Content uploaded by Michael White
Author content
All content in this area was uploaded by Michael White on Apr 05, 2022
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=rjqy20
Justice Quarterly
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/rjqy20
Investigating the Impacts of a Global Pandemic
and George Floyd’s Death on Crime and Other
Features of Police Work
Michael D. White, Carlena Orosco & Brice Terpstra
To cite this article: Michael D. White, Carlena Orosco & Brice Terpstra (2022): Investigating the
Impacts of a Global Pandemic and George Floyd’s Death on Crime and Other Features of Police
Work, Justice Quarterly, DOI: 10.1080/07418825.2021.2022740
To link to this article: https://doi.org/10.1080/07418825.2021.2022740
View supplementary material
Published online: 10 Mar 2022.
Submit your article to this journal
View related articles
View Crossmark data
Investigating the Impacts of a Global Pandemic and
George Floyd’s Death on Crime and Other Features of
Police Work
Michael D. White , Carlena Orosco and Brice Terpstra
School of Criminology and Criminal Justice, Arizona State University, Phoenix, AZ, USA
ABSTRACT
The current study involves a two-level examination of the impact
of COVID-19 and George Floyd’s death on police work in the
Tempe (AZ) Police Department. We employ interrupted time ser-
ies analysis to test weekly trends (January 2017–January 2021) in
crime and officer activity measures. We also examine body-worn
camera footage (n¼474) to compare dynamics of police-citizen
encounters over time. The impact on crime was mixed and varied
by crime type. When declines occurred, they were short-term. The
pandemic altered every aspect of police activity, and George
Floyd’s death led to separate impacts on a few of those measures.
Overall, the pandemic had a larger impact than did George
Floyd’s death. Routine activities theory proved to be a useful
framework for understanding the findings. Body-worn camera
footage highlighted the consistency in how officers handled citi-
zen encounters. The findings underscore the importance of local-
ized context for these events and their effects.
ARTICLE HISTORY
Received 22 July 2021
Accepted 13 December 2021
KEYWORDS
COVID-19 and the police;
global pandemic and
the police
Introduction
The COVID-19 pandemic dramatically altered many aspects of life in the United States,
and its consequences have been both shocking and deadly. Woolf et al. (2020: 133)
stated “by October 2020 COVID-19 had become the third leading cause of death for
persons aged 45 through 84 years and the second leading cause of death for those
aged 85 years or older.”Though vaccinations have ramped up significantly in 2021,
the U.S. crossed the grim threshold of 700,000 COVID-19 deaths on October 1, 2021
(Bosman & Leatherby, 2021).
Police have been at the forefront of the response to the pandemic and its conse-
quences. In many jurisdictions, police took the lead on enforcing stay-at-home orders
(White & Fradella, 2020). Departments quickly had to distribute protective equipment
to reduce risk of infection, and many altered their day-to-day operations, from cancel-
ing training and abandoning community-oriented outreach efforts (National Police
Foundation (NPF), 2020) to cutting back on officer proactivity (e.g. pedestrian stops;
ß2022 Academy of Criminal Justice Sciences
CONTACT Michael D. White mdwhite1@asu.edu
JUSTICE QUARTERLY
https://doi.org/10.1080/07418825.2021.2022740
Lum et al., 2020). Despite these changes, more than 200 police officers died from
COVID-19 in 2020 (Mourtgos & Adams, 2021; ODMP, 2021).
On May 25, 2020, just over two months after then-President Trump declared a
national emergency in response to the pandemic, Minneapolis police officer Derek
Chauvin knelt on George Floyd’s neck for nearly 9 minutes, killing him. Floyd’s death,
along with the police killings of Breonna Taylor (March 13, 2020), Rayshard Brooks
(June 12, 2020), Jacob Blake (August 23, 2020) and others, ignited outrage and unpre-
cedented scrutiny of the police, demonstrated by protests across the globe, riots in
numerous cities, boycotted games by professional athletes, legislative efforts to reform
the police at the local, state and federal level, and even calls to defund the police.
Though the national dialogue over police reform is not new, public outrage reached a
fever-pitch in the summer of 2020 as the police were dealing with the consequences
of COVID-19.
The global pandemic and George Floyd’s death (and the protests that followed)
coincided with a large increase in violent crime. In their report to the National
Commission on COVID-19 and Criminal Justice, Rosenfeld and Lopez (2020: 3) note
that “Homicides, aggravated assaults, and gun assaults rose significantly beginning in
late May and June of 2020.”Several studies also documented notable increases in
rates of domestic violence in the first part of 2020 (Bullinger et al., 2020; Nix &
Richards, 2021; Piquero et al., 2020). On September 27, 2021, the Federal Bureau of
Investigation Lucas, 2021 released crime statistics for 2020. Though property crime
declined by 7.8% (compared to 2019), violent crime increased by 5.6%. The violent
crime trend was driven by a 12.1% increase in aggravated assaults, and a nearly 30%
spike in homicide, which is “the largest single-year increase ever recorded in the coun-
try (Lucas, 2021:1).”
Existing research has failed to capture the degree to which COVID-19 and George
Floyd’s death have jointly affected crime and other features of police work. As a result,
there are unanswered questions regarding how these two unprecedented events may
have intersected to (1) produce a massive increase in violence, and (2) alter the basic
functioning of the police. This research gap has serious theoretical and practical impli-
cations. From a theoretical perspective, routine activities theory (Cohen & Felson,
1979) would suggest a decline in crime following issuance of mandatory stay-at-home
orders in early 2020, as two of the three components of the crime triangle were con-
strained (suitable targets and motivated offenders). Alternatively, that same theory
could explain an increase in domestic violence. A spike in crime and violence after
George Floyd’s death could be explained by an increase in legal cynicism (Kirk &
Matsuda, 2011; Sampson & Bartusch, 1998), especially when combined with frustration
over the government’s handling of the pandemic (Haeder & Gollust, 2020; Sauer et al.,
2021). Routine activities theory would also suggest a crime increase as stay-at-home
orders expired in summer 2020. The practical implications are equally profound, as
these events may have substantially increased risk of victimization and altered the
core functioning of the police in our country. There is a critical need for research to
investigate these issues.
The current study seeks to fill this gap through a two-level examination of police
work in the Tempe, Arizona Police Department. At the department level, we employed
2 M. D. WHITE ET AL.
interrupted time series analysis to investigate weekly trends (2017–2021) in several
features of police business including calls for police service, reported crime (overall
and by type), officer proactivity, traffic accidents, arrests, and use of force. At the inci-
dent level, we examine randomly selected body-worn camera (BWC) footage of 474
incidents involving Tempe officers and citizens. We capture more than 130 incident,
officer, and citizen-related variables for each encounter. At both levels, we are able to
examine and compare data in three time periods: pre-pandemic (before week of
March 8, 2020), pandemic only (weeks of March 8, 2020 - May 17, 2020), and
pandemic þGeorge Floyd’s death (week of May 24, 2020 and after).
1
This two-level,
time-staggered approach allows us to parse out the specific effects of each event, and
to assess their interactive effect on key outcomes. More specifically, we seek to answer
three basic research questions:
1. How did crime and other features of police work in Tempe change with the onset
of the COVID-19 global pandemic?
2. How did crime and other features of police work in Tempe change in the after-
math of George Floyd’s death?
3. To what extent was there an interactive effect among the COVID-19 pandemic
and George Floyd’s death on crime and other features of police work in Tempe?
Literature Review
The Impact of COVID-19 on Police Operations
Police departments across the U.S. have had to adapt policy and practice to meet the
challenges of the global pandemic (Laufs & Waseem, 2020). Many agencies eliminated
in-person roll calls (NPF, 2020), shifted officers from specialized units to patrol (Winton
& Tcheckmedyian, 2020), cancelled community-oriented policing and other outreach
efforts (Lum et al., 2020), and altered work schedules to limit officers’interactions with
each other (Police Executive Research Forum (PERF), 2020). The pandemic also
impacted hiring and training processes, as academy classes were cancelled and in-ser-
vice trainings were suspended (Jennings & Perez, 2020). Additionally, agencies
adopted policies to limit in-person contact between officers and citizens, from online
reporting and telephone responses to non-emergency calls to increased use of social
media (Lum et al., 2020). Some agencies even required their officers to stay in their
patrol vehicles as much as possible during citizen encounters (Hermannet al., 2021).
Many agencies were also pressured to reduce arrests for low-level crimes in an effort
to limit jail populations, where the potential for widespread infection is high
1
We acknowledge there is some subjectivity to the selection of the “break point”for the pandemic. We use March
13, 2020 as the start of the global pandemic as that is the day President Trump declared a national emergency
(Smith-Schoenwalder, 2020). Clearly, there was widespread coverage of COVID-19 prior to that day. There is less
subjectivity for George Floyd’s death. We use May 25, 2020 as that is the day George Floyd was killed. However,
public scrutiny of the police pre-dates Floyd’s death and the repercussions of his death were witnessed over the
months that followed. It is also fair to say that implying these are “events”suggests each is a one-time occurrence.
