target Hit Firearm Discharge Intended Unintended Intended Misdiagnosis Errors Misses

Unintended Misapplication Errors Unintentional Discharges

shooting errors and particularly those that result in injury or death are likely to be the most controversial

and least understood by the public. As a result, they are also more likely to have a disproportionate and

negative impact on the public’s perception of police legitimacy and competence when it comes to police

use of deadly force (Gau, 2014; Pickering & Klinger, 2016). Second, where we find human error in the

professional environment, we find an opportunity to improve future outcomes (Dekker, 2010, 2014;

Doyle, 2010; Green, 2017; Institute of Medicine, 2000; Snook, 2000; Sunstein & Thaler, 2008; Woods

et al., 2010).

It should be noted that even though the cases used to exemplify each category of the new typology

all resulted in a fatality, neither death nor injury is required to meet the definition of a police shooting

error. In fact, most police shooting errors occur when a bullet does not strike a human target at all. It

should also be noted that police shooting errors have significant officer safety implications in addition

to the more obvious public safety concerns. Shooting at but missing a subject who is intent on killing

an officer or another person could be catastrophic. In addition, all accidental blue-on-blue shootings—

incidents in which an officer shoots another police officer—fit neatly into the new typology. Under-

standing how and why these errors occur and under what circumstances they manifest themselves could

be used to improve outcomes for both the police and the public they serve. The remainder of this article

will be used to examine each category of the new typology.

6.1 Misdiagnosis errors Intended Firearm Discharge + Intended Target Hit = Unintended Outcome (Error)

A misdiagnosis error occurs when a police officer intends to discharge a firearm and hits the intended

target of his behavior and the result is an unintended outcome. This type of error would fall into

Reason’s (1990) rule-based errors and Scharf and Binder’s (1983) false-positive category of police

shooting errors. Police practitioners have a variety of names for these types of shootings, including

“cellphone shootings” (Police Executive Research Foundation [PERF], 2012, p. 8), “mistake-of-fact”

shootings (Fachner et al., 2015, p. 30), and “perception-only” shootings (Los Angeles Police Depart-

ment [LAPD], 2018).

An empirical example of a misdiagnosis shooting, from the four cases outlined earlier, is the shooting

of Francisco Serna. The police were dispatched to the call of a man with a gun. When they arrived, the

police confronted a man who matched the dispatched description of the person with a gun and had his

hands in his pockets. When the man rapidly removed one of his hands from his pocket with an object

in it, the police shot and killed him only to discover the object was a crucifix (Washington Post Staff,

2016). How often do these type of shootings occur?

At a PERF conference in 2012, then Los Angeles County Assistant Sheriff Cecil Rhambo said, “In

Los Angeles County we have 5 to 15 [police] shootings a year due to what we call perception issues”

(p. 8). Between 2013 and 2017, 14% (n = 30) of the shooting incidents reported by the Los Angeles

Police Department (N = 211) could be classified as misdiagnosis errors (LAPD, 2018). Fachner et al.

(2015) found similar numbers for the Philadelphia Police Department. Between 2007 and 2013, 10%

TAYLOR 815

(n = 35) of the shooting cases examined (N = 356) would fall under the misdiagnosis category. It

should be noted that both sets of numbers represent incident rather than individual bullet-level data. It

is likely that more than one officer fired and that individual officers fired more than one time in some

of the incidents captured in the LAPD and Philadelphia data. If these data are at all representative of

the prevalence of this type of police shooting error, the percentage of misdiagnosis errors is not trivial.

But why do they occur?

Although it is beyond the scope of this article to test the individual categories within the typology,

the research on heuristics and biases (e.g., Kahneman, 2011) and recognition-primed decision-making

(e.g., Klein, 2011) hold great promise for examining this type of error. Experience in any practical

domain allows for practitioners to develop patterns of key information within their realm of expertise.

They can then use these patterns or mental models to evaluate situations quickly and to act with less-

than-perfect information by systematically focusing on what is important while ignoring what is not

(Klein, 2011).

