I need two to three pages by tommorow January 14th, 2025 Please provide plagarism and AI free report. Please see all references and websites attached
Pinkdiva
DatainPolicingAnIntegrativeReview.pdf
International Journal of Public Administration
ISSN: 0190-0692 (Print) 1532-4265 (Online) Journal homepage: www.tandfonline.com/journals/lpad20
Data in Policing: An Integrative Review
Muhammad Afzal & Panos Panagiotopoulos
To cite this article: Muhammad Afzal & Panos Panagiotopoulos (2025) Data in Policing: An Integrative Review, International Journal of Public Administration, 48:7, 411-430, DOI: 10.1080/01900692.2024.2360586
To link to this article: https://doi.org/10.1080/01900692.2024.2360586
© 2024 The Author(s). Published with license by Taylor & Francis Group, LLC.
Published online: 30 May 2024.
Submit your article to this journal
Article views: 13199
View related articles
View Crossmark data
Citing articles: 9 View citing articles
Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=lpad20
Data in Policing: An Integrative Review Muhammad Afzal and Panos Panagiotopoulos
School of Business & Management, Queen Mary University of London, London, UK
ABSTRACT The article brings together contemporary research on how new forms of data are reshaping policing and police organizations. A systematic review of 192 papers (1970–2022) identifies the developing effects of data in areas like automated decision-making and predictive policing. We synthesize the findings in an integrative framework that links data sources, data-driven processes, and policing outcomes while highlighting the significance of institutional, organizational and individual-level mod- erators. The acceleration of data in policing raises critical questions for public administration scholars. We discuss the implications on the evolving nature of discretion and decision-making, the training of police officers, and tensions between data professionals and street-level officers. We further unpack the implications for organizational transformation and the changing nature of policing strategies.
KEYWORDS Data-driven policing; police management; systematic review; digital transformation; e- government
Introduction
Interest in data-driven policing can be traced back to the digitization of police records in the 1970s and the use of computerized databases for strategic planning and opera- tional allocations in the early 1990s (Danziger & Kraemer, 1985; Manning, 2008). The first significant milestone was the development of the CompStat (1994) performance management tool in New York, predicated upon crime analytics and modern data visualization (Brayne, 2020; Manning, 2008; Silverman & O’Connell, 1999). CompStat served as a precursor for hotspot and problem-oriented policing, which became internationally prominent strategies in the early 2000s. Meanwhile, sys- tems like CompStat were widely exported to other states in the US, establishing the foundation for data-intensive management strategies known as PerformanceStat (Behn, 2008; Nam, 2014; O’Malley, 2014; Pasha et al., 2023).
More recently, policing has witnessed the inception of the full array of technologies related to data analytics, new forms of data and algorithmisation (Brayne, 2020; Eubanks, 2017). Major advancements have concentrated on surveillance technologies, predictive algorithms and mining open information sources (Brayne & Christin, 2021). The events of 9/11 compelled police authorities to build expertise on surveillance and monitoring with technologies like smart CCTVs, dashboard cameras, body-worn video cameras, and automated number plate readers (Brayne, 2020; Ferguson, 2017a). Other popular policing technologies have deployed sensors in urban
settings to collect vast quantities of data on the movement of individuals and vehicles and shifts in environmental indicators like heat, sound, or pollution (Ferguson, 2017a; Levine et al., 2017). Further, open information networks like social media platforms have provided large data streams of text, photos, and videos that can be translated into collective reactions regarding any place, person, or event (Williams et al., 2013).
These new forms of data offer opportunities to trans- form policing by pursuing improvements in areas like operational efficiency, accurate decision-making, enhanced supervision, better regulation, and service deliv- ery (Agarwal, 2018; Allard et al., 2018; Barrett & Greene, 2023; Mergel et al., 2016; Peeters & Schuilenburg, 2018; Vogl et al., 2020; Wirtz et al., 2019). While some of these benefits can be observed in practice, there has also been increased emphasis on scrutinizing policing actions, spurred by the escalating occurrences of fatal shootings during stop-and-frisk operations involving racial and eth- nic minorities (Shoub, 2022). This prompts inquiries into the extent to which the use of data has transformed poli- cing, enhanced objectivity and transparency of police deci- sion-making, and how police culture views the integration of emergent data technologies.
Research has not systematically addressed the data- driven transformation of policing and its implications within the public administration literature. Studies have focused on topics like social media (J. P. Walsh &
CONTACT Muhammad Afzal [email protected] School of Business & Management, Queen Mary University of London, Room 3.33a, Francis Bancroft Building, Mile End Campus, London E1 4NS, UK
INTERNATIONAL JOURNAL OF PUBLIC ADMINISTRATION 2025, VOL. 48, NO. 7, 411–430 https://doi.org/10.1080/01900692.2024.2360586
© 2024 The Author(s). Published with license by Taylor & Francis Group, LLC. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent.
O’Connor, 2019) or whether predictive policing has been adopted productively (Benbouzid, 2019; Egbert & Krasmann, 2020; Hardyns & Rummens, 2017; Meijer & Wessels, 2019). A coherent view of data-driven policing remains opaque in the literature, and relevant research is fragmented across applications that concentrate on hotspot, problem-oriented, intelligence-led and predic- tive strategies. Building on this work, we aim to organize current knowledge in data-driven policing and join the evidence on its effects.
The study aims to inform a new approach to the role of data in policing from a public administration perspective. Our starting point is to examine data applications across the core policing functions of (1) order maintenance (riot suppression, preservation of communal peace and mor- ality), (2) law enforcement (crime prevention and inves- tigation, traffic regulation, patrols, stop-frisk or search- and-seize interventions) and (3) service provision (instances where officers serve as disaster managers, par- ole and probation officers, social support workers) (Badiora & Ntamark, 2020; Liederbach & Travis, 2008; Wilson, 1968). To consolidate existing research, we estab- lish the link between data-driven practices and core poli- cing functions, considering the effect on various police routines embedded in law enforcement, order mainte- nance, and service provision. We identify factors affecting the integration of data applications into core policing functions to provide a better understanding of the com- plexities and implications for police management.
The review is based on a systematic literature search, selection and analysis of 192 articles. The systematic review approach provides a comprehensive, transpar- ent, and replicable set of procedures to identify, evalu- ate, and synthesize the existing literature (Moher et al., 2009; Page et al., 2021). The findings are organized in the form of a framework that offers an integrated per- spective on data-driven applications and functions of policing. Building on these insights, we readdress areas of attention and future directions for public administra- tion research in policing.
Methodology
The systematic literature review follows the four phases of the PRISMA protocol: identification, screening, elig- ibility and inclusion (Moher et al., 2009; Page et al., 2021). The identification phase defined the scope of the study and directed the framing of the search phrase with inclusion-exclusion criteria. The search phrase combined domain and concept terms. The domain term (police OR policing) ensured the relevance of searches to policing. The concept term included data- related keywords (smart OR intelligence OR algorith* OR
predictive OR analytics OR “data science” OR “data driven” OR “big data” OR AI OR “social media”).
Inclusion-exclusion criteria for the selection of arti- cles were applied as follows: (i) published in the English language in peer-reviewed journals between 1970 and 2022, as the 1970s saw the onset of digitalization and integration of computers in police departments (Manning, 2008; Northrop et al., 1995; Nunn, 2001); (ii) empirical focus on the use of data and data practices in police organizations at local, state and federal levels; (iii) conceptual studies grounded in the empirical litera- ture and examples; (iv) technical studies with a prospective dimension, purely conceptual and theore- tical treatises, and other review studies were excluded.
Advanced Boolean searches were conducted using title, abstract and keyword options in Web of Science, Scopus and EbscoHost databases. The initial search pro- duced 8,125 articles, reduced to 5,512 by removing duplicates and applying the inclusion-exclusion criter- ion (i). Manual screening of titles and abstracts on criterion (ii) excluded 5,050 articles. In the eligibility phase, full-text perusal and assessment of the remaining 623 articles following criteria (ii)-(iv) resulted in the selection of 192 articles for full review and analysis. The process is shown in Figure 1.
Data analysis was conducted in two phases. First, the articles were structurally organized following Webster and Watson (2002)’s conceptual-matrix approach (see Table 1). Key concepts were compared and grouped into
Figure 1. PRISMA flow chart–steps of the systematic review.
412 M. AFZAL AND P. PANAGIOTOPOULOS
three research streams: (i) data-driven techniques and applications prevalent in policing (n = 78); (ii) the impact on core policing functions (n = 45); and (iii) the antecedents of efficacious use of data for organiza- tional transformation (n = 69).
Second, the articles were thoroughly perused and coded using the qualitative data analysis software NVivo 12. Abstract concepts were initially openly coded based on similarity, interchangeability and differ- ences. They were iteratively compared and grouped based on dimensional commonalities, causes-and- effects and part-and-whole relationships to form theo- retical concepts. The content analysis of research streams (i) and (ii) helped identify various forms of data and data-driven processes. The different forms of data were distinguished using Kitchin (2014)’s typology of directed data, automated data and volunteered data. Data-driven processes were extracted by coding data applications and analytical techniques. Antecedents (iii) were coded and grouped into institutional, organi- zational, and individual levels.
Overview of the articles
The extant literature is highly multidisciplinary, as shown in Table 2, and spans across policing, criminology, or criminal justice (n = 115) and other fields (n = 77) like information systems and technology, sociology and public policy and administration.