The COVID pandemic is ongoing and may have had a differential impact on policing over time. The same could be
said for George Floyd’s death and the police reform/defund movement it generated. Fortunately, ARIMA does permit
for the testing of interventions (or events) of varying onsets and durations. To the extent possible, our analyses will
explore alternative starts for the two unprecedented events.
JUSTICE QUARTERLY 3
(Nowotny et al., 2020). All of this occurred while police across the country were tasked
with enforcing mandatory stay-at-home orders and social distancing protocols (White
& Fradella, 2020).
The Impact of COVID-19 on the Tempe, Arizona Police Department
The current study examines the impact of COVID-19 on the Tempe Police Department
(TPD). Like other departments, TPD altered their policy and practice in important ways.
On March 19, 2020, the city closed public facilities, and TPD established notification
procedures for employees experiencing possible exposure, the presence of symptoms,
a positive test result, or direct contact within a designated workgroup. TPD also imple-
mented comprehensive testing and quarantine procedures. Work travel and training
were suspended, and all ride-alongs and department tours were cancelled. TPD also
enacted a tier system which allowed for personnel to be shifted to different assign-
ments as needed, and implemented telecommuting options for those not required to
work in the field. The department also emphasized the use of citations over arrest for
low-level offenses. Last, sworn personnel were directed to use PPE and decontamin-
ation procedures, and to prioritize calls for service in which there was danger to life
and/or property.
The Impact of COVID-19 on Police Business
A handful of studies have examined the impact of COVID-19 on calls for service and
reported crime, and the results are mixed. Researchers have focused specifically on
domestic violence given the nature and duration of stay-at-home orders (Leslie &
Wilson, 2020; Mohler et al., 2020). Piquero et al. (2020) found a short-term spike in
domestic violence calls for service (lasting two weeks) during the early stages of the
stay-at-home orders enforced in Dallas, TX. Mohler et al. (2020) and Bullinger et al.
(2020) also documented increases in domestic violence-related calls for service (in
Indianapolis and Los Angeles, and in Chicago, respectively). Nix and Richards (2021:1)
found an “immediate, significant spike”in domestic violence in five of six jurisdictions
examined, though “as stay-at-home orders were lifted throughout the remainder of
2020, domestic violence calls for service declined in every jurisdiction but Salt
Lake City.”
Early evidence suggested little increase in overall crime throughout the U.S during
the pandemic (Campedelli et al., 2020; Weichselbaum & Li, 2020). For instance, Ashby
(2020) examined calls-for-service data from ten U.S. cities during the initial stages of
the pandemic, and reported relatively steady rates in comparison to previous years.
Other studies have documented sizeable shifts post-pandemic (Bullinger et al., 2020;
Rosenfeld & Lopez, 2020). Abrams (2021) documented immediate decreases in crime
rates, including drug crimes, theft and residential burglary, and a majority of violent
crimes. Lum et al. (2020) conducted a nation-wide survey of 1,141 law enforcement
agencies, and reported an increase in mental distress calls for service among nearly
half of those agencies. The statistics released by the FBI (Lucas, 2021) in late
4 M. D. WHITE ET AL.
September 2021 ended any debate about crime trends: the U.S. experienced a stun-
ning year-to-year increase in violence, especially homicide, in 2020.
The Death of George Floyd
There is a long history of demands for police reform in the U.S., but the last decade
has been especially notable. August 9, 2021 marked the seven-year anniversary of
Michael Brown’s death in Ferguson, Missouri. In 2014–15, the police killings of Brown,
Freddie Gray, and others led to public outrage, riots, and demands for police reform.
In late 2014, President Obama announced the creation of the President’s Task Force
on 21
st
Century Policing. The blue-ribbon commission issued a final report in 2015
with nearly 60 recommendations for improving community trust and enhancing police
accountability. Many police departments across the country adopted this report as a
blueprint for reform (Abramson, 2015).
Flash-forward to May 25, 2020. Officer Derek Chauvin of the Minneapolis Police
Department knelt on George Floyd’s neck for nearly nine minutes, killing him. Floyd’s
horrific death, along with the police killings of Breonna Taylor, Jacob Blake, and others
reignited the national outrage over police use of force. The nation-wide indignation
led to protests in cities across America, calling for local and national police reform
(Taylor, 2021). These events also sparked protests in nations around the world, both in
solidarity with America and in outrage over their own social injustices (Silverstein,
2021). In response to the national and globalized demands for reform, 30 states in the
U.S. and Washington, D.C. have passed police reform legislation (Subramanian & Arzy,
2021). Additionally, federal legislators introduced the George Floyd Justice in Policing
Act of 2020. The House passed the Act in February 2021 but it has stalled in the
Senate since then (as of February 2022).
The Impact of George Floyd’s Death in Tempe, Arizona
Tempe experienced protests in the aftermath of George Floyd’s death. There were citi-
zen protests against police brutality in the city throughout June and July 2020. These
protests culminated in a clash between officers and protestors on July 27, 2020, result-
ing in seven arrests, and an officer injury (KTAR, 2020; Martinez, 2020). In August, TPD
experienced a high-profile incident involving an officer who pointed his firearm at an
unarmed Black hotel employee as the officer was searching for a white suspect. The
case generated significant media attention and public outcry, and led to a $300,000
out-of-court settlement (Blasius, 2020). In the wake of those protests, the Tempe City
Manager accepted Tempe Police Chief Sylvia Moir’s resignation on September 15,
2020, stating “he felt it was in the city’s best interest to hire a new leader to oversee
long-term changes within the Tempe Police Department (Burkitt & Pineda, 2020).”
The Current Study
Few studies have jointly examined the impact of the pandemic and George Floyd’s
death on crime and policing. The critical need for comprehensive study of these
JUSTICE QUARTERLY 5
events is underscored by the nearly 30% increase in homicide in 2020. To what extent
did each of these events contribute to the spike in violence? How did each event alter
other key measures of police work, from officer proactivity to use of force and arrests?
Moreover, there have been no examinations of individual-level interactions between
police and citizens before and during these dramatic events. Given the pressures and
stress caused by these events, were officers using the same amounts of procedural
justice? Were they using the same levels of professionalism?
The current study seeks to fill this gap through a comprehensive examination of
the impact of COVID-19 and George Floyd’s death on the prevalence, nature, and
dynamics of police work in Tempe, Arizona. Two theories guide our research: routine
activities and legal cynicism. Routine activities theory suggests that in order for that
crime to occur there must be the convergence in space and time of three elements:
(1) motivated offenders, (2) suitable targets, and (3) the absence of capable guardians
(Cohen & Felson, 1979). Crime is more likely to occur when a potential offender who
wants to commit a crime converges in time and space with a suitable target (potential
victim), with no guardian present (police). Changes in these factors may increase or
decrease the chances of crime occurring.
Legal cynicism refers to citizens’belief that the legal system (e.g. police) are "illegit-
imate, unresponsive, and ill equipped to ensure public safety (Kirk & Matsuda, 2011).”
Legal cynicism increases when citizens perceive the police abuse their authority in
interactions with citizens (direct or vicarious [George Floyd]; Sampson & Bartusch,
1998). Research has shown a direct effect of legal cynicism on criminal conduct (Kaiser
& Reisig, 2019). Drawing on routine activities and legal cynicism, we test sev-
eral hypotheses:
H1:Crime and department-level measures of police business (calls for service, reported
crime, officer proactivity, use of force, arrests, and traffic accidents) declined signifi-
cantly following the emergence of the COVID-19 pandemic.
H2:Incident-level dynamics between police officers and citizens (use of procedural just-
ice, de-escalation, aggression) became more negative following the emergence of
the COVID-19 pandemic.
H3:Crime and department-level measures of police business (calls for service, reported
crime, officer proactivity, use of force, arrests, and traffic accidents) increased signifi-
cantly following George Floyd’s death.
H4:Incident-level dynamics between police officers and citizens (use of procedural just-
ice, de-escalation, aggression) became more negative following George
Floyd’s death.
Methods and Data
Research Setting
Tempe, Arizona is a suburb located southeast of Phoenix in Maricopa County. With a
geography that spans roughly 40 square miles in size, Tempe is home to the main
campus of Arizona State University (and its approximately 70,000 students) and a bus-
tling entertainment district. Census estimates indicate the resident population is
6 M. D. WHITE ET AL.
approximately 195,805 (U.S. Census Bureau, 2021), with a service population of
268,000 (Tempe Police Department Quick Facts, 2021). TPD consists of approximately
337 sworn employees who are assigned to one of three divisions under the Office of
the Chief (Investigations, Support Staff and Field Operations).
Administrative Data Sources
We extracted data from the records management system (RMS) and other departmen-
tal data repositories from January 2017 through January 2021. This time frame covered
approximately 3 years before the pandemic and 10 months after its onset in the U.S. It
also covers approximately eight months after George Floyd’s death. Each data set is
described below.
Citizen and Officer-Initiated Calls
We extracted calls for service (CFS) in aggregate form to determine raw weekly counts,
and separated the data into citizen-initiated (911 calls) and officer-initiated (proactive
contacts). There were a number of exclusions in the CFS data, specifically calls that
were canceled prior to arrival, test calls, non-interactive call types used for transport or
surveillance, and security checks performed by security units.