The police are no different (e.g., Stalans & Finn, 1995), and the classic observational studies of the

police are brimming with references to this type of rapid pattern assessment followed by decisive and

often consequential action. James Q. Wilson (1968, pp. 38–39) wrote, “[O]fficers are routinely called

upon to ‘prejudge’ persons by making quick decisions about what their behavior has been in the past

or is likely to be in the future.” He noted that officers seemed to be particularly attuned to two types

of cues: “those that signal danger and those that signal impropriety.” Jerome Skolnick (1967, p. 45)

postulated:

The policeman, because his work requires him to be occupied continually with potential violence, develops a perceptual shorthand to identify certain kinds of people as symbolic assailants, that is, as persons who use gesture, language, and attire that the policeman has come to recognize as a prelude to violence.

William Muir (1977, p. 153) called this process “pigeonholing” and wrote:

To anticipate what was going to happen, policemen developed a sense for the patterns in human affairs. They formed concepts, or classifications, which helped them to assim- ilate and distinguish between discrete persons and events. Concepts were attended by visual procedures by which policemen processed the details of the moment into these abstractions.

Writing specifically about police-involved shootings, David Klinger (2004, p. 83) noted that:

The line separating close calls from shootings is razor thin … when officers do shoot, it is because something – the way armed individuals stand, the way they hold their weapons, the way they move, the words they speak, the look on their faces, some cue – tells them that this moment is different.

Although scholars in the policing literature have yet to adopt a cohesive term for the type of decision-

making described earlier, scholars in the psychological and behavioral economics literature would call

them “heuristics” and “biases.” Daniel Kahneman (2011, p. 98), defined heuristics as “simple [mental]

procedures that help find adequate, though often imperfect, answers to difficult questions.” As noted

earlier, there are upsides and downsides to this type of implicit pattern-based decision-making. On

the down side, because they shortcut much of the available information to reach a suitable answer

816 TAYLOR

quickly, heuristics can and regularly do result in error. If a person attends to the wrong information and

ignores or mistakenly interprets the right information, heuristics can lead to systematic and predictable

errors (Reason, 1990). In a randomized, controlled experiment, using a video simulator, audio dispatch

tapes, and a sample of more than 300 active law enforcement officers, Taylor (2016) found that the

percentage of misdiagnosis shooting errors—shooting a person who rapidly produced a cellphone from

his pocket—increased more than 100% from the control when officers received pre-event dispatch

information that the subject they would be dealing with appeared to have a gun.

6.2 Misapplication errors Unintended Firearm Discharge + Intended Target Hit = Unintended Outcome (Error)

A misapplication error occurs when a police officer does not intend to discharge a firearm but does

discharge a firearm, the bullet hits the intended target of the officer’s behavior, and the result is an

unintended outcome. This type of error falls under Reason’s (1990) skill-based errors of slips and

lapses. Practitioners call these shootings “weapon confusion shootings” (Americans for Effective Law

Enforcement [AELE], 2012) or “TASER-confusion shootings” (Martin, 2016). Although these types

of shooting errors are primarily associated with electronic-control devices like TASER, they may also

occur with flashlights and other tools an officer might activate during a potential incident.

An empirical example of a misapplication shooting, from the four cases outlined earlier, is the shoot-

ing of Oscar Grant. An officer who was struggling with Grant on the ground announced to fellow

officers that he was going to “Tase” Grant, stood up, drew his firearm, and shot Grant one time in

the back killing him (State of California v. Johannes Mehserle, 2010). During the subsequent trial of

the involved officer, experts who had analyzed the video evidence from the shooting noted that the

officer’s firearm did not have a thumb-activated safety mechanism like the TASER does and that the

officer can be seen making a motion with his thumb as he drew his firearm, which is consistent with

the motion that a person would make to deactivate the safety on a TASER (Martin, 2016). As of 2012,

there were 13 documented cases of misapplication shooting errors involving electronic-control devices

like TASER (AELE, 2012).

This type of slip, or unintended action, is well documented in the psychological literature (e.g., Nor-

man, 1981). According to Reason (1990), “A necessary condition for the occurrence of a slip of action

is the presence of attentional ‘capture’ associated with either distraction or preoccupation” (p. 56).

“Slip-and-capture” errors are often associated with switching to a new tool or process after significant

training and/or experience with an ergonomically similar but functionally different tool or process. A

benign example would be after years of driving a vehicle with an ignition that requires a key, a person

switches to a vehicle with a keyless ignition. If the person’s attention was on something else as he or

she attempted to start the car, the person might attempt to insert the key fob into the steering column

and this would be classified as a slip-and-capture error. These types of errors are well documented in

the fields of medicine (e.g., Zhang, Patel, Johnson, & Shortliffe, 2004) and aviation (e.g., Nagel, 1988)

and have led to accidental fatalities in both professions.