Approximately 70% (n = 133) of the articles were published after 2013, showcasing the surging interest in empirically exploring data-related policing practices during the last 10 years (Figure 2). The studies included in the review have mainly presented evidence from the USA, followed by the UK, Europe, Canada, Australasia, Asia and Africa (Figure 3). Almost all empirical studies (n = 149 or 97%) focus on research in a single country.
Most studies are qualitative (n = 128) and based on case study research employing interviews, focus group dis- cussions, analysis of policy documents and field obser- vations. Quantitative studies (n = 65) apply experimental, quasi-experimental and survey-based research designs.
Evolution of data in policing
A timeline of major developments in data in policing is shown in Figure 4 following the analysis of articles published over the last 52 years. Though police organi- zations have maintained manual records and data for a long time, the evolution of data practices in policing can be traced back to the adoption of computers and the digitalization of police records in the 1970s (Colton, 1979; Manning, 2008; Northrop et al., 1995; Nunn, 2001). Around 90% of US city police departments used computers in 1985, while minicomputers, dashboard- mounted and hand-held data terminals for data entry, searching, and visualization were introduced in the early 1990s (Northrop et al., 1995; Nunn, 2001). Digitalized data consisted of arrest records, calls for service, crim- inal histories, vehicle registration, driver’s license, stolen property and vehicles, traffic accidents and violations, warrants and summons (Northrop et al., 1995; Ratcliffe, 2000).
Most US police departments were using computers and databases by the early 1990s (Northrop et al., 1995). At the strategic level, police managers adopted new approaches to planning workforce allocation, criminal analysis, and other administrative functions using data (Nunn, 2001). At the operational levels, patrol officers and detectives had access to databases for manual lookup searches during stop-and-frisk operations to verify vehicle and license status, outstanding warrants, and stolen property. In addition to lookup searches, detectives manually scanned through digitalized crim- inal history files, interrogation reports, and accomplices’ files for collating crime-related information. Files were mainly stored on different computers in disparate loca- tions, which consumed significant time and caused frustration (Brayne, 2017; Northrop et al., 1995).
Databases and digital technologies have underpinned the transition from conventional policing to more proactive approaches. Historically, police institutions were established to enforce social and political control through force, arrests, and surveillance (Brodeur, 1983). Growing criticism of the state’s inability to manage increasing crime rates in the post-1960s prompted a shift toward crime prevention and detection (Bratton & Malinowski, 2008; Brayne, 2017; Heaton, 2000; Lum & Koper, 2017; Sherman, 2013; W. F. Walsh & Vito,
Table 1. Concept matrix. Concepts No. of articles
Data-driven techniques and applications 78 (41%) Impact on core policing functions 45 (23%) Antecedents of efficacious use of data in policing 69 (36%) Total (N) 192 (100%)
Table 2. Multidisciplinary nature of publications.
Journals related to: No. of
journals No. of articles
Policing, Criminology and Criminal Justice 41 (42%) 115 (60%) Others 56 (58%) 77 (40%)
Sociology and Cultural 17 (18%) 21 (11%) Information Systems and Technology 19 (19%) 22 (11%) Public Management and Administration 12 (12%) 26 (13%) Others 8 (8%) 8 (4%)
Total (N) 97 192
INTERNATIONAL JOURNAL OF PUBLIC ADMINISTRATION 413
Figure 2. Graphical trend of publications and methodologies employed.
Region/Country US UK EU Canada International Australia Asia & Africa Total
No. of studies 83 33 32 13 14 10 7 192
USA 43%
Europe 17%
UK 17%
Canada 7%
International 7%
Australasia 5% Asia & Arica 4%
Figure 3. Geographical distribution of articles included in the review.
414 M. AFZAL AND P. PANAGIOTOPOULOS
Fi gu
re 4
. E vo
lu tio
n of
d at
a in
p ol
ic in
g: a
t im
el in
e.
INTERNATIONAL JOURNAL OF PUBLIC ADMINISTRATION 415
2004). When policing practices like random patrolling proved ineffective, new evidence-based approaches like problem-oriented policing (1979), community-oriented policing (1980), and broken-windows policing (1982) were developed (Heaton, 2000; Lum & Koper, 2017; Sherman, 2013; W. F. Walsh & Vito, 2004). Their reli- ance on crime analysis, human intelligence and com- munity collaboration initially yielded mixed results in reducing crime (Bratton & Malinowski, 2008; Heaton, 2000; Lum & Koper, 2017; Magers, 2004; Maguire, 2000; W. F. Walsh & Vito, 2004). Police managers came to a better realization of the true intelligence potential of data when CompStat was introduced in 1994 to mark the beginning of hotspot policing by visualizing crime hotspots (Bratton & Malinowski, 2008; Dannels & Smith, 2001; Weisburd et al., 2003).
In the 2000s, three pivotal developments transformed data use. First, information and communication tech- nologies (ICTs) replaced traditional face-to-face data collection methods and paper systems (Brayne, 2017). Second, in the aftermath of the 9/11 events, the wide- spread adoption of surveillance technologies was inten- sified and legitimized (Brayne, 2017; Ferguson, 2017b; Lyon, 2003). Surveillance networks expanded beyond criminal activities to encompass diverse data sources. Third, advancements in computing power and techni- ques like machine learning significantly enhanced real- time collection, storage, communication, integration, and analysis of vast amounts of data, leading to the big data phenomenon (Ferguson, 2017c; Kitchin, 2014). Subsequently, the unstructured, complex, dynamic, and heterogeneous nature of large data sets led to the development of data analytics (Janssen & Kuk, 2016; Mergel et al., 2016; Ruijer et al., 2023).
In contemporary law enforcement, ICTs, new forms of data and data analytics have become central to surveil- lance and intelligence efforts. Further to transforming existing strategies like problem-oriented policing, they gave rise to new data-driven policing paradigms such as predictive policing, surveillance-based policing, and intelligence-led policing (Brayne, 2017; D. L. Carter & Carter, 2008; Ferguson, 2017c; Meijer & Wessels, 2019). Examples include the Real-Time Analysis Critical Response (RACR) Division of the Los Angeles Police Department and the Domain Awareness System (DAS) of the New York Police Department (Ferguson, 2017c; Levine et al., 2017; Meijer & Wessels, 2019). These sys- tems integrate databases, automated license-plate readers, CCTV cameras, facial recognition technology, and envir- onmental sensors, incorporating various data types such as neighborhood histories, repeat offenders, gang affilia- tions, crime records, vehicle ownership, personal infor- mation, and environmental sensor data.
A framework of data-driven policing
The review findings were consolidated around different data types and data-driven processes, which were sub- sequently mapped onto the core policing functions, as visualized in the framework shown in Figure 5 and explained below.
Data types
The literature brings forth three broader data types deployed in policing: directed, automated, and crowd- sourced data. Directed data are intentionally captured by police officers or other public and private organiza- tions. The former include crime and disorder records, crash reports, offender profiles, service calls, commu- nity intelligence, warrants and summons (Adderley & Musgrove, 2001; Brayne, 2017; Hardyns & Rummens, 2017; Versteegh et al., 2013). The latter include weather patterns, socio-demographics, infrastructure condi- tions, school schedules, traffic patterns, accident and emergency calls and passport data (Egbert, 2019; Hardyns & Rummens, 2017).
Automated data are collected by sensors and smart electronic devices or systems used by the police or the public like smart CCTVs, body-worn cameras, dash- board cameras, automatic number plate readers (ANPRs), smart-phones, smart cards and environmen- tal sensors (Brayne, 2020; Chan & Moses, 2017; Duxbury & Andrabi, 2022; Levine et al., 2017).
Crowdsourced data are captured through open- source devices, platforms or systems, with the most substantial volumes originating primarily from social networking sites like Facebook, YouTube and Twitter (Burnap et al., 2015; Rønn et al., 2021; Williams et al., 2013, 2017).
Data-driven processes
Table 3 summarizes the various analytical techniques applied for extracting information from different data sources. Depending upon the nature of the information extracted, data-driven processes can be categorized into crime construction, crime sensing, surveillance automa- tion, and disorder automation. Their applications were found in the policing functions of law enforcement and order maintenance, with fewer applications related to service provision.
Crime construction Crime construction involves using directed data to proactively visualize crime clusters and criminal beha- vior, prioritize targets, and reallocate resources for
416 M. AFZAL AND P. PANAGIOTOPOULOS
A nt
ec ed
en ts
D at
a
In st
it ut
io na
l (P
ol iti
ca l &
re gu
la to
ry e
nv ir
on m
en t,
D at
a sh
ar in
g, In
te ro
pe ra
bi lit
y, E
xt er
na l
co lla
bo ra
tio ns
e tc
)
D ir
ec te
d D
at a
A ut
om at
ed D
at a
C ro
w d-
so ur
ce d
D at
a
D at
a- dr
iv en
P ro
ce ss
es C
ri m
e C
on st
ru ct
io n
C ri
m e
Se ns
in g
Su rv
ei lla
nc e
A ut
om at
io n
D is
or de
r A
ut om
at io
n
C or
e po
lic in
g fu
nc ti
on s
L aw
E nf
or ce
m en
t
O rd
er -m
ai nt
en an
ce
Se rv
ic e
Pr ov
is io
n
O rg
an iz
at io
na l
(T ec
hn ol
og ic
al d
if fe
re nt
ia ls
b et
w ee
n an
al ys
ts a
nd s
w or
n of
fi ce
rs , C
iv ili
an
an al
ys ts
’ i nt
eg ra
tio n
in to
m ai
ns tr
ea m
, Sl
ac k
re so
ur ce
s, L
ea de
rs hi
p, P
ol ic
y- ro
ut in
e co
ng ru
en cy
et c)
In di
vi du
al (A
na ly
st s’
S ki
ll- le
ve l,
A na
ly st
s’ ro
le
cl ar
ity ,L
im ita
tio ns
o f d
at a
co lle
ct io
n an
d da
ta a
cc es
s et
c)
Fi gu
re 5
. F ra
m ew
or k
of d
at a-
dr iv
en p
ol ic
in g.