Reported Crime
We extracted crime data in raw form to calculate total offenses for the study period.
Using a final offense classification variable to account for any changes made to the
offense type after further investigation, data were then separated into totals per
offense type. Given there are more 100 different crime types, we limit our analyses to
total offenses and NIBRS classifications of crimes against persons, crimes against prop-
erty, and crimes against society.
2
We also examine separately several individual crimes
of interest: assaults (simple and aggravated), burglary (including vehicle burglary), and
domestic violence (which has been a focus of other pandemic-related research). We
do not examine homicides separately because of its low occurrence in Tempe. In
2019, there were 14 homicides. In 2020, there were 9.
Arrests
We also extracted from the RMS distinct counts of arrests using a unique arrest book-
ing number. We limit our analyses to arrest totals.
Use of Force
We gathered use of force data using IAPro, a software that serves as a case manage-
ment system for use of force incidents that occur throughout the department. A
report builder function captured all use of force incidents during the study period,
2
NIBRS classifications differ from UCR. Crimes against persons include: murder, aggravated assault, forcible sex
offenses, non-forcible sex offenses, kidnapping/abduction, simple assault, intimidation. Crimes against property
include: arson, bribery, burglary, counterfeiting/forgery, criminal mischief/damaged property, embezzlement,
extortion, fraud, larceny, theft from motor vehicle, motor vehicle theft, robbery, stolen property. Crimes against
society include: drugs/narcotics violations, gambling, child pornography, prostitution, weapon law violations.
JUSTICE QUARTERLY 7
with the officer as the unit of analysis. We parsed out a total use of force count for
each week.
Traffic Collisions
We extracted collisions from the RMS and included both injury and non-injury acci-
dents that occurred during the study period.
Body-Worn Camera (BWC) Footage
We also extracted and reviewed BWC footage covering the time frame before
President Trump’s national emergency declaration (August 3, 2019–March 12, 2020;
n¼230), during the pandemic only period (March 13, 2020–May 24, 2020; n¼160),
and during the pandemic þGeorge Floyd’s death period (May 25, 2020–July 16, 2020;
n¼84).
3
BWC footage allows for a unique examination of the dynamics of police-citi-
zen encounters before and during these two events. For the BWC review, we randomly
selected 10 officers each week, and for each of those officers, we then randomly
selected a BWC video the officer had uploaded to Evidence.com (the TPD’s cloud-
based storage solution) within the prior seven days. We developed a detailed coding
instrument that collects a wide range of variables regarding the call. For the current
study, we focus on approximately 50 variables capturing incident and demographic
details; officers’use of procedural justice; officers’use of de-escalation tactics; citizen
behavior; citizen drug and alcohol use; citizen mental illness; citizen aggression; officer
use of force, and encounter resolution. One inherent limitation of BWC footage is that
the video does not show the officer (i.e. the officer is not in view of the camera).
Coders recorded information about the dynamics of the encounter based on what the
officer said and did. For example, if the officer asked detailed questions, took time to
listen to the citizen’s perspective, and ensured that the citizen would be in a good
place at the conclusion of the interaction (e.g. had a plan to get home), coders would
record an affirmative for the variable "did the officer appear invested."
Four research assistants coded the interactions, and we rigorously tested accuracy
and consistency of coding using dual coding inter-rater reliability testing. More specif-
ically, a senior coder conducted inter-rater reliability (IRR) testing using a randomly
selected sample of videos coded by each of the other coders. This process continued
for the duration of the study period to ensure that IRR videos were double coded and
assessed on a weekly basis. All low-agreement items were discussed in detail and
inconsistencies were resolved before proceeding. The senior coder conducted IRR test-
ing for 35% of the entire sample of videos.
3
The BWC review includes incidents from a random selection of patrol officers who consented to participate in a
grant-funded project to evaluate de-escalation training. This consent included voluntary participation in all aspects
of the evaluation, including officer surveys, review of administrative data, and review of BWC footage. The BWC data
examined here are the same data collected under the de-escalation project consent, except that we do not separate
by study group and we split the data into the three sub-periods of interest. TPD owns the BWC data (not the
individual officers) and has given permission for its use in this study.
8 M. D. WHITE ET AL.
Analytical Plan
We employ a multi-pronged set of analyses to investigate the impact of COVID-19 and
George Floyd’s death on police business. First, we conducted an aggregate-level
descriptive examination during the study period (January 2017 to January 2021) to
describe trends during the three sub-periods (pre-pandemic, pandemic only,
pandemic þGeorge Floyd) in crime and other police activity measures. Second, we
employed interrupted time series analysis using AutoRegressive Integrated Moving
Average (ARIMA) in SPSS to test whether the weekly count of each measure changed
significantly with the onset of the global pandemic and George Floyd’s death (the
data begin with the week January 1–7, 2017 and end the week January 24–31, 2021).
ARIMA is a quasi-experimental time series design that allows for comparison of pre-
intervention and post-intervention values of an outcome. ARIMA includes a two-stage
process. The first stage, model identification, involves selecting the best-fitting trend
model for the time series, composed of three model components: p,d, and q(refer-
ring to the autoregressive component, the trend component, and the moving average
component of the model, respectively). The second stage involves inclusion of an
intervention variable in the model with a hypothesized onset (abrupt or gradual) and
duration (temporary or permanent; represented as “0”and “1”data points). ARIMA
overcomes several threats to internal validity and violations of the independence
assumption (e.g. serial correlation; McCleary, et al., 2017; McDowall & McCleary, 2014).
Given our hypotheses, we test a very limited set of onsets that include an immediate
impact (week of March 8 for the pandemic; week of May 24 for George Floyd) and
gradual impacts (up to nine months after weeks of March 8 and May 24). We also
tested varying durations from impacts of one month to impacts lasting through the
end of the study period. The syntax, model identification diagnostics, and output for
all ARIMA modeling are included in the Methodological Appendix.
Third, we conduct bivariate analysis of incident-level data collected via examination
of randomly selected BWC footage. Since nearly all of the variables of interest are
dichotomous, we use cross-tabs and chi-square to assess whether the prevalence of
each variable changed significantly from the pre-pandemic (n¼230), pandemic only
(n¼160), and pandemic þGeorge Floyd time periods (n¼84). The only exception is
length of encounter, measured in minutes and seconds. We test for differences in this
measure using an independent samples t-test.
Results
For each outcome, we present both the descriptive examination (figures) and the
ARIMA results. Descriptive results are shown only for the general crime categories (per-
son, property, society) given space constraints.
4
Table 1 shows for each outcome: the
best-fitting model; the onset and duration of both interventions; and the model
4
We do not show graphs of assaults, burglary, robbery, and domestic violence, given space constraints. These are
found in the Methodological Appendix. We do describe the weekly trend and report ARIMA results for these crimes.
All ARIMA syntax and output are reported in the Methodological Appendix.
JUSTICE QUARTERLY 9
diagnostics for each model. Results from the body-worn camera footage analysis are
presented last (Table 2).
Calls for Service and Crime
Citizen-Initiated Calls for Service
Figure 1 shows calls for police service, separated into citizen-initiated and officer-initi-
ated call types. For citizen-initiated calls, the weekly average pre-pandemic (1,673.9)
drops to 1,309.8 during the pandemic only period - a percent-change decrease of
21.8. The average trends upward during the remainder of the study period, though
there are drops in late May and November/December (1,409.9). In Table 1, the best-fit-
ting ARIMA model identifies an immediate statistically significant decline in calls
beginning the week of March 8, 2020 and lasting through the end of the study period.
Table 1. ARIMA results.
Variable Rsquared BStandard Error T(pvalue)
Citizen-Initiated Calls for Service (1,0,0)
Pandemic only: 3/8/20–1/24/21 .718 289.150 30.507 9.478 (.000)
Pandemic þGeorge Floyd: 5/31/20–1/24/21 .686 237.684 50.411 4.715 (.000)
Reported Crime (1,0,0)
Pandemic only: 3/8/20–8/16/20 .386 71.957 8.692 8.279 (.000)
Pandemic þGeorge Floyd: 5/24/20–8/16/20 .280 51.751 14.802 3.496 (.001)
Crimes Against Persons (0,1,1)
Pandemic only: –– – –
Pandemic þGeorge Floyd: –– – –
Simple & Aggravated Assaults (0,1,1)(0,0,1)
Pandemic only: –– – –
Pandemic þGeorge Floyd: –– – –
Crimes Against Property (0,1,1)
Pandemic only: 3/8/20–3/15/20 .358 27.799 6.959 3.995 (.000)
Pandemic þGeorge Floyd: –– – –
Robbery (0,1,1)
Pandemic only: –– – –
Pandemic þGeorge Floyd: –– – –
Burglary (0,1,1)
Pandemic only: 3/8/20 .394 13.847 6.332 2.187 (.030)
Pandemic þGeorge Floyd: –– – –
Crimes Against Society (0,1,1)
Pandemic only: 3/8/20–3/22/20 .320 7.475 1.931 3.872 (.000)
Pandemic þGeorge Floyd: –– – –
Domestic Violence (0,1,1)
Pandemic only: –– – –
Pandemic þGeorge Floyd: –– – –
Officer-Initiated Calls for Service (1,0,0)
Pandemic only: 3/8/20–11/1/20 .607 214.073 46.766 4.578 (.000)
Pandemic þGeorge Floyd: 5/24/20–11/1/20 .601 230.488 51.078 4.512 (.000)
Use of Force (0,1,1)
Pandemic only: 3/8/20 1/24/20 .460 .395 .123 3.200 (.000)
Pandemic þGeorge Floyd: 7/19/20 1/24/21 .467 .719 .202 3.565 (.000)
Arrests: (1,1,0)
Pandemic only: 3/8/20–3/22/20 .305 39.397 6.972 5.651 (.000)
Pandemic þGeorge Floyd: –– – –
Vehicle Accidents (1,0,0)
Pandemic only: 3/22/20–1/24/21 .728 58.921 5.972 9.867 (.000)
Pandemic þGeorge Floyd: 6/7/20–1/24/21 .696 47.949 9.597 4.996 (.000)
10 M. D. WHITE ET AL.
Table 2. Comparison of pre- and post-pandemic police-citizen encounters.