In talking about the risk for these types of errors, Karl Weick (1985, p. 40) said:

[S]kills trained just to the point of sufficiency may be potentially the most dangerous, since trainees, trainers and [administrators] alike assume the skill is available. In real- ity, that skill will be one of the first to disappear under pressure. It will be replaced by a much more primitive action that has been practiced for a much longer period of time.

TAYLOR 817

When it comes to firearms and TASER training for police officers, there seems to be just such a training

discrepancy. New police officers spend an average of 60 hours on firearms and shooting skills during

their initial academy training (Morrison, 2006b). After the academy, firearms training continues to be

emphasized throughout an officer’s career through in-service training, often on an annual or quarterly

basis. TASER certification, on the other hand, requires officers to fire two practice rounds during the

initial certification and two practice rounds for recertification every 2 years (Axon, 2018). From this

perspective, if an officer’s attentional resources were fully engaged—captured—by a struggle with a

suspect and, intending to draw an electronic-control device, the officer unintentionally drew a firearm

and shot the suspect instead, then this type of error might also be framed in terms of the frequency-

gambling heuristic (Kahneman, 2011). Under time constraints and uncertainty, with System 2 fully

engaged in other things, System 1 may be prone to go with the most frequent associated behavioral

response. Regardless, additional research is certainly needed.

6.3 Misses Intended Firearm Discharge + Unintended Target Hit = Unintended Outcome (Error)

A miss occurs when a police officer intends to discharge a firearm but does not hit the intended target

of the officer’s behavior and the result is an unintended outcome. This type of error falls into Reason’s

(1990) skill-based errors. An empirical example of a miss, from the cases described earlier, is the Bettie

Jones shooting. A Chicago police officer, responding to the report of a domestic disturbance, shot and

killed a man who was attacking him with a baseball bat but also unintentionally shot and killed Bettie

Jones (Johnson, 2018).

Unlike the other categories of this typology, a significant body of literature has been produced on

police shooting accuracy in the field. The predominant finding from this body of research is that hit

ratios are very low and that they have not significantly changed over the years (e.g., Donner & Popovich,

2018; Geller & Karales, 1981; Geller & Scott, 1992; Landman, Nieuwenhuys, & Oudejans, 2016;

White, 2006). Despite variance in the literature, departments rarely achieve hit rates above 50% and

most fall well below (Copay & Charles, 2001). In fact, Morrison (2006a) wrote, “[H]it rates hovering

around one in five shots have persisted from the earliest measurements in the 1970s through the 1990s,

this despite the many changes in such training during the period” (p. 332). Researchers examining

364 officer-involved shootings by the Philadelphia Police Department between 2007 and 2013 found

a police hit rate of 18% (Frachner et al., 2015). The Los Angeles Police Department self-reported that

between 2013 and 2017 its officers fired a total of 1,900 rounds in the field and 509 of those rounds hit

their intended target for an average hit rate of 27%. This ranged from a low in 2017 of 18% to a high

in 2016 of 42% (LAPD, 2018).

Regardless of the variance and reporting issues noted by scholars (e.g., Donner & Popovich, 2018),

it is pretty clear that, as Geller and Scott (1992) so elegantly put it, “The numbers of wounded

and slain criminal suspects in the United States pale by comparison to the numbers shot at but

missed by police” (p. 100). Even with the highest recorded hit ratios from the research that has

been done—approximately 50%—when it is combined with the other error categories from the typol-

ogy, we find that error is a much more common outcome than correct action when it comes to

police shootings. This type of finding has rarely if ever been documented in the professional envi-

ronment (e.g., Dekker, 2010, 2014; Reason, 1990; Woods et al., 2010), and significantly more research

needs to be done on both the suitability and the transferability of current police firearms training

programs.

818 TAYLOR

6.4 Unintentional discharges Unintended Firearm Discharge + Unintended Target Hit = Unintended Outcome (Error)

An unintentional discharge occurs when a police officer does not intend to discharge a firearm but

a firearm discharges and the bullet hits an unintended target of the officer’s behavior and the result

is an unintended outcome. This type of error also falls under Reason’s (1990) skill-based errors. In

addition to unintentional discharges, practitioners also call these shootings “accidental” or “negligent”

discharges (O’Neill, O’Neill, & Lewinski, 2017).