INTERNATIONAL JOURNAL OF PUBLIC ADMINISTRATION 417
Ta bl
e 3.
D at
a- dr
iv en
p ro
ce ss
es a
nd il
lu st
ra tiv
e ex
am pl
es .
Pr oc
es se
s D
at a
so ur
ce s
In di
ca tiv
e ap
pl ic
at io
ns In
te rv
en tio
n( s)
Co re
p ol
ic in
g fu
nc tio
n( s)
Re fe
re nc
es
Cr im
e co
ns tr
uc ti
on ●
Ex pl
or at
or y
cr im
e m
ap pi
ng ●
Cr im
e st
at is
tic s
● Co
m pS
ta t
● Ci
tiS ta
t ●
M AP
S ●
M ap
In fo
● G
eo D
as h
● Ar
cV ie
w /G
IS ●
CP AS
● D
ire ct
ed
pa tr
ol lin
g ●
En vi
ro nm
en ta
l ●
Co m
m un
ity
O ut
re ac
h
La w
e nf
or ce
m en
t Ar
ie l e
t al
. ( 20
16 ) C
ap la
n et
a l.
(2 02
1) O
at le
y et
a l.
(2 01
9)
Ra tc
liff e
(2 00
0) A
dd er
le y
an d
M us
gr ov
e (2
00 1)
B eh
n (2
00 8)
● Ex
pl or
at or
y pr
of ili
ng -d
o- ●
O RI
O N
● D
ire ct
ed
pa tr
ol lin
g ●
Pr ev
en ta
tiv e
ar re
st
La w
e nf
or ce
m en
t Ra
tc liff
e an
d Ki
ku ch
i ( 20
19 ) A
dd er
le y
an d
M us
gr ov
e (2
00 1)
● Pr
ed ic
tiv e
cr im
e m
ap pi
ng (R
is k
te rr
ai n
m od
el in
g, N
ea r
re pe
at s)
● Cr
im e
st at
is tic
s ●
So ci
o- de
m og
ra ph
ic s
● Tr
an sp
or t
hu bs
/t ra
ns it
ce nt
er s
● W
ea th
er P
at te
rn s
● G
an g
da ta
ba se
s ●
So ci
al m
ed ia
● Pr
ed Po
l ●
H un
ch L
ab ●
PR EC
O Bs
● Bl
ue Cr
us h
● AR
M O
R ●
Q lik
S en
se ●
Kr im
Pr o
● Ke
yC rim
e ●
CA S
● D
ire ct
ed
pa tr
ol lin
g La
w e
nf or
ce m
en t
Be nb
ou zi
d (2
01 9)
M oh
le r
et a
l. (2
01 5)
M os
es a
nd C
ha n
(2 01
8) A
nd re
je vi
c et
a l.
(2 02
0) L
or en
z et
a l.
(2 02
1)
Br ay
ne a
nd C
hr is
tin (2
02 1)
Ja in
e t a
l. (2
01 0)
M ei
je r e
t a l.
(2 02
1)
● Pr
ed ic
tiv e
pr of
ili ng
(e .g
., us
in g
m ac
hi ne
le ar
ni ng
o r
ra nd
om fo
re st
s) -d
o- ●
Pr oK
id ●
H AR
T ●
D AS
H ●
CR AF
● D
ire ct
ed
pa tr
ol lin
g ●
D et
er re
nc e
ta ct
ic s
La w
e nf
or ce
m en
t O
sw al
d et
a l.
(2 01
8) S
au nd
er s
et a
l. (2
01 6)
W ie
nt je
s et
a l.
(2 01
7) S
pi va
k et
a l.
(2 02
1)
Cr im
e se
ns in
g ●
Pu bl
ic In
te rv
ie w
s ●
So ci
al m
ed ia
● iN
SI ●
eP O
O LI
CE La
w e
nf or
ce m
en t
O rd
er -m
ai nt
en an
ce Ed
w ar
ds a
nd U
rq uh
ar t
(2 01
6) L
ow e
an d
In ne
s (2
01 2)
● Si
gn al
c rim
es p
er sp
ec tiv
e ●
D ire
ct ed
pa
tr ol
lin g
● En
vi ro
nm en
ta l
Su rv
ei lla
nc e
au to
m at
io n
● Se
ns or
a le
rt in
g (T
hr es
ho ld
-lo gi
c ru
le s,
o pt
ic al
c ha
ra ct
er re
co gn
iti on
, ac
ou st
ic s
co rr
el at
io ns
, m ac
hi ne
le ar
ni ng
fo r p
at te
rn
re co
gn iti
on )
● CC
TV s
● AN
PR s
● En
vi ro
nm en
ta l s
en so
rs ●
Sh ot
-s po
tt er
s ●
BW Cs
e tc
● D
AS ●
RA CR
● D
ire ct
ed
pa tr
ol lin
g ●
Re sp
on se
-t im
e ●
D et
ec tiv
e- ar
re st
La w
e nf
or ce
m en
t Le
vi ne
e t
al . (
20 17
)W iig
(2 01
8) B
ul lo
ck e
t al
. ( 20
20 )
● Ra
di o-
ce ll
an al
ys is
(D at
a m
in in
g) ●
Ca ll
re co
rd s
● eF
AS ●
i2 AN
● D
et ec
tiv e-
ar re
st La
w e
nf or
ce m
en t
Ku bl
er (2
01 7)
P aa
sc he
(2 01
3) A
dd er
le y
an d
M us
gr ov
e (2
00 1)
● So
ci al
m ed
ia m
ap pi
ng ●
So ci
al m
ed ia
● EU
V irt
uo so
● Cl
ea rV
ie w
A I
● D
et ec
tiv e-
ar re
st La
w e
nf or
ce m
en t
Tr ot
tie r
(2 01
7) P
au ls
en (2
02 0)
D is
or de
r au
to m
at io
n (N
at ur
al la
ng ua
ge p
ro ce
ss in
g, F
or m
al c
on ce
pt a
na ly
si s,
D
ee p
le ar
ni ng
) ●
So ci
al m
ed ia
● G
an g
da ta
ba se
s ●
CO SM
O S
● EM
O TI
VE ●
AT H
EN A
● H
at er
N et
● Co
m m
un ity
O
ut re
ac h
● D
et er
re nc
e ta
ct ic
s ●
En vi
ro nm
en ta
l
O rd
er -m
ai nt
en an
ce Bu
rn ap
e t
al . (
20 15
) D om
do uz
is e
t al
. ( 20
16 ) P
er ei
ra -
Ko ha
ts u
et a
l. (2
01 9)
418 M. AFZAL AND P. PANAGIOTOPOULOS
crime prevention in an exploratory or predictive way (Santos, 2014).
Exploratory crime construction uses spatial, spatio- temporal and network analysis techniques for crime mapping and criminal profiling. Exploratory crime mapping identifies hotspots with high victimization risk or crime density using directed data like reported crime and calls for service (Ariel et al., 2016; Bowers et al., 2004; Innes et al., 2005). Typical examples include approaches designated as hotspot policing (Clavell, 2018; Ratcliffe & McCullagh, 2001; Rydberg et al., 2018; Versteegh et al., 2013) and problem-oriented poli- cing (Nunn & Quinet, 2002; Uchida & Swatt, 2013; White & Katz, 2013).
Exploratory criminal profiling employs offender data to calculate harm scores to identify potentially harmful offenders. A common technique is a weighted aggregation of static factors like the offense gravity, sentence severity and time decay (Marciniak, 2023; Ratcliffe & Kikuchi, 2019). In the case of organized crimes, link and network analyses are used. Link analy- sis is employed to develop association matrices and link diagrams, and network analysis is used to identify the existence of a criminal organization and its central fig- ure (Adderley & Musgrove, 2001; Harper & Harris, 1975; Schroeder et al., 2007). Network analysis generates metrics like capital scores for identifying influential targets within the gangs (Adderley & Musgrove, 2001; Hashimi & Bouchard, 2017; Innes et al., 2005).
Predictive crime construction uses machine learning tools for crime mapping and criminal profiling. Predictive crime mapping involves forecasting crimes within micro-locations called dynamic hotspots (Mohler et al., 2015). Commonly used techniques include risk terrain and near repeat modeling of spatio- temporal crime and socio-demographic data (Moses & Chan, 2018). Predictive profiling incorporates static and dynamic risk factors to predict a person’s future offend- ing tendency by calculating the predictive risk scores available to police officers on their dashboards (Kakade, 2020; Oswald et al., 2018; Saunders et al., 2016; Wientjes et al., 2017).