Variable
Pre-pandemic
(n¼230)
%/Mean, N
Pandemic only
(n¼160)
%/Mean, N
Pandemic þGeorge
Floyd (n¼84)
%/Mean, N
Length of encounter (minutes) 15.7 14.112.2
Procedural justice
Officer answered the citizen’s questions 99.3% (147) 97.2% (104) 98% (50)
Office offered solutions to the
citizen’s questions
86.7% (130) 92.1% (93) 94% (47)
Officer appeared invested 93.3% (208) 94.9% (149) 98.8% (82)
Officer explained actions that could be
taken to reach a safe resolution for
all parties
92% (193) 85% (96) 92.3% (72)
Officer spoke to the citizen in a
respectful manner
94.3% (214) 92.9% (145) 98.8% (82)
a
Officer De-escalation
Used friendly physical contact on
the citizen
5.4% (12) 2.6% (4) 8.4% (7)
a
Used friendly language 53.6% (118) 48.1% (75) 39.8% (33)
Explained their decision-making to
the citizen
92% (206) 88.3% (136) 83.6% (67)
Appeared to call for back-up 3.1% (7) 2.6% (5) 0% (0)
Appeared to call a supervisor 2.2% (5) 1.9% (5) 2.4% (2)
Used condescending language or a
patronizing tone when talking to
the citizen
5.7% (13) 5.7% (9) 2.4% (2)
Spoke in a calm manner 94.7% (216) 96.8% (152) 92.8% (77)
Was patient 91.7% (209) 94.2% (147) 95.2% (79)
Appeared empathetic to the citizen 64.2% (136) 35.5% (54) 25.9% (21)
Appeared to compromise with
the citizen
53.2% (67) 44.3% (39) 20.6% (15)
b
Encouraged the citizen that a positive
outcome could be reached
60.7% (125) 42.3% (63) 19.5% (15)
b
Was clear in their commands to
the citizen
100% (142) 100% (131) 97.8% (45)
Appeared to maintain sufficient
personal space between themselves
and the citizen
99.6% (226) 99.3% (146) 97.6% (81)
Appeared to use active listening 91.4% (202) 91.5% (139) 96.3% (78)
Appeared to have a calm
commanding presence
93.4% (213) 94.8% (147) 91.5% (75)
Attempted to build a common ground
with the citizen
77.9% (176) 65.8% (102) 50.6% (42)
a
Attempted to talk through the issue
at hand
92.9% (210) 88.7% (134) 87.5% (70)
Attempted to build rapport 50.7% (115) 41.8% (64) 19.5% (16)
b
Used “we”language 22.6% (51) 23.2% (36) 6.1% (5)
b
Kept a clear head 97.8% (222) 98.7% (154) 98.8% (82)
Controlled their emotions 94.3% (215) 95.5% (148) 97.6% (81)
Transferred control to another officer,
if necessary
53.8% (57) 93.9% (62) 100% (50)
Prioritized officer safety 99.1% (223) 98.5% (135) 100% (83)
Was in sync with other officers
on scene
99.2% (128) 98% (99) 100% (53)
Used time to their advantage 94.1% (209) 96.2% (150) 97.6% (81)
Acted impersonally 11.4% (26) 10.8% (17) 2.4% (2)
a
Used charged/imposing body language 2.7% (6) 2.6% (4) 1.2% (1)
Had a “go go go”attitude 5.7% (13) 5.1% (8) 3.6% (3)
Ignored what the citizen was
communicating
3.1% (7) 2.6% (4) 3.8% (3)
Only asked questions necessary for
a report
42.1% (96) 43.2% (64) 47.4% (36)
(continued)
JUSTICE QUARTERLY 11
Through ARIMA, we also identified a separate statistically significant decline starting
the week of May 31, 2020 (one week after George Floyd’s death) which also lasted
through the end of the study period.
Table 2. (Continued)
Variable
Pre-pandemic
(n¼230)
%/Mean, N
Pandemic only
(n¼160)
%/Mean, N
Pandemic þGeorge
Floyd (n¼84)
%/Mean, N
Unnecessarily yelled at the citizen 12.2% (5) 14.3% (2) 40% (2)
Was reactive, angry, or abrasive 4.8% (11) 5.1% (8) 1.2% (1)
Let ego control their response 2.6% (6) 1.9% (3) 0% (0)
Lost their patience 5.3% (12) 5.8% (9) 4.8% (4)
Pressed the situation unnecessarily 4.6% (10) 2.6% (4) 2.4% (2)
Ignored cues from other officers 0% (0) .99% (1) 0% (0)
Used force when it was possible to use
other approaches
3% (1) 3.3% (1) 0% (0)
Officer used force 5.7% (13) 8.1% (13) 0.0 (0)
b
Citizen drug use 0.9% (2) 5.2% (8)3.7% (3)
Citizen alcohol use 7.9% (18) 8.4% (13) 7.2% (6)
Citizen mental illness 3.2% (7) 7.6% (12)4.8% (4)
Incident resolution
No formal action 66.1% (144) 66% (97) 71.8% (56)
Ticket/citation 11.5% (25) 11.6% (17) 5.1% (4)
Detained/arrested/force used 15.1% (33) 12.2% (18) 14.1% (11)
Medical/other 7.3% (16) 10.2% (15) 9% (7)
Group differences are assessed using cross-tabs and chi-square, with the exception of length of encounter (inde-
pendent samples t-test).
The number of missing cases varies across the data presented above. Percentages are based on the total number of
cases for each variable.
p<.05, p<.01 (pre-pandemic comparison to pandemic only, pandemic þGeorge Floyd).
a
p<.05,
b
p<.01 (pandemic only comparison to pandemic þGeorge Floyd).
Figure 1. Citizen and officer-initiated calls, by week.
12 M. D. WHITE ET AL.
Total Crime
Figure 2 shows that reported crime pre-pandemic was relatively stable (average of
433.0). In the pandemic only period, reported crime dropped substantially to 352 per
week (percent-change of 18.7). There is a notable one-week spike during the week of
May 24 (when George Floyd was killed) and an overall increasing trend during the rest
of 2020 (412.7, percent-change of 17.2). The ARIMA results are consistent. Table 1
shows a best-fitting model with significant immediate drop in reported crime begin-
ning the week of March 8 and lasting through the week of August 16, 2020 (about
5 months). Again, there is a separate statistically significant drop beginning the week
of George Floyd’s death (May 24), also ending the week of August 16, 2020.
Crimes against Persons
Figure 3 shows that crimes against persons averaged 60.1 per week during the pre-
pandemic period. Person crimes drop slightly to 56.0 during the pandemic only period
(percent-change of 6.8), but increased to 64.0 per week after George Floyd’s death
(percent-change of 14.3). Table 1 shows there was no significant change in crimes
against persons associated with the onset of the global pandemic or George
Floyd’s death.
Assaults (Simple and Aggravated)
Assaults (see Methodological Appendix) follow a similar pattern, declining slightly
from the pre-pandemic (49.4) to pandemic only (44.8) periods (percent-change of 9.3),
followed by an increase after George Floyd’s death (51.2, percent-change of 14.3).
Table 1 shows there was no significant change in simple and aggravated assaults asso-
ciated with the global pandemic or George Floyd’s death.
Figure 2. Reported crime by week.
JUSTICE QUARTERLY 13
Crimes against Property
Figure 4 shows the weekly trend for crimes against property. Pre-pandemic, the city of
Tempe averaged 212.8 property crimes per week, and that average dropped to 178.1
during the pandemic only period (percent-change of 16.3). From May 24 through the
end of the study period, the weekly average increased to pre-pandemic levels (205.6,
percent-change of 15.4). ARIMA results in Table 1 indicate that the decline in crimes
against property coincided with the COVID-19 pandemic and is statistically significant,
beginning the week of March 8, 2020 but lasting only two weeks. Table 1 shows there
was no significant change in property crimes associated with George Floyd’s death.
Figure 3. Crimes against persons by week.
Figure 4. Crimes against property by week.