The empirical example of this type of error, from the cases described earlier, is the shooting of Akai

Gurley. A newer officer who was walking up a poorly lit high-rise apartment stairwell with his gun

drawn, apparently an informal department practice for New York City Police Department (NYPD)

officers in this particular building, accidently discharged his weapon. The bullet ricocheted off a wall

before striking and killing Gurley who had just entered the stairwell at a lower level (Wilson, 2014).

Between 2013 and 2017, the LAPD reported 44 unintentional discharges—approximately 17% of its

reported shooting incidents (LAPD, 2018). Between 2006 and 2016, the NYPD reported 195 unin-

tentional discharges out of 1,037 shooting incidents or approximately 19% of its reported shooting

incidents (NYPD, 2017).

Heim, Schmidtbleicher, and Niebergall (2006a) found that 20% of officers touched the trigger of a

drawn firearm during experimental scenarios when it was inappropriate for them to shoot and none of

the officers in the study remembered doing it when they were interviewed after the fact. In a follow-

up study, Heim, Schmidtbleicher, and Niebergall (2006b) found that various body movements like

jumping, kicking, or grabbing onto objects could cause enough sympathetic pressure to be exerted by

the trigger finger to cause an unintentional discharge. This type of muscle co-activation was found to

be the cause of approximately 24% of the unintentional discharges in two separate studies (n = 137 and

n = 171) that analyzed official police reports of the incidents (O’Neill, O’Neill et al., 2017; O’Neill,

Hartman, O’Neill, & Lewinski, 2018). O’Neill, O’Neill et al. (2017) and O’Neill et al. (2018) also found

that approximately 59% and 47%, respectively, of the unintentional discharges they analyzed occurred

during “routine firearms tasks” (i.e., cleaning, putting the gun away at the end of shift, placing the

gun in a locker when entering a courthouse or jail, and loading and unloading) and that approximately

11% and 13%, respectively, occurred during the use of an unfamiliar firearm or holster. These are

fascinating findings from the perspective of Kahneman and Tversky’s (1979) dual-process model and

Reason’s (1990) levels of performance.

As people gain experience with a task or problem, the tendency is to move from effortful cognitive

processing to automaticity. Cognitive automaticity – System 1 processing – allows people to devote

attentional resources to things outside the task or problem they are working on, but it is not likely opti-

mal for tasks or problems involving a loaded firearm and, according to Reason (1990), will inevitably

result in error.

7 DISCUSSION AND CONCLUSION

The purpose of this article has been to present a typology of police shooting errors and thereby to

provide a new framework for the discussion, research, and understanding of these tragic outcomes.

Each of the example cases discussed in this article and all of the shootings they exemplify could be,

and indeed often are, for purposes of research, indiscriminately aggregated with all other available

cases that fit under the heading of “police use of deadly force” and then statistically analyzed for

TAYLOR 819

causal correlations. Yet the underlying causal mechanisms for each of these cases are much different

and would be missed using such a coarse approach.

It would be equally problematic to treat each of these cases as independent and isolated events. We

understand that people with similar training and experience tend to respond to similar situations in

similar ways and that situations that result in error tend to result in repeated errors across time and

people. This means that where we find professional error, we find an opportunity to improve practice

and reduce the potential for future errors. In the case of officer-involved shootings, where error occurs

more often than correct action, this would mean reducing some of the most visible and controversial

shootings along with their impact on the public’s perception of police competence and legitimacy.

Because error and correct action are so often “two sides of the same coin” when it comes to the

cognitive processes used to reach the ultimate outcome, it is critical that we understand how police

practitioners make decisions in the midst of the uncertain and time-compressed situations they face

(Reason, 1990, p. 2). The errors that police officers systematically make in their decisions to use deadly

force and in their handling of firearms provide a window into both police practice and the cognitive

decision-making strategies officers employ. If, as Peter Manning (1992) wrote, “[T]he core technology

of the police is situated decision making with the potential for the application of violence” (p. 354), then

understanding these decision-making strategies, whether they are explicit or implicit, is fundamental

to understanding the police themselves. Given the risk, uncertainty, and speed with which these types

of decisions have to be made in the field and the body of literature noting the pattern-based nature of

police decision-making in general, a research framework in which heuristics and biases are employed

from behavioral economics and the psychological sciences and/or Klein’s (2011) recognition primed

decision model would likely be fruitful, particularly when like cases are used as either the foundation

for research questions or the subjects of the research themselves.