Numerous applications and automated risk-assess- ment tools for crime construction have been developed through police-private collaboration (see Table 3) (Brayne, 2020). These collaborations support police authorities in implementing intervention strategies for violence prevention (murder and armed assault) and property crimes (burglary, motor vehicle theft and rob- bery). Outcomes include prominent interventions like environmental or target hardening (removal of graffiti, addition or fixation of lighting, altering bus routes, posting signs prohibiting trespassing, sanctioning illegal
sub-letting or rooming, etc.), enforcement (increased surveillance, home visits, patrolling, etc.) and outreach interventions (community recreational events, tackling homelessness, etc.) (Braga & Schnell, 2013; Brayne, 2020). Predictive techniques have been limited to enfor- cement interventions, while exploratory crime con- struction has been tested through different types of interventions (see Table 3).
Quantitative evaluations of crime construction appli- cations have produced mixed results. Some found no significant crime reductions for enforcement interven- tions (Chilvers & Weatherburn, 2004; Galiani & Jaitman, 2023; Massarotti, 2012; Ratcliffe et al., 2021; Rydberg et al., 2018; Worrall, 2016). In contrast, others found differential decreases in violence and property crime for enforcement and multi-prong interventions and positive spill-over on collective efficacy and infor- mal social control (Badiora & Ntamark, 2020; Bond & Hajjar, 2013; Braga & Schnell, 2013; Braga et al., 2019; Caplan et al., 2021; Kennedy et al., 2022; Kochel & Weisburd, 2019; Mazerolle et al., 2007; Meijer & Wessels, 2019; Mohler et al., 2015; Saunders et al., 2016; R. B. Taylor & Ratcliffe, 2020; Uchida & Swatt, 2013; Weisburd et al., 2015).
Crime sensing Crime sensing involves proactively gathering commu- nity intelligence from the public and social media to extract physical and social indicators for developing area signal profiles of crime and disorder events. These include indicators like excessive littering, graffiti and verbal harassment that encourage local criminals and attract outside offenders (Skogan, 2015).
Public interviews and social media constitute essen- tial data sources of disorder indicators. Computer- assisted personal interviewing software, like i-NSI, are used to record public interviews, extract disorder indi- cators and geocode them onto neighborhood maps for developing area profiles comprising crime, fear-and- avoidance and societal-tension hotspots (Innes et al., 2009; Lowe & Innes, 2012). The Lancashire and Sutton police departments in the UK have implemented this technique to detect signals of societal tensions (Innes et al., 2009). An enforcement intervention achieved a 28% reduction in signal events across risk hotspots (Lowe & Innes, 2012).
The interview-based instruments are costly and can capture cross-sectional data only. Social media data present a low-cost alternative for longitudinal analysis (Williams et al., 2017). Social media posts may contain witness accounts of physical and social disorder events in their neighborhoods or otherwise relate to live and past incidents of social disorder. The intensity of such
INTERNATIONAL JOURNAL OF PUBLIC ADMINISTRATION 419
tweets and posts is associated with the offline incidence of burglary, criminal damage and violence (Williams et al., 2017). Integrating police records and social media data has been used to identify patterns of human smuggling and drug trafficking (Edwards & Urquhart, 2016).
Surveillance automation Surveillance automation is the application of data mining tools and visual analytics to automated and crowdsourced data for law enforcement through real- time monitoring. Prominent approaches include sensor alerting, automated pattern recognition, radio-cell ana- lysis and social media mapping.
Sensor alerting uses video analytics, optical character recognition and acoustics correlations to generate alerts from CCTVs, environmental sensors, ANPRs and gun- fire alert technologies (Brayne, 2020; Levine et al., 2017). These sensors are usually integrated with pattern recog- nition software that produces spatiotemporal routing patterns for detecting and predicting the future loca- tions of suspected or under-watch vehicles (Levine et al., 2017). The domain awareness system (DAS) installed in New York is a classic example of an integrated system that helped reduce police response time, prevent kid- nappings and arrest offenders redhandedly (Bullock et al., 2020; Levine et al., 2017; Young et al., 2019).
Radio-cell analysis refers to mining mobile phone data to identify suspicious connections involved in riot- ing and heinous crimes like murder and kidnapping (Adderley & Musgrove, 2001; Lee, 2003; Paasche, 2013). Human investigators delve deeper into these suspicious connections to establish potential links to crime or rioting. Using this technique, the German police successfully traced 379 suspects related to the 2011 Dresden riots (Paasche, 2013).
Automated social mapping involves analyzing social media data of suspected individuals, groups, or events during criminal investigations. Commonly used techni- ques include temporal mapping (like a Facebook time- line), facial mapping (photos and images in conjunction with facial recognition technology (FRT) using data mining tools), and relational mapping (connection and communication patterns) (Andrews, 2019; Edwards & Urquhart, 2016; Paulsen, 2020; Trottier, 2017; Wirtz et al., 2019).
Only a few formal impact evaluations of automated surveillance processes are available and limited to sensor alerting and FRT. For instance, traffic enforcement using smart CCTVs significantly increased seat belt compliance in Shanghai (Li et al., 2018), and sensor alerting helped to reduce crime and response time in Camden (USA) (Wiig, 2018). Various police forces
worldwide have employed FRT to identify rioters, but it reinforced black-white arrest disparities in the USA (Busuioc, 2021; Johnson et al., 2022; Trottier, 2017).
Disorder automation Disorder automation applies a combination of data mining and machine learning techniques to crowd- sourced data, mainly from social media, for assessing communal tensions and identifying groups and actors involved in sparking social unrest. Natural language processing (NLP), formal concept analysis and double deep learning are commonly used tools (Burnap et al., 2015; Domdouzis et al., 2016; Pereira-Kohatsu et al., 2019; Williams et al., 2013).
Prominent platforms and applications include the Cardiff Online Social Media Observatory (COSMOS), EMOTIVE, ATEHNA and HaterNet. COSMOS accu- rately measured racial tensions using NLP following the 2011 Evra-Suarez abuse incident (Edwards & Urquhart, 2016; Williams et al., 2013). ATHENA is a European crisis management system that uses formal concept analysis for data clustering and NLP to categorize senti- ment scores into positive, negative, or neutral (Domdouzis et al., 2016). HaterNet, a Spanish hate speech detection system, employs double deep learning for clustering, NLP for tweet classification, and social network analysis for pattern, actor and community identification through word clouds and ranked graphs (Pereira-Kohatsu et al., 2019). These applications are more recent developments, and our search did not yield empirical impact evaluations on core policing functions.
Antecedents of data-driven policing
A consolidated view of data in policing sets the context for establishing their implications for police manage- ment. Few articles have explicitly linked the integration of data applications into different policing functions. These connections are necessary to develop a better understanding of the complexities involved in the data-driven transformation of policing. We identified 69 relevant articles and grouped their insights into institutional, organizational, and individual levels. Depending on the context and application, different factors act as barriers or enablers of data-driven policing.
Institutional Police organizations operate in complex environments often firmly anchored upon public demands and media pressures (Trottier, 2015). Different cycles of positive and negative feedback affect public sensitivities and
420 M. AFZAL AND P. PANAGIOTOPOULOS
accordingly determine policy directions and funding decisions for data-driven policing (Clavell, 2018; Oswald et al., 2018; Sanders & Condon, 2017).
The benefits of data-driven policing have concen- trated on improved response times, better decisions through enhanced situational awareness, more precise risk assessments, increased police visibility in troubled areas, and even better transparency behind policy deci- sions (Brayne, 2017; Ratcliffe et al., 2020; Willis et al., 2007). The unintended consequences of new data sources and algorithms have been associated with amplifying racial, ethnic, income and other systematic biases embedded within the historical police data (Andrews, 2019; Brayne, 2020; Dencik et al., 2017; Eubanks, 2017; Moses & Chan, 2018; Wirtz et al., 2019). Public concerns undermine police legitimacy and exert political control, external accountability and data regulations (Edwards & Urquhart, 2016; Pasha, 2018). Greater access to data lowers the threshold for inclusion in surveillance schemes and raises arrest rates for petty offenses, especially for already marginalized groups (Brayne, 2017; Busuioc, 2021; Wiig, 2018). For instance, 80% of individuals in London’s gang databases are Black, which makes this group more vulnerable to policing interventions (Andrews, 2019; Densley & Pyrooz, 2020). Similarly, CompStat has been found to result in disproportionate targeting of nonwhite popu- lations (Pasha et al., 2023). These limitations have made police officers skeptical about applications under pre- dictive policing (Alon-Barkat & Busuioc, 2022; Rønn et al., 2021; Sandhu & Fussey, 2021; Urquhart & Miranda, 2022).
In a less explored but promising area, data advance- ments have been influential in the multijurisdictional nature of crime and criminal networks when multiple agencies join efforts (Adderley & Musgrove, 2001; Burcher & Whelan, 2018; Jackson & Brown, 2007; Lewandowski et al., 2018; Sheptycki, 2004). Such efforts, for example, have been prominent in the USA and Canada, where new state and regional data fusion cen- ters have been established. However, their performance has been limited by technological incompatibilities and data fragmentation (Barrett & Greene, 2023; Brayne, 2020; J. G. Carter & Phillips, 2015; Chang, 2015; Lewandowski et al., 2018). The lack of advanced tech- nical and data storage capabilities has stimulated colla- borations between police and international or private vendors, like Palantir and PredPol, as well as academic experts (Brayne, 2020; Clavell, 2018; Trottier, 2017).