14 M. D. WHITE ET AL.
Robbery
Robberies (see Methodological Appendix) remained relatively stable throughout the
study period, as the weekly average changed little during the pre-pandemic (4.1), pan-
demic only (3.6), and pandemic þGeorge Floyd (4.1) periods. Table 1 shows there was
no significant change in robbery associated with the onset of the pandemic or George
Floyd’s death.
Burglary
During the pandemic only period, burglaries (see Methodological Appendix) declined
by 19.3% (from 54.4 pre-pandemic to 43.9), but the weekly average then increased by
12.3% following George Floyd’s death (49.3). ARIMA results in Table 1 show a best-fit-
ting model with a statistically significant decline occurring the week of March 8, 2020
but lasting just one week. There was no significant change in burglary associated with
George Floyd’s death.
Crimes against Society
Figure 5 shows a significant decline in crimes against society during the pandemic
only period (27.9%, from 42.3 pre-pandemic to 30.5), followed by an increase after
May 24, 2020 (average of 34.7, a percent-change of 13.8%). There is a considerable
drop in crimes against society near the end of the study period. ARIMA results are
consistent: Table 1 shows a best-fitting model with a statistically significant decline
coinciding with the onset of the pandemic (March 8, 2020) and lasting for three weeks
(through week of March 22, 2020). There was no significant change in crimes against
society associated with George Floyd’s death.
Figure 5. Crimes against society by week.
JUSTICE QUARTERLY 15
Domestic Violence
Domestic violence (see Methodological Appendix) increased slightly from pre-pan-
demic to pandemic only (23.5 to 25.0, a percent-change of 6.4), and increases again
after George Floyd’s death (27.3, a percent-change of 9.2). Table 1 shows there was no
change in domestic violence associated with the onset of the pandemic or George
Floyd’s death.
Officer Activity
Officer-Initiated Calls for Service
Figure 1 shows weekly officer-initiated calls (i.e. proactivity), and there is a sharp
decline as the pandemic sets in (from 901.7 per week to 805.5, percent-change of
10.7). There is an immediate, large decline in the weeks following George Floyd’s
death (percent-change of 42.6) though officer-initiated activity increases throughout
the rest of 2020. ARIMA results confirm this visual inspection. Table 1 shows a best-fit-
ting pandemic only model with a statistically significant decline beginning the week
of March 8, 2020 that persists through the week of November 1, 2020. There is also a
separate statistically significant decline beginning the week of May 24, 2020, also last-
ing through November 1, 2020.
Arrests
Weekly arrests decline precipitously once the global pandemic starts (by 45.5%; see
Figure 6). Weekly arrests increase slightly after April 2020 but remain well below pre-
pandemic totals during the remainder of the study period. ARIMA results are again
consistent: Table 1 shows a model with a statistically significant decline in arrests start-
ing the week of March 8, 2020 but only lasting three weeks (ending week of March 22
Figure 6. Arrests by week.
16 M. D. WHITE ET AL.
when arrests plateau). There is no significant change in arrests associated with George
Floyd’s death.
Use of Force
Figure 7 shows a flat trend in weekly use of force until early 2019. During the pan-
demic only and pandemic þGeorge Floyd periods, use of force continues to tick up
(33.5 and 36.0, respectively; increases of 22.3% and 7.5%). Consistent with this assess-
ment, the ARIMA analysis identified best-fitting models with a statistically significant
increase in use of force starting the week of March 8, 2020 and a separate significant
increase beginning the week of July 19, 2020 (both lasting through the end of the
study period).
5
The increase in use of force is especially notable given the immediate
significant declines in other measures of police business (citizen and officer-initiated
calls, reported crime, arrests).
Accidents
Figure 8 shows the weekly number of traffic accidents in Tempe during the study
period. Pre-pandemic, police responded to an average of 110.6 accidents per week.
During the pandemic only period, the average number of traffic accidents dropped by
56.2% (48.5). Accidents increase slightly in mid-May but remain well below pre-pan-
demic levels through the end of the study period. ARIMA analysis confirmed this
result: Table 1 shows a best-fitting model with statistically significant drop starting the
week of March 22, 2020 and lasting through the end of the study period. There is also
a separate significant decline starting the week of June 7 and lasting through the end
of January 2021.
Figure 7. Use of force by week.
5
We conducted the same analyses with an incident-based measure of use of force (number of incidents with a use
of force). The results do not change.
JUSTICE QUARTERLY 17
Body-Worn Camera Review
Table 2 shows a comparison of nearly 50 variables across police-citizen encounters during
the pre-pandemic (n¼230), pandemic only (n¼160), and pandemic þGeorge Floyd
(n¼84) periods. These variables represent the entire sum of dynamics captured for each
incident (White et al., 2021). Several important findings emerge. First, the average length
of encounters declined by 3.5 minutes over time (from 15.7 minutes to 12.19 minutes; a
22.5% decline). The reasons for this are not clear, though it may be that officers resolved
encounters more quickly to reduce infection risk to themselves and the citizens present,
as well as in response to the intense scrutiny following George Floyd’s death. Second, offi-
cers displayed high levels of procedural justice with citizens regardless of the two extraor-
dinary events. For example, officers nearly always answered citizens’questions
(97.2%–99.3%), appeared invested (93.3%–98.8%), and spoke to citizens in a respectful
manner (92.9%–98.8%). In fact, the positive change in increased respect from the pan-
demic only to pandemic þGeorge Floyd periods reaches statistical significance. In plain
terms, these two extraordinary events did not reduce the level of procedural justice pro-
vided to Tempe citizens. Third, citizen drug/alcohol use and mental illness were rarely evi-
dent, though both drug use and mental illness spiked temporarily during the pandemic
only period (5.2% and 7.6%, respectively).
Fourth, we captured 37 different variables that reflect dynamics of how officers
handled encounters, including use of de-escalation tactics. There were no significant
differences in 26 of the 37. Across the three time periods, officers mostly continued to
handle encounters in ways that minimized the potential for a violent outcome. For
example, officers continued to:
speak in a calm manner (92.8%–96.8%);
display patience (91.7%–95.2%);
maintain sufficient personal space (97.6%–99.6%);
Figure 8. Accidents by week.
18 M. D. WHITE ET AL.
use active listening (91.4%–96.3%);
control their emotions (94.3%–97.6%);
prioritize officer safety (98.5%–100%);
use time to their advantage (94.1%–97.6%).
Officers also continued to avoid behaviors that would potentially aggravate a citi-
zen unnecessarily. Officers rarely: used charged/imposing body language (1.2%–2.7%);
ignored what the citizen was communicating (2.6%–3.8%); lost their patience
(4.8%–5.8%); or pressed the situation unnecessarily (2.4%–4.6%).
6
There was also no
significant change in how officers resolved encounters, though issuance of tickets/cita-
tions declined by half. Along these 26 dimensions, officers’handling of citizen encoun-
ters was largely unaffected by the global pandemic and George Floyd’s death.
However, there were a handful of significant differences over time that reflect both
negative and positive change. After George Floyd’s death, officers were:
less likely to use friendly language (39.8% compared to 53.6% pre-pandemic);
7
less likely to explain their decision-making (though this remained very common;
83.6% compared to 92.0% pre-pandemic);
less likely to appear empathetic (25.9% compared to 64.2% pre-pandemic);
less likely to compromise (20.6% compared to 53.2% pre-pandemic);
less likely to encourage citizens that a positive outcome could be reached (19.5%
compared to 60.7% pre-pandemic);
less likely to attempt to build common ground (50.6% compared to 77.9% pre-pan-
demic) or rapport (19.5% compared to 50.7% pre-pandemic);
less likely to use “we”language (6.1% compared to 22.6% pre-pandemic).
Alternatively, several dimensions trended positively after George Floyd’s death.
Officers were:
more likely to speak in a respectful manner (98.8% compared to 92.9% pan-
demic only);
more likely to use friendly physical contact (8.4% compared to 2.6% pan-
demic only);
more likely to transfer control to another officer, if necessary (100% compared to
53.8% pre-pandemic);
less likely to act impersonally (2.4% compared to 11.4% pre-pandemic).
Also, none of the cases we observed after May 24, 2020 involved an officer use of
force, compared to 5.7% and 8.1% of incidents in the earlier periods (both statistically
significant).
6
We acknowledge that the coding of many of these measures is subjective. As a result, we carried out intensive
checks of inter-rater reliability, described in the Methods section.
7
For the pre-pandemic v. pandemic only comparison, statistical significance is displayed in the pandemic only
column (p<.05; p<.01). For the pre-pandemic v. pandemic þGeorge Floyd comparison, statistical significance is
displayed in the pandemic þGeorge Floyd column (p<.05; p<.01). For the pandemic only v. pandemic þGeorge
Floyd comparison, statistical significance is displayed in the pandemic þGeorge Floyd column (
a
p<.05;
b
p<.01).
JUSTICE QUARTERLY 19
Discussion
Existing research has not captured the full impact of the global pandemic and George
Floyd’s death on police work. The current study sought to investigate the impact of
those events through an intensive examination of the nature, prevalence, and dynam-
ics of crime and police work in Tempe, Arizona. Drawing on routine activities and legal
cynicism, we hypothesized that crime and other measures of police business would
decline after the onset of the global pandemic, and those same measures would
increase following the death of George Floyd. In other words, we anticipated that
these events would exert opposite effects on police work. We also hypothesized that
incident-level dynamics measured through BWC footage would become more negative
following both extraordinary events.