Ultimately, as Woods et al. (2010) pointed out, it is not enough to search for and identify human

error in the workplace. Instead, the goal of accident- or error-related research should always be to

learn from error, reduce complexity, and “engineer resilience” into the decision environment for front-

end operators (p. 83). By providing a framework for more precise research while fostering a systems

perspective on particular types of error, the hope for this typology is that it will be used to improve both

the research on police use of deadly force and the practitioner policy, training, and practice surrounding

these dynamic and often controversial encounters.

REFERENCES

Alpert, G. P., & Fridell, L. A. (1992). Police vehicles and firearms: Instruments of deadly force. Prospect Heights:

Waveland Press.

Americans for Effective Law Enforcement (AELE). (2012). Weapon confusion and civil liability. AELE Monthly Law Journal, 6(101), 101–111.

Axon. (2018). TASER training: TASER X26P user course manual (Version 20.2).

Bailey, K. D. (1994). Typologies and taxonomies: An introduction to classification techniques (Vol. 102). Thousand

Oaks: Sage.

Bittner, E. (1990). Aspects of police work. Boston: Northeastern University Press.

Chabris, C. F., & Simons, D. J. (2010). The invisible gorilla: And other ways our intuitions deceive us. New York: Crown.

Collier, D., LaPorte, J., & Seawright, J. (2012). Putting typologies to work: Concept formation, measurement, and analytic

rigor. Political Research Quarterly, 65(1), 217–232.

Copay, A. G., & Charles, M. T. (2001). Handgun shooting accuracy in low light conditions: The impact of night sights.

Policing: International Journal of Police Strategies & Management, 24(4), 595–604.

Dekker, S. W. A. (2010). The field guide to human error investigations. Burlington: Ashgate.

Dekker, S. W. A. (2014). The field guide to understanding ‘human error’. Burlington: Ashgate.

820 TAYLOR

Donner, C. M., & Popovich, N. (2018). Hitting (or missing) the mark: An examination of police shooting accuracy in

officer-involved shooting incidents. Policing: An International Journal, 42(3), 474–489.

Doyle, J. M. (2010). Learning from errors in American criminal justice. The Journal of Criminal Law and Criminology,

100(1), 109–147.

Fachner, G., & Carter, S. & Collaborative Reform Initiative. (2015). An assessment of deadly force in the Philadelphia police department (pp. 1–173). Washington: Office of Community Oriented Policing Services.

Fyfe, J. J. (1978). Shots fired: An examination of New York City police firearms discharges (No. 78–14335 UMI). State

University of New York at Albany.

Fyfe, J. J. (1979). Administrative interventions on police shooting discretion: An empirical examination. Journal of Criminal Justice, 7(4), 309–323.

Fyfe, J. J. (1988). Police use of deadly force: Research and reform. Justice Quarterly, 5, 165–205.

Gau, J. M. (2014). Procedural justice and police legitimacy: A test of measurement and structure. American Journal of Criminal Justice, 39(2), 187–205.

Geller, W. A., & Karales, K. J. (1981). Split-second decisions: Shootings of & by Chicago police. Chicago: Chicago Law

Enforcement Study Group.

Geller, W. A., & Scott, M. (1992). Deadly force: What we know: A practitioner’s desk reference on police-involved shootings. Washington: Police Executive Research Forum.

Green, M. (2017). Roadway human factors: From science to application.Tucson: Lawyers & Judges.

Heim, C., Schmidtbleicher, D., & Niebergall, E. (2006a). The risk of involuntary firearms discharge. Human Factors,

48(3), 413–421.

Heim, C., Schmidtbleicher, D., & Niebergall, E. (2006b). Towards an understanding of involuntary firearms discharges:

Possible risks and implications for training. Policing: An International Journal of Police Strategies & Management, 29(3), 434–450.

Hollway, J. (2014). A systems approach to error reduction in criminal justice. Penn Law Public Law and Legal Theory Research Paper Series, 14(6), 3–31.

Institute of Medicine. (2000). To err is human: Building a safer health system. Washington: National Academy Press.

International Association for Chiefs of Police (IACP). (2012). Officer-involved shootings, in-custody deaths and serious

uses of force. In Model Policy. Washington: Author.

Johnson, E. T. (2018). Log #1078616: Non-concurrence with findings and penalty of Officer Robert Rialmo (Letter to

Patricia Banks: Interim Chief Administrator of the Civilian Office of Police Accountability). Chicago: Chicago Police

Department.