Organizational Organizational antecedents underline the structural and cultural peculiarities of police organizations.
Realizing high expectations about data in policing requires investments to acquire resources like analyti- cal platforms. Lack of resources restricts essential upgrades to systems, upskilling of staff and reallocation to technical divisions (Behn, 2008; Burcher & Whelan, 2018; Dannels & Smith, 2001; Nunn, 2001). Smaller police organizations are often compelled to use out- dated technology and place “soon-to-retire” and “troubled coppers” as analysts with limited productiv- ity (Burcher & Whelan, 2018; Cope, 2004; Dannels & Smith, 2001; Darroch & Mazerolle, 2013; Nunn, 2001; Sanders & Condon, 2017; Sanders & Hannem, 2012). Larger police organizations usually have more slack resources to acquire modern equipment, hire profes- sional analysts, and train their workforce (J. G. Carter, 2016; J. G. Carter & Phillips, 2015; Nunn, 2001; Sheptycki, 2004).
Some studies have informed on the existence of ten- sions between analysts and sworn police officers that hinder the integration of data insights into decision- making (Burcher & Whelan, 2018; Sanders & Condon, 2017; Sanders & Hannem, 2012; B. Taylor et al., 2007). Working relationships between analysts and police offi- cers may be unstable and issue-driven because officers believe their experiential knowledge provides a better rationale for resource allocation than the outputs of data-driven tools and algorithms (Meijer et al., 2021; Ratcliffe et al., 2020; B. Taylor et al., 2007). These aspects challenge the development of specific career paths and incentives for integrating data analysts into mainstream policing (Belur & Johnson, 2018; Sanders & Condon, 2017; Trottier, 2015). Moreover, data professionals employ various arguments, like disconnect with citizens and organizational complexity, to distance themselves from or minimize their responsibilities (Fest et al., 2023).
Studies have also reported on leadership and man- agement factors. Effective leadership involves winning commitment by inspiring the staff through personal insights, coaching, fostering participation and acknowl- edging success rather than compliance (Behn, 2008; J. G. Carter & Phillips, 2015; Catte & Linden, 2021; Chang, 2015; Darroch & Mazerolle, 2013; Eterno & Silverman, 2006; Meijer et al., 2021; Ratcliffe, 2000). Open and decentralized management styles and clarity and coherence of goals spark innovative uses of data in policing (J. G. Carter & Phillips, 2015; Catte & Linden, 2021; Darroch & Mazerolle, 2013). Critical management decisions involve resolving tensions between granting data access for prompt decision-making and navigating data regulation hierarchies (Belur & Johnson, 2018; Lewandowski et al., 2018; Ratcliffe & McCullagh, 2001; Sanders & Hannem, 2012).
INTERNATIONAL JOURNAL OF PUBLIC ADMINISTRATION 421
Individual The prominence of data has brought challenges and opportunities in the policing profession. Fewer studies have focused on individuals within the data ecosystem, such as addressing the baseline data skills and qualifi- cations that police analysts need to develop with tech- nologies like database management, geographical information systems, and analytical products (Evans & Kebbell, 2012; Fest et al., 2023; O’Connor et al., 2022a; Weston et al., 2020). Various studies have found that police analysts and their supervisors strug- gle to keep their data literacy up-to-date and otherwise lack sufficient training and technological resources amidst the constant evolution of analytical systems, which has outpaced the skill level of civilian analysts (Belur & Johnson, 2018; Burcher & Whelan, 2018; J. G. Carter & Phillips, 2015; Weston et al., 2020). Consequently, the data workflows of analysts often meet limitations when working with legacy products and otherwise dealing with data quality and integrity issues due to errors and reporting delays (Barrett & Greene, 2023; Belur & Johnson, 2018; Busuioc, 2021; Chang, 2015; O’Connor et al., 2022b; Sanders & Condon, 2017).
Ethical and privacy considerations of data-driven policing
Alongside its potential benefits, data-driven policing is commonly associated with ethical and privacy concerns (Brayne, 2020; Busuioc, 2021; Eubanks, 2017; Ferguson, 2017b; O’Neil, 2016; Oswald et al., 2018). The prolifera- tion of real-time sensors, data interception techniques and algorithmic profiling for monitoring individuals, populations, and public spaces to identify potential threats have become commonplace, ushering in an era of heightened surveillance and surveillance-based poli- cing (Broeders et al., 2017; Ferguson, 2017a; Kubler, 2017; Urquhart & Miranda, 2022). This has raised ques- tions about the extent of individual privacy in public spaces (Eneman et al., 2022). Similarly, using predictive analytics to anticipate criminal behavior raises concerns about the presumption of innocence, unwarranted scru- tiny of individuals and expansion of surveillance lists (Duxbury & Andrabi, 2022; Urquhart & Miranda, 2022; Van Brakel, 2020). Social media monitoring of activists and dissenting voices jeopardizes free speech and demo- cratic freedoms (Egawhary, 2019; Owen, 2017). Moreover, numerous cases of data misuse, such as unauthorized access to personal information and the exploitation of data for discriminatory profiling pur- poses, have been reported (e.g., Brayne, 2020; Das, 2023; Oates, 2019).
Excessive dependence on historical data and proxies has provoked doubts about the reliability and accuracy of predictive policing interventions (Busuioc, 2021; Eubanks, 2017; O’Neil, 2016). Systematic flaws like underreporting, selective crime logging, and dispropor- tionate policing in minority neighborhoods compro- mise police data integrity (Behn, 2008; Craglia et al., 2005; Drenth & van Steden, 2020; Eterno et al., 2016; Kakade, 2020; Moy, 2021; O’Connor et al., 2022; Richardson et al., 2019; Sanders & Hannem, 2012; Sanders et al., 2015; Sheptycki, 2004; Thomas & Wolff, 2021). Predictive models relying on such biased or dirty data will likely provide prejudiced outputs or recom- mendations (see e.g., Richardson et al., 2019). Similarly, incorporating flawed proxies or imperfect indicators, like zip codes or minor crimes (e.g., nuisance) fre- quently occurring in impoverished areas, in risk-assess- ment algorithms to gauge criminal tendencies can result in the over-policing of individuals from specific ethnic or racial backgrounds (Dencik et al., 2017; O’Neil, 2016; Oswald et al., 2018). Over-reliance on such tools may overlook contextual factors, undermining the discretion of officers to cater to individual circumstances (Busuioc, 2021; Eneman et al., 2022). Algorithms may fall into detrimental feedback loops that are challenging to detect, break or rectify (Busuioc, 2021; O’Neil, 2016). For instance, an algorithm mislabeling a specific area as high crime may result in more arrests. This new training data will reinforce initial predictions, which become “self-fulfilling prophecies” (Busuioc, 2021, p. 826).
These privacy, accuracy and reliability issues under- score the need for transparency and accountability of predictive tools. However, the inherent opacity and complexity of algorithms complicate these aspects (Brayne, 2020; Oswald et al., 2018; Young et al., 2019). While the outputs of predictive algorithms are apparent —such as risk assessments or crime predictions—the implicit process and criteria it uses for predictions remain obscure for several reasons. First, deep learning algorithms and the underlying neural networks operate as black boxes, concealing the decision-making logic that is challenging even for system engineers to decipher (Busuioc, 2021; Oswald et al., 2018; Young et al., 2019). Second, predictive tools generally developed by private companies, like Palantir and PredPol, are shielded as trade secrets, and their inner workings are often with- held from public scrutiny (Brayne & Christin, 2021; Dencik et al., 2017; Egbert, 2019; Eneman et al., 2022; McQuade, 2016). Third, understanding the interplay between hundreds or thousands of features contributing to predictive outcomes is daunting (Brayne, 2020; Busuioc, 2021; Oswald et al., 2018). The technical exper- tise required to assess their validity is a barrier for law
422 M. AFZAL AND P. PANAGIOTOPOULOS
enforcement officers (Brayne, 2020; Dencik et al., 2017; Oswald et al., 2018). Fourth, algorithmic systems embody the implicit value choices their designers make (Bovens & Zouridis, 2002; Busuioc, 2021; Joh, 2014; Young et al., 2019). These choices, whether prior- itizing accuracy over fairness or certain demographics over others, have profound implications for system out- comes but yet remain obscured from public view (Bovens & Zouridis, 2002; Busuioc, 2021; Young et al., 2019).
Thus, data-driven policing may create an illusion of neutrality and mathematical objectivity while encoding societal biases concealed in algorithmic black-boxes (O’Neil, 2016; Oswald et al., 2018). Police officers uncriti- cally relying on them without understanding the under- lying assumptions and feature interactions may perpetuate biases. For example, Chicago’s Strategic Subjects List (SSL) employed for predicting violence propensity has been cri- ticized for racial profiling and over-policing in minority neighborhoods (Saunders et al., 2016). The HART risk assessment model in Durham Constabulary fails to con- sider crucial aspects such as family circumstances and job significance, potentially perpetuating biases (Oswald et al., 2018). Despite efforts like end-user interfaces and indepen- dent audits, achieving a fully comprehensible explanation remains elusive due to the intricate nature of algorithms (Oswald et al., 2018). Algorithms used for predicting reci- divism in both bail and sentencing decisions have been flagged as biased against black defendants (Epp et al., 2017; Kakade, 2020; Shoub, 2022). Facial recognition algorithms have also been found to target minorities, leading to false arrests in marginalized communities (Johnson et al., 2022; Shoub, 2022).