The Impact on Crime
The impacts of the pandemic and George Floyd’s death on crime were mixed, provid-
ing partial support for our hypotheses. All reported crime dropped immediately and
significantly after the onset of COVID-19 but the decline lasted five months—well after
George Floyd’s death. Crimes against property and society, and burglary, experienced
dramatic declines associated with the pandemic, but those effects lasted less than a
month (ending weeks before George Floyd’s death). Crimes against persons, assaults,
robbery, and domestic violence were not affected at all by the pandemic. None of the
specific crime types or offenses changed in the immediate aftermath of George
Floyd’s death, though there was a significant decline in all reported crime. These find-
ings suggest that the decline in all offenses (lasting through August 2020) was driven
by reductions in misdemeanor and other offenses not included in the general NIBRS
crime categories. More generally, Tempe did not experience the large spike in violence
and homicide documented in national UCR data for 2020, but most crime types
returned to or exceeded pre-pandemic levels by the end of 2020.
The Impact on Other Measures of Police Activity
The pandemic had a much more consistent effect on other measures of police work,
supporting our hypothesis. Citizen-initiated calls for service declined by 22%. Officer-
initiated calls for service declined by 11%. Arrests declined by 46%. Traffic accidents
declined by 56%. Not only were these changes statistically significant, they were dra-
matically so (see the ARIMA diagnostics in Table 1).
Several (but not all) measures dropped again following George Floyd’s death,
including citizen calls for service and traffic accidents. Officer proactivity declined pre-
cipitously after George Floyd’s death (suggestive of a de-policing argument; Shjarback
et al., 2017). The large declines in officer proactivity after both events is evidence of
an interactive effect between the pandemic and George Floyd’s death, though pro-
activity climbed back to pre-pandemic levels by November 2020. The one notable out-
lier is officer use of force. Use of force spiked in the weeks following the global
pandemic and continued to rise in the summer and fall of 2020, which also suggests a
possible interactive effect. We address this finding below.
20 M. D. WHITE ET AL.
Making Sense of the Findings
Localized Effects
The findings from Tempe highlight the localized effects of these events. Cities, coun-
ties, and states varied significantly in how they responded to the pandemic (White &
H
ebert-Dufresne, 2020). Jurisdictions differed in the length and duration of stay-at-
home orders, how those orders were enforced, when non-essential businesses re-
opened, the imposition of mask mandates, etc. The local variation in COVID-19
response likely produced differing impacts on crime and other features of police busi-
ness. The same can be said for George Floyd’s death. Citizens in some communities
reacted violently to George Floyd’s death. In other places, protests were peaceful. In
some places, there were no protests at all. In plain terms, the impact of these events
may differ from one city to the next. This study tells the story of one city.
The extent to which Tempe’s story aligns with the experiences of other jurisdictions
remains unknown. However, Tempe has characteristics common to many other juris-
dictions in both Arizona and across the U.S. It is medium-sized city with a police
department of just under 350 sworn officers. It borders a much larger city and is part
of a sprawling metropolitan area. It is home to a large university. It experiences few
homicides but has higher-than-average property crime rates. Tempe was subject to
state-level executive orders related to the pandemic, and the city experienced protests
following George Floyd’s death. Local context is clearly important, but we believe the
findings here have implications, both theoretical and policy-related, that extend
beyond the city of Tempe.
Immediate Pandemic-Driven Declines
Why did Tempe experience immediate and dramatic declines in some crimes and
other measures of police activity after the emergence of the global pandemic? We
believe routine activities theory has particular relevance. Stay-at-home orders, closures
of schools and non-essential businesses, and fear of infection greatly reduced the
opportunity for crime by limiting the availability of victims and restricting the move-
ment of offenders (two central components of the crime triangle). At the same time,
the capable guardians were still present, visible, and responsive (though they too
were affected by the pandemic). Routine activities theory offers a valuable lens by
which to explain the immediate reductions in all offenses, crimes against property and
society, and burglary in the weeks after the emergence of COVID-19. The theory also
would explain the dramatic reductions in other measures of police business, such as
arrests, calls for service, officer proactivity, and accidents.
Short-Term (or No) Pandemic-Driven Declines
Why were many of the effects on crime short-term? And why was there no impact at
all on serious person crimes, assaults, robbery, and domestic violence? We hypothe-
sized that the effects of the pandemic would last at least until George Floyd’s death in
May 2020, but in most cases, the declines had rebounded within a month. There are a
number of potential explanations. First, it is possible that the stay-at-home orders
were only effective in restricting the movement of people in public spaces for a few
weeks. Most police departments, including TPD, adopted an educational/informational
JUSTICE QUARTERLY 21
approach to individuals who violated the stay-at-home order, rather than an enforce-
ment approach. People may have quickly decided to not abide by the order.
Moreover, Arizona’s stay-at-home order expired on May 15, 2020 (Husch Blackwell,
2021), thereby easing limits on movement.
Second, the consequences of the pandemic likely left many people with large
amounts of free time. People were out of work; schools, gyms, and businesses were
closed; restaurants, movie theaters, malls, hotels, casinos, and air travel were shut
down. Opportunity theories such as crime pattern theory (Brantingham &
Brantingham, 1993) and routine activities (Cohen & Felson, 1979) would predict a
crime increase under such circumstances. Available evidence also shows drug and
alcohol use increased significantly in 2020 (Abramson, 2021), and there is an estab-
lished connection between substance use and crime (Bennett et al., 2008). The
increases in unencumbered time combined with greater substance use may have
increased victimization risk and reversed the initial suppressing effect of COVID-19
on crime.
Third, TPD’s ability to prevent and respond to crime may have been inhibited by
the operational changes that were implemented to reduce risk of infection, such as
increased use of telephone and online reporting, shift and unit adjustments, reduced
informal contacts, and less self-initiated activity. Table 1 and Figure 1 show large
cumulative drops in officer proactivity immediately following the onset of the pan-
demic and after George Floyd’s death. The decline in proactivity in early March is
explained by policy changes made by the department to reduce infection risk. The
decline after George Floyd’s death is not explained by such a policy change. Rather, it
appears that the outrage and scrutiny after George Floyd’s death led to an informal
pullback in proactivity consistent with the de-policing thesis. Many police leaders and
researchers have pointed to reductions in preventive policing and proactivity as con-
tributors to the spike in violence across the U.S. (MacDonald, 2015; Rushin & Edwards,
2017). Though Tempe did not experience a large spike in violence after Floyd’s death,
nearly all crime types had returned to or exceeded pre-pandemic levels by the end of
the year. From a routine activities perspective, these changes reduced the presence of
capable guardians and increased opportunities for motivated offenders. This explan-
ation raises particular concerns about how police can maintain their effectiveness
when external events cause forced (pandemic) and internally driven (George Floyd’s
death) reductions in officer proactivity.
George Floyd’s Impact
George Floyd’s death had an independent impact on some measures of crime and
police activity but those effects were less significant than the pandemic (with the
exception of officer proactivity). In particular, crime and violence did not spike as we
hypothesized. Why? Our review of BWC footage may offer insights on this question.
We compared nearly 50 variables in 474 police-citizen encounters before and after
both events. Before March 12, 2020, Tempe police officers treated citizens with high
levels of procedural justice. That did not change after March 12, 2020, or after George
Floyd’s death. We also examined 37 different officer behavioral and verbal measures,
and 30 of the 37 either did not change or changed in a positive direction after the
22 M. D. WHITE ET AL.
emergence of a global pandemic and George Floyd’s death. The high levels of proced-
ural justice and the stability in most encounter-dynamics may have served as protect-
ive factors that minimized the impact of George Floyd’s death in Tempe. There may
also be other intervening variables we did not measure.
The Increase in Use of Force
Despite large declines in citizen calls for service, officer-initiated activity, arrests, and
some types of crime, police use of force increased significantly after both events. In
effect, there were far fewer police-citizen encounters after the onset of the global pan-
demic and George Floyd’s death, but those encounters were more likely to involve a
use of force. The troubling association between these two extraordinary events and
increased use of force may be explained by a few factors. As the global pandemic
took hold, TPD focused their response on the most serious calls involving threats to
property and life. In effect, TPD officers responded to a greater proportion of higher-
risk calls. Also, citizen frustration with the global pandemic, stay-at-home orders, other
consequences of COVID-19, and anti-police sentiment after George Floyd’s death may
have led to greater aggression toward the police (partially supporting our George
Floyd-related hypothesis). Those same factors could have similarly affected officers,
leading to more aggression toward citizens. Though the explanation remains elusive,
it is clear that Tempe officers used force more frequently after the onset of the pan-
demic and George Floyd’s death (compared to before).