Kahneman, D. (2011). Thinking, fast and slow. New York: Farrar, Straus & Giroux.

Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrics, 47, 263–291.

Kania, R. R., & Mackey, W. C. (1977). Police violence as a function of community characteristics. Criminology, 15 (1),

27–48.

Klein, G. A. (2011). Streetlights and shadows: Searching for the keys to adaptive decision making. Cambridge: The MIT

Press.

Klinger, D. (2004). Into the kill zone: A cop’s eye view of deadly force. San Francisco: Jossey-Bass.

Klinger, D. (2012). On the problems and promise of research on lethal police violence: A research note. Homicide Studies,

16, 78–96.

Klinger, D., Rosenfeld, R., Isom, D., & Deckard, M. (2015). Race, crime, and the micro-ecology of deadly force. Crim- inology & Public Policy, 15(1), 193–222.

Landman, A., Nieuwenhuys, A., & Oudejans, R. R. D. (2016). Decision-related action orientation predicts police officers’

shooting performance under pressure. Anxiety, Stress, & Coping, 29(5), 570–579.

Los Angeles Police Department (LAPD). (2018). Use of force year-end review. Los Angeles: Author.

Lipsky, M. (2010). Street corner bureaucracy. New York: Russell Sage Foundation.

Mach, E. (1975). Knowledge and error (English translation). Dordrecht, the Netherlands: Reidel. (Original work pub-

lished 1905).

Manning, P. K. (1992). Information technologies and the police. Crime and Justice, 15, 349–398.

Martin, J. A. (2016). Applied human error theory: A police TASER-confusion shooting case study. Proceedings of the Human Factors and Ergonomics Society 2016 Annual Meeting, 475–479.

Morrison, G. B. (2006a). Deadly force programs among larger US police departments. Police Quarterly, 9(3), 331–360.

TAYLOR 821

Morrison, G. B. (2006b). Police department and instructor perspectives on pre-service firearm and deadly force training.

Policing: An International Journal of Police Strategies & Management, 29(2), 226–245.

Muir, W. K. (1977). Police: Street corner politicians. Chicago: The University of Chicago Press.

Nagel, D. C. (1988). Human error in aviation operations. In E. L. Weiner, D. C. Nagel, M. P. Friedman, & E. C. Carterette

(Eds.), Human factors in aviation (pp. 263–303). New York: Academic Press.

New York Police Department (NYPD). (2017). Annual use-of-force report 2016. Retrieved from https://www1.nyc.gov/

assets/nypd/downloads/pdf/use-of-force/use-of-force-2016.pdf

Nix, J., Campbell, B. A., Byers, E. H., & Alpert, G. P. (2017). A bird’s eye view of civilians killed by police in 2015:

Further evidence of implicit bias. Criminology & Public Policy, 16(1), 309–340.

Norman, D. A. (1981). Categorization of action slips. Psychological review, 88(1), 1–15.

O’Neill, J., Hartman, M. E., O’Neill, D. A., & Lewinski, W. J. (2018). Further analysis of the unintentional discharge of

firearms in law enforcement. Applied ergonomics, 68, 267–272.

O’Neill, J., O’Neill, D. A., & Lewinski, W. J. (2017). Toward a taxonomy of the unintentional discharge of firearms in

law enforcement. Applied Ergonomics, 59, 283–292.

Police Executive Research Forum (PERF). (2012). An integrated approach to de-escalation and minimizing use of force.

Washington: Author.

Pickering, J. C., & Klinger, D. A. (2016). Enhancing police legitimacy by promoting safety culture. In M. Deflem (Ed.),

The politics of policing: Between force and legitimacy (pp. 21–39). Bingley, England: Emerald Group.

Rasmussen, J., & Jensen, A. (1974). Mental procedures in real-life tasks: A case study of electronic troubleshooting.

Ergonomics, 17, 293–307.

Reason, J. (1990). Human error. Cambridge, England: Cambridge University Press.

Reason, J. (2008). The human contribution: Unsafe acts, accidents, and heroic recoveries. Burlington: Ashgate.

Ross, C. T. (2015). A multi-level Bayesian analysis of racial bias in police shootings at the county-level in the United

States, 2011–2014. PLoS One, 10(11), e0141854.