Implications and future research perspectives
The review brings together evidence on the role of data as a pivotal force in transforming policing with several points of consideration for public administration scho- lars and practitioners. First, following the rapid expan- sion of algorithmic decision-making tools (like PredPol, ARMOR, HART, and Qlik Sense), big data technologies have become deeply integrated into law enforcement operations, especially planning interventions and resource allocation decisions (Andrejevic et al., 2020; Brayne, 2020; Levine et al., 2017; Marciniak, 2023; Peeters & Schuilenburg, 2018). Evidence of the effec- tiveness of these applications remains mixed. Human interpretation remains critical because of contextual complexity and—among others—biases embedded within data, data overload and data quality limitations, which otherwise can make police decision-making more opaque and exacerbate inequalities (Andrews, 2019;
Brayne, 2020; Burcher & Whelan, 2018; Busuioc, 2021; Clavell, 2018; Eubanks, 2017; Fussey et al., 2021). Future research could address the emerging interactions between algorithmic decision-making and the changing role of professional judgment in policing. New approaches can consider the active involvement of human agents to augment data-driven decisions in poli- cing, like the human-in-the-loop configurations (Vogl et al., 2020).
Second, data-driven applications are catalyzing the anticipated transition from street-level to screen-level and system-level policing, where computers and algo- rithms replace interactions with the public (Bovens & Zouridis, 2002; Egbert, 2019; Vogl et al., 2020). Findings remain inconclusive on whether the transition has enhanced operational efficiency, decision-making and control of street-level discretion (Marciniak, 2023; Meijer et al., 2021; Ratcliffe et al., 2020). While data and algorithms do challenge prevailing discretionary prac- tices, they have not impacted decision-making so as to visibly restrict street-level officers’ ability to apply tacit knowledge to real-world scenarios. Police officers will generally tend to resist data-driven technologies when they are used to control their discretion (Brayne, 2020). Future research can elaborate on the complexities of transitioning from street-level to system-level policing and the ensuing impact on organizational processes and policing functions. We propose that more attention is necessary at the individual level and the baseline data skills that policing training should aim to develop. For example, police officers in various studies described the effects of data-driven decision-making as deskilling or experiential discounting, reducing them to mere line workers (Andrejevic et al., 2020; Brayne, 2020; Wenzelburger et al., 2022; Young et al., 2019). To mitigate these effects, it is suggested to focus on new methods of real-time data updates and training. We recommend that training explores the best ways to integrate police officers’ cognitive biases and perceptions on tasks with different complexities. This, for example, would allow police offi- cers to build confidence in data-driven processes and outcomes beyond compliance-focused training.
Third, our knowledge of how data-driven transfor- mation occurs within police departments appears lim- ited. The shift from traditionally reactive practices like random patrolling to more proactive police interven- tions has led to new roles, hierarchies and competencies within police departments (Coulthart & Riccucci, 2022; Fest et al., 2023; Mergel et al., 2016). The costs and benefits of using police resources to deploy smart tech- nologies and new data infrastructures are not always clear. Future research can examine how police managers cope with tensions arising from an attitudinal and
INTERNATIONAL JOURNAL OF PUBLIC ADMINISTRATION 423
cultural aversion toward data-driven approaches, ana- lysts, and budgetary constraints and their underlying causes. These effects need better validation through longitudinal evaluations of mature technology imple- mentations across more police departments.
Fourth, surveillance and disorder automation are established applications enabling police to monitor pub- lic spaces. Empirical assessments have not yet looked at their influence on law enforcement and order mainte- nance. For instance, Levine et al. (2017) reported a 6% reduction in New York’s crime with the implementation of DAS but could not establish the causal effect between crime reduction and DAS. On the other hand, surveil- lance analytics and automation are widening the scope of suspicion and incrimination by normalizing the practice of tracking beyond active criminals to innocent civilians or bystanders engaged in public activities (Brayne, 2020; Clavell, 2018; Eneman et al., 2022; Urquhart & Miranda, 2022). This practice has evoked ethical apprehensions relating to improper sharing, unauthorized access, or data misuse for non-law enforcement purposes (Brayne, 2020; Clavell, 2018). Concerns are exacerbated by the involvement of enterprises providing software and cloud infrastructures for data storage (Brayne, 2020; Busuioc, 2021; Clavell, 2018). Scholars need to examine the impacts on police legitimacy and how public man- agers address these concerns at the organizational level.
Fifth, from the standpoint of core policing functions, the role of data vis-à-vis service provision has not been adequately explored, although assisting vulnerable indi- viduals and dealing with emergencies and accidents constitute approximately a third of street-level police activities (Langton et al., 2021; Millie, 2012; Wilson, 1968). Future research can evaluate the link between data-driven processes and service provision as an essen- tial function that entails non-punitive interventions, which may help to improve trust in police-citizen inter- actions and reinforce police legitimacy (Badiora & Ntamark, 2020; Brayne, 2020).
Conclusion
The review integrated research on the use of data in policing through the cross-examination of multidisci- plinary studies. Establishing the connection between data-driven processes and fundamental policing func- tions highlights the complexities of incorporating data into policing practice. The framework in Figure 5 out- lines a cohesive view of data as the building blocks upon which rest the pillars of exploration, sensing and antici- pation of criminal and disorder events. These pillars act as springboards to implement intervention strategies within crime prevention and order maintenance.
The review broadens the scope of data-driven poli- cing beyond the identification of novel data sources. It serves as a point of reference for public administration scholars to research the data-driven transformation of police management. Previous studies like Meijer and Wessels (2019) focused on predictive policing as the dominant data-driven policing application with ambig- uous effects. Our findings capture the heterogeneity of techniques and the diversity of ways in which data shapes policing processes and transforms the spectrum of suspicion-led interventions. Further to updating empirical evidence, the current review positions predic- tive policing as a unique crime construction approach separate from surveillance and disorder management.
The study is limited to the spectrum of articles reviewed that do not include conference publications and non-academic literature. Our focus on core policing functions does not fully capture data-driven police man- agement practices. For the same reasons, the review cannot take into consideration technical details and implementation challenges. Finally, heterogeneity in documentation across the literature of diverse policing traditions and sources of information results in different levels of reporting on how data are used in policing.
The consolidation of prevailing research holds signifi- cant importance within a domain that has rapidly gained attention. Current research designs are limited due to the absence of longitudinal studies and other difficulties in applying robust methods to real-life policing applica- tions. Most empirical evaluations come from the USA, with additional evidence originating predominantly from the developed world and limited comparative studies. Exploring other contexts will bring new perspectives on differences in institutional structures and socio-cultural values embedded in policing. For instance, police autho- rities in developing countries have comparatively more centralized structures, less vigorous data regulations, and more restricted resources, and they tend to experience lower trust from the public (Farrell et al., 2001; Goldsmith, 2016). Studies involving developing countries and more comparative contexts could yield important insights for understanding the data-driven transforma- tion of policing.
Disclosure statement
No potential conflict of interest was reported by the author(s).