The increase in use of force is inconsistent with our review of BWC footage, which
showed a reduction in use of force among randomly selected incidents after George
Floyd’s death. This may be explained by a few factors. First, though the selection was
random, the 474 videos examined here represent a very small proportion of the total
number of encounters during the study period. Second, our BWC review was limited
to a sample of patrol and specialty unit officers who participated in a randomized con-
trolled trial study of de-escalation training. The training was delivered in February and
the first week of March 2020, just before the emergence of the global pandemic. The
evaluation showed that both the Treatment and Control groups used less force (meas-
ured through the same administrative data examined here) in the six months follow-
ing training delivery (incorporating both extraordinary events examined here; White
et al., 2021). As a result, the increase in use of force documented in administrative
data is driven by officers who did not participate in the de-escalation experiment (i.e.
were not assigned to the general patrol function).
The inconsistency between BWC and administrative data may also be explained by
the randomized sampling process for BWC videos and the rare prevalence of use of
force generally. For example, during the pre-pandemic period there were 3,283 uses
of force across 427,543 police-citizen interactions (citizen- and officer-generated com-
bined). Use of force occurred in less than one percent of encounters (.8%). This rate
increased to 1.6% in the pandemic only period (368 uses of force in 23,268 encoun-
ters) and 1.7% after George Floyd’s death (1,295 uses of force in 74,690 encounters),
but overall, use of force remains rare. The random selection of BWC footage may cap-
ture incidents that are more reflective of the 98% of encounters between Tempe
police officers and citizens that did not involve a use of force.
JUSTICE QUARTERLY 23
Limitations
The current study has a number of limitations that warrant discussion. We report on
findings from one police department in the southwestern U.S. The extent to which
these findings are generalizable to other jurisdictions is not known (see the discussion
above on localized effects). Each data source examined here also has limitations. For
example, the shortcomings of administrative police data are well-known (Logan &
Ferguson, 2016). Similarly, body-worn camera footage has a number of shortcomings,
from officers’failure to record encounters to obstructed views (White & Coldren,
2017). The 474 videos reviewed here represent a very small proportion of the total
encounters during the study period. Also, our random selection of footage only cap-
tured encounters that were recorded on BWC, though available evidence suggests the
TPD has a high activation rate (Patterson & White, 2021; Tempe, 2021). Also, we col-
lected BWC footage from a group of 216 officers who participated in a de-escalation
training study during 2019–2020. The administrative data examined here includes all
officers in the department (n350) and encompasses a longer time period. Moreover,
our coding scheme is inherently subjective for a number of officer and citizen-related
variables, though we did employ a robust inter-rater reliability system to minimize this
concern. Last, we treat the global pandemic and George Floyd’s death as events with
discrete start dates. This may imply that each is a one-time occurrence (which they
are not), and in the case of the pandemic, the selection of a start date is subjective.
We did employ different intervention onsets and durations in ARIMA to minimize
this concern.
Conclusion
The current study represents one of the most comprehensive efforts to date to cap-
ture the impact of COVID-19 and George Floyd’s death on police work. The impact on
crime was mixed and varied notably by crime type. When declines occurred, they
were short-term. The pandemic immediately and dramatically altered nearly every
measure of citizen and officer activity, and about two months later, George Floyd’s
death led to additional significant impacts on a few of those measures. Overall, the
pandemic had a much larger impact on crime and other features of police work than
did George Floyd’s death. Routine activities theory proved to be a valuable theoretical
lens for interpreting these findings. In the face of these dramatic changes, review of
body-worn camera footage highlighted the consistency in how officers effectively
handled citizen encounters. In the nearly 500 encounters we examined, they contin-
ued to use procedural justice. They also continued to use (or increased the use of) 30
of 37 different de-escalation tactics. All of the findings underscore the importance of
localized context for these events and their effects.
Disclosure statement
No potential conflict of interest was reported by the authors.
24 M. D. WHITE ET AL.
Notes on contributors
Michael D. White is a Professor in the School of Criminology and Criminal Justice at Arizona
State University, and is Associate Director of ASU’s Center for Violence Prevention and
Community Safety. Prior to entering academia, he worked as a deputy sheriff in Pennsylvania.
Dr. White’s primary research interests involve the police, including use of force, technology, and
misconduct. His recent work has been published in Criminology and Public Policy, and Journal of
Experimental Criminology, and Injury Prevention.
Carlena Orosco is a third-year PhD Candidate in the School of Criminology and Criminal Justice
at Arizona State University, and a Research Assistant in the Center for Violence Prevention and
Community Safety. She is also employed full time as a Crime Analyst in the Strategic Planning,
Analysis and Research Center (SPARC) at Tempe Police Department. Carlena’s primary research
interests include police decision-making and discretion, police dispatchers, as well as crime ana-
lysis and the spatial dynamics of crime. Carlena’s work can be found in the Journal of Criminal
Justice,Policing: A Journal of Policy and Practice, and Policing: An International Journal.
Brice Terpstra is a doctoral student in the School of Criminology and Criminal Justice at Arizona
State University, and a Research Assistant in the Center for Violence Prevention and Community
Safety. His primary research interests include the intersection of mental illness and the criminal
justice system, criminal justice policy, problem-solving courts, and justice-involved Veterans. His
work has appeared in Justice Quarterly, Crime and Delinquency, and International Journal of
Offender Therapy and Comparative Criminology.
ORCID
Michael D. White http://orcid.org/0000-0002-6151-6399
Carlena Orosco http://orcid.org/0000-0002-7491-0897
Brice Terpstra http://orcid.org/0000-0002-9550-1080
References
Abrams, D. S. (2021). COVID and crime: An early empirical look. Journal of Public Economics,194,
104344. https://doi.org/10.1016/j.jpubeco.2020.104344
Abramson, J. (2015, May 18). 10 cities making real progress since the launch of the 21st century
policing task force.https://obamawhitehouse.archives.gov/blog/2015/05/18/10-cities-making-
real-progress-launch-21st-century-policing-task-force
Abramson, A. (2021). Substance abuse during the pandemic. American Psychological Association.
https://www.apa.org/monitor/2021/03/substance-use-pandemic
Ashby, M. P. (2020). Changes in police calls for service during the early months of the 2020 cor-
onavirus pandemic. Policing: A Journal of Policy and Practice,14(4), 1054–1072. https://doi.org/
10.1093/police/paaa037
Bennett, T., Holloway, K., & Farrington, D. (2008). The statistical association between drug misuse
and crime: A meta-analysis. Aggression and Violent Behavior,13(2), 107–118. https://doi.org/10.
1016/j.avb.2008.02.001
Blasius, M. (2020, December 10). Tempe disciplines officer, settles hotel gunpoint claim. ABC, 15.
https://www.abc15.com/news/region-southeast-valley/tempe/tempe-disciplines-officer-settles-
hotel-gunpoint-claim
Bosman, J., & Leatherby, L. (2021). U.S. coronavirus death toll surpasses 700,000 despite wide
availability of vaccines. New York Times.https://www.nytimes.com/2021/10/01/us/us-covid-
deaths-700k.html
Brantingham, P. L., & Brantingham, P. J. (1993). Environment, routine, and situation: Toward a
pattern theory of crime. In: R. V. Clarke and M. Felson (Eds.), Routine activity and rational
choice. Advances in Criminological Theory (Vol. 5). Transaction Publications.
JUSTICE QUARTERLY 25
Bullinger, L. R., Carr, J. B., & Packham, A. (2020). COVID-19 and crime: Effects of stay-at-home
orders on domestic violence. National Bureau of Economic Research. https://doi.org/10.3386/
w27667
Burkitt, B., & Pineda, P. (2020, September 17). Tempe police chief Sylvia Moir abruptly resigns.
Arizona Central.https://www.azcentral.com/story/news/local/tempe-breaking/2020/09/15/
tempe-police-chief-sylvia-moir-abruptly-resigns/5806567002/
Campedelli, G. M., Aziani, A., & Favarin, S. (2020). Exploring the immediate effects of COVID-19
containment policies on crime: An empirical analysis of the short-term aftermath in Los
Angeles. American Journal of Criminal Justice. Advance online publication. https://doi.org/10.
1007/s12103-020-09578-6 .https://doi.org/10.31219/osf.io/gcpq8
Cohen, L., & Felson, M. (1979). Social change and crime rate trends: A routine activity approach.
American Sociological Review,44(4), 588–608. https://doi.org/10.2307/2094589
Haeder, S. F., & Gollust, S. E. (2020). From poor to worse: Health policy and politics scholars’
assessment of the U.S. COVID-19 response and its implications. World Medical & Health Policy,
12(4), 454–481. https://doi.org/10.1002/wmh3.371
Hermann, P., Morse, D., & Jackman, T. (2021, February 17). Patrolling the beat while keeping a
distance: Police officers face new rules, rhythms amid pandemic. The Washington Post.https://
www.washingtonpost.com/local/public-safety/patrolling-the-beat-while-keeping-a-distance-
police-officers-face-new-rules-rhythms-amid-pandemic/2020/03/27/334b3602-6afc-11ea-abef-
020f086a3fab_story.html
Husch Blackwell. (2021). State-by-state Covid-19 guidance.https://www.huschblackwell.com/ari-
zona-state-by-state-covid-19-guidance
Jennings, W. G., & Perez, N. M. (2020). The immediate impact of COVID-19 on law enforcement
in the United States. American Journal of Criminal Justice,45(4), 690–701. https://doi.org/10.