Scharf, P., & Binder, A. (1983). The badge and the bullet: Police use of deadly force. New York: Praeger.

Schmidt, R. A., & Lee, T. D. (2014). Motor learning and performance: From principle to application. Champaign:

Human Kinetics.

Shane, J. M. (2013). Learning from error in policing: A case study in organizational accident theory. New York: Springer.

Shane, J. M., & Swenson, Z. (2018). Unarmed and dangerous: Patterns of threats by citizens during deadly force encoun- ters with police. New York: Routledge.

Sherman, L. W. (1980). Perspectives on police and violence. The ANNALS of the American Academy of Political and Social Science, 452(1), 1–12.

Sherman, L. W. (2018). Reducing fatal police shootings as system crashes: Research, theory, and practice. Annual Review of Criminology, 1, 421–449.

Sherman, L. W., & Langworthy, R. H. (1979). Measuring homicide by police officers. The Journal of Criminal Law & Criminology, 70(4), 546–560.

Skolnick, J. H. (1966). Justice without trial: Law enforcement in democratic society. New York: Wiley.

Skolnick, J. H., & Fyfe, J. J. (1993). Above the law: Police and the excessive use-of-force. New York: Free Press.

Snook, S. A. (2002). Friendly fire: The accidental shootdown of US Black Hawks over Northern Iraq. Princeton: Prince-

ton University Press.

Stalans, L. J., & Finn, M. A. (1995). How novice and experienced officers interpret wife assaults: Normative and effi-

ciency frames. Law & Society Review, 29(2), 287–321.

State of California v. Johannes Mehserle, Superior Court, County of Los Angeles, No. 161210. Testimony of: Mehserle,

Johannes (June 24 & 25, 2010).

Stinson, P. M. (2015). Police shootings: A new problem or business as usual. Criminal Justice Faculty Publications,

51, 1.

Sunstein, C., & Thaler, R. (2008). Nudge: Improving decisions about health, wealth, and happiness. New Haven: Yale

University Press.

Taylor, P. L. (2016). Dispatch priming and the police decision to use deadly force. Paper presented at the American

Society of Criminology Conference, New Orleans, LA.

Van Maanen, J. J. E. (1973). Working the street: A developmental view of police behavior (MIT Working Paper). Cam-

bridge: MIT Press.

822 TAYLOR

Von Bertalanffy, L. (1973). General system theory: Foundations, development, applications. New York: George

Braziller.

Washington Post Staff. (2016). Fatal force: 963. The Washington Post, Retrieved from https://www.washing

tonpost.com/graphics/national/police-shootings-2016/

Weick, K. E. (1985). A stress analysis of future battlefields. In J. G. Hunt & J. D. Blair (Eds.), Leadership on future battlefield (pp. 32–46). Washington: Pergamon-Brassey’s International Defense Publishers.

White, M. D. (2006). Hitting the target (or not): Comparing characteristics of fatal, injurious, and noninjurious police

shootings. Police Quarterly, 9(3), 303–330.

White, M. D. (2015). Transactional encounters, crisis-driven reform, and the potential for a national police deadly force

database. Criminology & Public Policy, 15(1), 223–235.

Wilson, J. Q. (1978). Varieties of police behavior: The management of law and order in eight communities. Boston:

Harvard University Press.

Wilson, M. (2014). Officer’s errant shot kills unarmed Brooklyn man. The New York Time, Retrieved from

https://www.nytimes.com/2014/11/22/nyregion/new-york-police-officer-fatally-shoots-brooklyn-man.html?_r=0

Woods, D. D., Dekker, S., Cook, R., Johannesen, L., & Sarter, N. (2010). Behind human error. Burlington: Ashgate.

Zhang, J., Patel, V. L., Johnson, T. R., & Shortliffe, E. H. (2004). A cognitive taxonomy of medical errors. Journal of Biomedical Informatics, 37(3), 193–204.

Zimring, F. E. (2017). When police kill. Cambridge: Harvard University Press.

AUTHOR BIOGRAPHY

Paul L. Taylor is a Ph.D. candidate in the School of Criminal Justice at the University at Albany,

State University of New York. His research interests are focused on situated decision-making within

the context of police use of force.

How to cite this article: Taylor PL. Beyond false positives: A typology of police shooting

errors. Criminol Public Policy. 2019;18:807–822. https://doi.org/10.1111/1745-9133.12460

Copyright of Criminology & Public Policy is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.