ORCID
Muhammad Afzal http://orcid.org/0000-0003-3497-3399 Panos Panagiotopoulos http://orcid.org/0000-0001-5408- 4842
424 M. AFZAL AND P. PANAGIOTOPOULOS
References
References marked with an asterisk (*) indicate studies included in the systematic review
*Adderley, R. W., & Musgrove, P. (2001). Police crime record- ing and investigation systems – a user’s view. Policing: An International Journal of Police Strategies & Management, 24 (1), 100–114. https://doi.org/10.1108/13639510110382287
Agarwal, P. K. (2018). Public administration challenges in the world of AI and bots. Public Administration Review, 78(6), 917–921. https://doi.org/10.1111/puar.12979
Allard, S. W., Wiegand, E. R., Schlecht, C., Datta, A. R., Goerge, R. M., & Weigensberg, E. (2018). State agencies’ use of administrative data for improved practice: Needs, challenges, and opportunities. Public Administration Review, 78(2), 240–250. https://doi.org/10.1111/puar.12883
Alon-Barkat, S., & Busuioc, M. (2022). Human–AI interac- tions in public sector decision making: “Automation bias” and “selective adherence” to algorithmic advice. Journal of Public Administration Research & Theory, 33(1), 153–169. https://doi.org/10.1093/jopart/muac007
*Andrejevic, M., Dencik, L., & Treré, E. (2020). From pre-emption to slowness: Assessing the contrasting tem- poralities of data-driven predictive policing. New Media and Society, 22(9), 1528–1544. https://doi.org/10.1177/ 1461444820913565
Andrews, L. (2019). Public administration, public leadership and the construction of public value in the age of the algorithm and ‘big data’. Public Administration, 97(2), 296–310. https://doi.org/10.1111/padm.12534
*Ariel, B., Weinborn, C., & Sherman, L. W. (2016). “Soft” policing at hot spots—do police community support offi- cers work? A randomized controlled trial. Journal of Experimental Criminology, 12(3), 277–317. https://doi.org/ 10.1007/s11292-016-9260-4
Badiora, A. I., & Ntamark, J. J. (2020). Dynamics of public support for community policing (CP): Findings from high-crime and low-crime residential neighborhoods. International Journal of Public Administration, 43(16), 1397–1412. https://doi.org/10.1080/01900692.2019. 1669178
Barrett, K., & Greene, R. (2023). As the quantity of data explodes, quality matters. International Journal of Public Administration, 46(16), 1189–1190. https://doi.org/10. 1080/01900692.2023.2197171
*Behn, R. D. (2008). Designing Performancestat: Or what are the key strategic choices that a jurisdiction or agency must make when adapting the CompStat/CitiStat class of perfor- mance strategies? Public Performance & Management Review, 32(2), 206–235. https://doi.org/10.2753/ PMR1530-9576320202
*Belur, J., & Johnson, S. (2018). Is crime analysis at the heart of policing practice? A case study. Policing and Society, 28 (7), 768–786. https://doi.org/10.1080/10439463.2016. 1262364
*Benbouzid, B. (2019). To predict and to manage. Predictive policing in the United States. Big Data & Society, 6(1), 205395171986170. https://doi.org/10.1177/20539517198 61703
*Bond, B. J., & Hajjar, L. M. (2013). Measuring congruence between property crime problems and response strategies:
Enhancing the problem-solving process. Police Quarterly, 16(3), 323–338. https://doi.org/10.1177/1098611113497041
Bovens, M., & Zouridis, S. (2002). From street-level to system-level bureaucracies: How information and commu- nication technology is transforming administrative discre- tion and constitutional control. Public Administration Review, 62(2), 174–184. https://doi.org/10.1111/0033- 3352.00168
*Bowers, K. J., Johnson, S. D., & Pease, K. (2004). Prospective hot-spotting: The future of crime mapping? The British Journal of Criminology, 44(5), 641–658. https://doi.org/10. 1093/bjc/azh036
*Braga, A. A., & Schnell, C. (2013). Evaluating place-based policing strategies. Police Quarterly, 16(3), 339–357. https:// doi.org/10.1177/1098611113497046
*Braga, A. A., Turchan, B., Papachristos, A. V., & Hureau, D. M. (2019). Hot spots policing of small geo- graphic areas effects on crime. Campbell Systematic Reviews, 15(3), e1046. https://doi.org/10.1002/cl2.1046
Bratton, W. J., & Malinowski, S. (2008). Police performance management in practice: Taking COMPSTAT to the next level. Policing-An International Journal of Police Strategies & Management, 2(3), 259–265. https://doi.org/10.1093/ police/pan036
*Brayne, S. (2017). Big data surveillance: The case of policing. American Sociological Review, 82(5), 977–1008. https://doi. org/10.1177/0003122417725865
Brayne, S. (2020). Predict and surveil: Data, discretion and the future of policing. Oxford University Press.
*Brayne, S., & Christin, A. (2021). Technologies of crime prediction: The reception of algorithms in policing and criminal courts. Social Problems, 68(3), 608–624. https:// doi.org/10.1093/socpro/spaa004
Brodeur, J.-P. (1983). High policing and low policing: Remarks about the policing of political activities. Social Problems, 30(5), 507–520. https://doi.org/10.2307/800268
Broeders, D., Schrijvers, E., van der Sloot, B., van Brakel, R., de Hoog, J., & Hirsch Ballin, E. (2017). Big data and security policies: Towards a framework for regulating the phases of analytics and use of big data. Computer Law & Security Review, 33(3), 309–323. https://doi.org/10.1016/j.clsr.2017.03. 002
*Bullock, J., Young, M. M., Wang, Y.-F., Giest, S., & Grimmelikhuijsen, S. (2020). Artificial intelligence, bureau- cratic form, and discretion in public service. Information Polity, 25(4), 491–506. https://doi.org/10.3233/IP-200223
*Burcher, M., & Whelan, C. (2018). Intelligence-led policing in practice: Reflections from intelligence analysts. Police Quarterly, 22(2), 139–160. https://doi.org/10.1177/ 1098611118796890
*Burnap, P., Rana, O. F., Avis, N., Williams, M., Housley, W., Edwards, A., Sloan, L. (2015). Detecting tension in online communities with computational Twitter analysis. Technological Forecasting and Social Change, 95, 96–108. https://doi.org/10.1016/j.techfore.2013.04.013
*Busuioc, M. (2021). Accountable artificial intelligence: Holding algorithms to account. Public Administration Review, 81(5), 825–836. https://doi.org/10.1111/puar.13293
*Caplan, J. M., Kennedy, L. W., Drawve, G., & Baughman, J. H. (2021). Data-informed and place-based violent crime prevention: The Kansas City, Missouri
INTERNATIONAL JOURNAL OF PUBLIC ADMINISTRATION 425
risk-based policing initiative. Police Quarterly, 24(4), 438–464. https://doi.org/10.1177/10986111211003205
*Carter, J. G. (2016). Institutional pressures and isomorphism: The impact on intelligence-led policing adoption. Police Quarterly, 19(4), 435–460. https://doi.org/10.1177/ 1098611116639536
Carter, D. L., & Carter, J. G. (2008). Intelligence-led policing: Conceptual and functional considerations for public policy. Criminal Justice Policy Review, 20(3), 310–325. https://doi. org/10.1177/0887403408327381
*Carter, J. G., & Phillips, S. W. (2015). Intelligence-led poli- cing and forces of organisational change in the USA. Policing and Society, 25(4), 333–357. https://doi.org/10. 1080/10439463.2013.865738
*Catte, R., & Linden, R. (2021). Leadership and change in winnipeg’s smart policing initiative. Policing (Oxford), 15 (1), 181–196. https://doi.org/10.1093/police/pay077
*Chang, J. (2015). The maturity model in practice: Chaos, reaction, evolution, and inclusion in law enforcement ana- lytic models. Police Practice & Research, 16(2), 108–120. https://doi.org/10.1080/15614263.2014.972617
*Chan, J., & Moses, L. B. (2017). Making sense of big data for security. British Journal of Criminology, azw059. https://doi. org/10.1093/bjc/azw059
*Chilvers, M., & Weatherburn, D. (2004). The New South Wales “Compstat” PROCESS: Its impact on crime. Australian & New Zealand Journal of Criminology, 37(1), 22–48. https://doi.org/10.1375/acri.37.1.22
*Clavell, G. G. (2018). Exploring the ethical, organisational and technological challenges of crime mapping: A critical approach to urban safety technologies. Ethics and Information Technology, 20(4), 265–277. https://doi.org/ 10.1007/s10676-018-9477-1
*Colton, K. W. (1979). The impact and use of computer technology by the police. Communications of the ACM, 22 (1), 10–20. https://doi.org/10.1145/359046.359049
*Cope, N. (2004). ‘Intelligence-led policing or policing-led intelligence’: Integrating volume crime analysis into poli- cing. The British Journal of Criminology, 44(2), 188–203. https://doi.org/10.1093/bjc/44.2.188
*Coulthart, S., & Riccucci, R. (2022). Putting big data to work in government: The case of the United States border patrol. Public Administration Review, 82(2), 280–289. https://doi. org/10.1111/puar.13431
*Craglia, M., Haining, R., & Signoretta, P. (2005). Modelling high-intensity crime areas: Comparing police perceptions with offence/offender data in sheffield. Environment and Planning A: Economy and Space, 37(3), 503–524. https:// doi.org/10.1068/a36143
*Dannels, D., & Smith, H. (2001). Implementation challenges of intelligence-led policing in a quasi-rural county. Journal of Crime and Justice, 24(2), 103–112. https://doi.org/10. 1080/0735648X.2001.9721137
*Danziger, J. N., & Kraemer, K. L. (1985). Computerized data-based systems and productivity among professional workers: The case of detectives. Public Administration Review, 45(1), 196–209. https://doi.org/10.2307/3110149
*Darroch, S., & Mazerolle, L. (2013). Intelligence-led policing: A comparative analysis of organizational factors influen- cing innovation uptake. Police Quarterly, 16(1), 3–37. https://doi.org/10.1177/1098611112467411
Das, S. (2023). Revealed: Metropolitan police shared sensitive data about crime victims with Facebook. The Guardian.
*Dencik, L., Hintz, A., & Carey, Z. (2017). Prediction, pre-emption and limits to dissent: Social media and big data uses for policing protests in the United Kingdom. New Media & Society, 20(4), 1433–1450. https://doi.org/10. 1177/1461444817697722
*Densley, J. A., & Pyrooz, D. C. (2020). The matrix in context: Taking stock of police gang databases in London and beyond. Youth Justice, 20(1–2), 11–30. https://doi.org/10. 1177/1473225419883706
*Domdouzis, K., Akhgar, B., Andrews, S., & Gibson, H. (2016). A social media and crowdsourcing data mining system for crime prevention during and post-crisis situations. Journal of Systems & Information Technology, 18(4), 364–382. https://doi.org/10.1108/JSIT-06-2016-0039
*Drenth, A. R., & van Steden, R. (2020). Everyday patrol work for a data-driven flying squad: Advancing theoretical think- ing on police craftsmanship in interacting with civilians. Journal of Crime and Justice, 43(4), 486–501. https://doi. org/10.1080/0735648X.2020.1722202
*Duxbury, S. W., & Andrabi, N. (2022). The boys in blue are watching you: The shifting metropolitan landscape and big data police surveillance in the United States. Social Problems. https://doi.org/10.1093/socpro/spac044
*Edwards, L., & Urquhart, L. (2016). Privacy in public spaces: What expectations of privacy do we have in social media intelligence? International Journal of Law and Information Technology, 24(3), 279–310. https://doi.org/10.1093/ijlit/ eaw007 .