1007/s12103-020-09536-2
Kaiser, K., & Reisig, M. (2019). Legal socialization and self-reported criminal offending: The role of
procedural justice and legal orientations. Journal of Quantitative Criminology,35(1), 135–154.
https://doi.org/10.1007/s10940-017-9375-4
Kirk, D. S., & Matsuda, M. (2011). Legal cynicism, collective efficacy, and the ecology of arrest.
Criminology,49(2), 443–472. https://doi.org/10.1111/j.1745-9125.2011.00226.x
KTAR. (2020, June 13). Protestors head to Tempe for 15th consecutive day of demonstrations.
https://ktar.com/story/3272526/protesters-head-to-tempe-for-15th-consecutive-day-of-
demonstrations/
Laufs, J., & Waseem, Z. (2020). Policing in pandemics: A systematic review and best practices for
police response to COVID-19. International Journal of Disaster Risk Reduction,51,
101812–101817. https://doi.org/10.1016/j.ijdrr.2020.101812
Leslie, E., & Wilson, R. (2020). Sheltering in place and domestic violence: Evidence from calls for
service during COVID-19. Journal of Public Economics,189, 104241–104247. https://doi.org/10.
1016/j.jpubeco.2020.104241
Logan, W. A., & Ferguson, A. (2016). Policing criminal justice data. Minnesota Law Review,101(2),
541–616.
Lucas, R. (2021). FBI data shows an unprecedented spike in murders nationwide in 2020. NPR.
https://www.npr.org/2021/09/27/1040904770/fbi-data-murder-increase-2020
Lum, C., Maupin, C., & Stoltz, M. (2020). The impact of COVID-19 on law enforcement agencies
(Wave 2). Center for evidence-based crime policy. https://www.theiacp.org/sites/default/files/
IACP_Covid_Impact_Wave2.pdf
MacDonald, H. (2015, May 29). Wall Street Journal: The new nationwide crime wave. Retrieved
from https://www.wsj.com/articles/the-new-nationwide-crime-wave-1432938425
Martinez, J. (2020, July 28). Tense moments during protest over police brutality in Tempe. Fox10
Phoenix.https://www.fox10phoenix.com/news/tense-moments-during-protest-over-police-bru-
tality-in-tempe
McCleary, R., McDowall, D., & Bartos, B. J. (2017). Design and analysis of time series experiments.
Oxford University Press.
26 M. D. WHITE ET AL.
McDowall, D., & McCleary, R. (2014). Interrupted time series models. Encyclopedia of Criminology
and Criminal Justice (pp. 2653–2665). Springer New York.
Mohler, G., Bertozzi, A. L., Carter, J., Short, M. B., Sledge, D., Tita, G. E., Uchida, C. D., &
Brantingham, P. J. (2020). Impact of social distancing during COVID-19 pandemic on crime in
Los Angeles and Indianapolis. Journal of Criminal Justice,68, 101692–101697. https://doi.org/
10.1016/j.jcrimjus.2020.101692
Mourtgos, S. M., & Adams, I. T. (2021). COVID-19 vaccine program eliminates law enforcement
workforce infections: a Bayesian structural time series analysis. Police Practice and Research,
22(5), 1557–1565. https://doi.org/10.1080/15614263.2021.1894937
National Police Foundation (NPF). (2020). COVID-19 Resources for Law Enforcement.https://www.
policefoundation.org/covid-19/
Nix, J., & Richards, T. N. (2021). The immediate and long-term effects of COVID-19 stay-at-home
orders on domestic violence calls for service across six US jurisdictions. Police Practice and
Research,22(4), 1443–1451. https://doi.org/10.1080/15614263.2021.1883018
Nowotny, K., Bailey, Z., Omori, M., & Brinkley-Rubinstein, L. (2020). COVID-19 exposes need for
progressive criminal justice reform. American Journal of Public Health,110(7), 967–968.
ODMP. (2021, January 7). Law enforcement line of duty deaths in 2020.https://www.odmp.org/
search/year?year=2020
Patterson, Q., & White, M. D. (2021). Is there a civilizing effect on citizens? Testing the pre-condi-
tions for body worn camera-induced behavior change. Police Quarterly,24(4), 411–437.
https://doi.org/10.1177/1098611121996716
Piquero, A. R., Riddell, J. R., Bishopp, S. A., Narvey, C., Reid, J. A., & Piquero, N. L. (2020). Staying
home, staying safe? A short-term analysis of COVID-19 on Dallas domestic violence. American
Journal of Criminal Justice,45(4), 601–635. https://doi.org/10.1007/s12103-020-09531-7
Police Executive Research Forum (PERF). (2020). Responding to the COVID-19 Coronavirus.https://
www.policeforum.org/coronavirus
Rosenfeld, R., & Lopez, E. (2020). Pandemic, social unrest, and crime in US Cities. Washington,
DC: Council on Criminal Justice.https://cdn.ymaws.com/counciloncj.org/resource/resmgr/
covid_commission/Crime_in_US_Cities_-_October.pdf
Rushin, S., & Edwards, G. (2017). De-policing. Cornell Law Review,102, 721–782.
Sampson, R. J., & Bartusch, D. J. (1998). Legal cynicism and (subcultural?) tolerance of deviance:
the neighborhood context of racial differences. Law & Society Review,32(4), 777–804. https://
doi.org/10.2307/827739
Sauer, M. A., Truelove, S., Gerste, A. K., & Limaye, R. J. (2021). A failure to communicate? How
public messaging has strained the COVID-19 response in the United States. Health Security,
19(1), 65–74. https://doi.org/10.1089/hs.2020.0190
Shjarback, J. A., Pyrooz, D. C., Wolfe, S. E., & Decker, S. H. (2017). De-policing and crime in the
wake of Ferguson: Racialized changes in the quantity and quality of policing among Missouri
police departments. Journal of Criminal Justice,50,42–52. https://doi.org/10.1016/j.jcrimjus.
2017.04.003
Silverstein, J. (2021, July 10). The global impact of George Floyd: How Black Lives Matter protests
shaped movements around the world. CBS News.https://www.cbsnews.com/news/george-
floyd-black-lives-matter-impact/
Smith-Schoenwalder, C. (2020, March 18). Trump declares national emergency amid coronavirus
pandemic.https://www.usnews.com/news/national-news/articles/2020-03-13/trump-declares-
national-emergency-amid-coronavirus-pandemic
Subramanian, R., & Arzy, L. (2021, July 13). State policing reforms since George Floyd’s Murder.
Brennan Center. https://www.brennancenter.org/our-work/research-reports/state-policing-
reforms-george-floyds-murder
Taylor, D. B. (2021, July 12). George Floyd protests: A timeline. The New York Times.https://www.
nytimes.com/article/george-floyd-protests-timeline.html
Tempe Police Department Quick Facts. (2021, March 3). Tempe Police Department.https://www.
tempe.gov/home/showpublisheddocument?id=84401
JUSTICE QUARTERLY 27
Tempe. (2021). Police body cameras (dashboard).https://covid19.tempe.gov/datasets/
c762d80604894b3caede88e9418a69b4
U.S. Census Bureau. (2021, March 8). City of Tempe, U.S. Census Bureau Quick Facts.https://www.
census.gov/quickfacts/fact/table/tempecityarizona,US/PST045219
Weichselbaum, S., & Li, W. (2020, November 5). As coronavirus surges, crime declines in some cit-
ies. The marshall project. https://www.themarshallproject.org/2020/03/27/as-coronavirus-
surges-crime-declines-in-some-cities
White, E. R., & H
ebert-Dufresne, L. (2020). State-level variation of initial COVID-19 dynamics in
the United States. PLoS One,15(10), e0240648 https://doi.org/10.1371/journal.pone.0240648
White, M. D., & Fradella, H. F. (2020). Policing a pandemic: stay-at-home orders and what they
mean for the police. American Journal of Criminal Justice,45(4), 702–717. https://doi.org/10.
1007/s12103-020-09538-0
White, M. D., & Coldren, J. (2017, February 12). Body-worn police cameras: Separating fact from
fiction. PM Magazine (March), 6–9. https://icma.org/articles/pm-magazine/pm-article-body-
worn-police-cameras-separating-fact-fiction
White, M. D., Orosco, C., Pooley, M., & Sorensen, D. (2021). Testing the impact of de-escalation
training on officer behavior: The Tempe (AZ) Smart Policing Initiative. Final Report. Center for
Violence Prevention and Community Safety, Arizona State University.
Winton, R., Tcheckmedyian, A. (2020, March 19). Coronavirus has authorities putting more police
on streets, releasing inmates from jails. Los Angeles Times.https://www.latimes.com/california/
story/2020-03-17/coronavirus-has-authorities-putting-more-police-on-streets-releasing-inmates-
from-jails
Woolf, S. H., Chapman, D. A., Sabo, R. T., Weinberger, D. M., Hill, L., & Taylor, D. D. (2020). Excess
deaths from COVID-19 and other causes, March-July 2020. JAMA,324(15), 1562–1564. https://
doi.org/10.1001/jama.2020.19545
28 M. D. WHITE ET AL.