*Egawhary, E. M. (2019). The surveillance dimensions of the use of social media by UK police forces. Surveillance & Society, 17(1/2), 89–104. https://doi.org/10.24908/ss.v17i1/ 2.12916
*Egbert, S. (2019). Predictive policing and the platformization of police work. Surveillance & Society, 17(1/2), 83–88. https://doi.org/10.24908/ss.v17i1/2.12920
*Egbert, S., & Krasmann, S. (2020). Predictive policing: Not yet, but soon preemptive? Policing and Society, 30(8), 905–919. https://doi.org/10.1080/10439463.2019.1611821
*Eneman, M., Ljungberg, J., Raviola, E., Rolandsson, B., ČČAs, J., De Hert, P., Porcedda, M. G., & Raab, C. D. (2022). The sensitive nature of facial recognition: Tensions between the Swedish police and regulatory autho- rities 1. Information Polity, 27(2), 219–232. https://doi.org/ 10.3233/IP-211538
*Epp, C. R., Maynard-Moody, S., & Haider-Markel, D. (2017). Beyond profiling: The institutional sources of racial dispa- rities in policing. Public Administration Review, 77(2), 168–178. https://doi.org/10.1111/puar.12702
*Eterno, J. A., & Silverman, E. B. (2006). The New York City police department’s Compstat: Dream or nightmaRE? International Journal of Police Science & Management, 8 (3), 218–231. https://doi.org/10.1350/ijps.2006.8.3.218
*Eterno, J. A., Verma, A., & Silverman, E. B. (2016). Police manipulations of crime reporting: Insiders’ revelations. Justice Quarterly, 33(5), 811–835. https://doi.org/10.1080/ 07418825.2014.980838
Eubanks, V. (2017). Automating inequality: How high-tech tools profile, police, and punish the poor (1st ed.). St. Martin’s Press.
426 M. AFZAL AND P. PANAGIOTOPOULOS
*Evans, J. M., & Kebbell, M. R. (2012). The effective analyst: A study of what makes an effective crime and intelligence analyst. Policing and Society, 22(2), 204–219. https://doi. org/10.1080/10439463.2011.605130
Farrell, G., Lane, E., Clark, K., & Tseloni, A. (2001). What does the world spend on policing? International Journal of Comparative Sociology, 42(1/2), 59–73. https://doi.org/10. 1177/002071520104200103
Ferguson, A. G. (2017a). Big data surveillance: the conver- gence of big data and law enforcement. In D. Gray & S. E. Henderson (Eds.), The Cambridge handbook of sur- veillance law (pp. 171–197). Cambridge University Press.
Ferguson, A. G. (2017b). Big data’s watchful eye: The rise of data surveillance. In The rise of big data policing (pp. 7–19). NYU Press.
Ferguson, A. G. (2017c). Introduction: Big data policing. In The rise of big data policing (pp. 1–6). NYU Press.
*Fest, I., Schäfer, M., van Dijck, J., & Meijer, A. (2023). Understanding data professionals in the police: A qualitative study of system-level bureaucrats. Public Management Review, 25(9), 1–21. https://doi.org/10.1080/ 14719037.2023.2222734
*Fussey, P., Davies, B., & Innes, M. (2021). ‘Assisted’ facial recognition and the reinvention of suspicion and discretion in digital policing. The British Journal of Criminology, 61 (2), 325–344. https://doi.org/10.1093/bjc/azaa068
*Galiani, S., & Jaitman, L. (2023). Predictive policing in a developing country: Evidence from two randomized con- trolled trials. Journal of Quantitative Criminology, 39(4), 805–831. https://doi.org/10.1007/s10940-022-09551-y
Goldsmith, A. (2016). Police reform and the problem of trust. Theoretical Criminology, 9(4), 443–470. https://doi.org/10. 1177/1362480605057727
Hardyns, W., & Rummens, A. (2017). Predictive policing as a new tool for law enforcement? Recent developments and challenges. European Journal on Criminal Policy and Research, 24(3), 201–218. https://doi.org/10.1007/s10610- 017-9361-2
*Harper, W. R., & Harris, D. H. (1975). The application of link analysis to police intelligence. Human Factors: The Journal of the Human Factors & Ergonomics Society, 17(2), 157–164. https://doi.org/10.1177/001872087501700206
*Hashimi, S., & Bouchard, M. (2017). On to the next one? Using social network data to inform police target prioritization. Policing: An International Journal of Police Strategies & Management, 40(4), 768–782. https://doi.org/ 10.1108/PIJPSM-06-2016-0079
Heaton, R. (2000). The prospects for intelligence‐led policing: Some historical and quantitative considerations. Policing and Society, 9(4), 337–355. https://doi.org/10.1080/ 10439463.2000.9964822
*Innes, M., Abbott, L., Lowe, T., & Roberts, C. (2009). Seeing like a citizen: Field experiments in ‘community intelli- gence‐led policing’. Police Practice & Research, 10(2), 99–114. https://doi.org/10.1080/15614260802264545
*Innes, M., Fielding, N., & Cope, N. (2005). ‘The appliance of science?’: The theory and practice of crime intelligence analysis. The British Journal of Criminology, 45(1), 39–57. https://doi.org/10.1093/bjc/azh053
*Jackson, A. L., & Brown, M. (2007). Ensuring efficiency, interagency cooperation, and protection of civil liberties: Shifting from a traditional model of policing to an
intelligence‐led policing (ILP) paradigm. Criminal Justice Studies, 20(2), 111–129. https://doi.org/10.1080/ 14786010701396855
*Jain, M., Tsai, J., Pita, J., Kiekintveld, C., Rathi, S., Tambe, M., & Ordóñez, F. (2010). Software assistants for randomized patrol planning for the LAX airport police and the federal air marshal service. Interfaces, 40(4), 267–290. https://doi. org/10.1287/inte.1100.0505
Janssen, M., & Kuk, G. (2016). The challenges and limits of big data algorithms in technocratic governance. Government Information Quarterly, 33(3), 371–377. https://doi.org/10. 1016/j.giq.2016.08.011
*Joh, E. E. (2014). Policing by numbers: Big data and the fourth amendment. Wash. L. Rev, 89, 35.
*Johnson, T. L., Johnson, N. N., McCurdy, D., & Olajide, M. S. (2022). Facial recognition systems in policing and racial disparities in arrests. Government Information Quarterly, 39(4), 101753. https://doi.org/10.1016/j.giq.2022.101753
*Kakade, N. (2020). Sloshing through the factbound morass of reasonableness: Predictive algorithms, racialized policing, and fourth amendment use of force. George Washington Law Review, 88(3), 788–818
*Kennedy, L. W., Caplan, J. M., & Drawve, G. (2022). Data- informed crime prevention at convenience stores in Atlantic City. Police Practice & Research, 23(2), 125–142. https://doi.org/10.1080/15614263.2021.1917399
Kitchin, R. (2014). The data revolution: Big data, open data, data infrastructures & their consequences. SAGE Publications.
*Kochel, T. R., & Weisburd, D. (2019). The impact of hot spots policing on collective efficacy: Findings from a randomized field trial. Justice Quarterly, 36(5), 900–928. https://doi.org/10.1080/07418825.2018.1465579
*Kubler, K. (2017). State of urgency: Surveillance, power, and algorithms in France’s state of emergency. Big Data & Society, 4(2), 205395171773633. https://doi.org/10.1177/ 2053951717736338
Langton, S., Bannister, J., Ellison, M., Haleem, M. S., & Krzemieniewska-Nandwani, K. (2021). Policing and mental ill-health: Using big data to assess the scale and severity of, and the frontline resources committed to, mental ill-health- related calls-for-service. Policing: A Journal of Policy and Practice, 15(3), 1963–1976. https://doi.org/10.1093/police/ paab035
Lee, L. T. (2003). Can police track your wireless calls - Call location information and privacy law. Cardozo Arts and Entertainment Law Journal, 21(Issues 2 & 3), 381–406.
*Levine, E. S., Tisch, J., Tasso, A., & Joy, M. (2017). The New York City police department’s domain awareness sys- tem. INFORMS Journal on Applied Analytics, 47(1), 70–84. https://doi.org/10.1287/inte.2016.0860
*Lewandowski, C., Carter, J. G., & Campbell, W. L. (2018). The utility of fusion centres to enhance intelligence-led policing: An exploration of end-users. Policing: A Journal of Policy and Practice, 12(2), 177–193. https://doi.org/10. 1093/police/pax005
Liederbach, J., & Travis, L. F. (2008). Wilson Redux: Another look at varieties of police behavior. Police Quarterly, 11(4), 447–467. https://doi.org/10.1177/1098611108314567
*Li, Q., Peng, J., Chen, T., Yu, Y., & Hyder, A. A. (2018). Seatbelt wearing rate in a Chinese city: Results from multi-round cross-sectional studies. Accident; Analysis
INTERNATIONAL JOURNAL OF PUBLIC ADMINISTRATION 427
and Prevention, 121, 279–284. https://doi.org/10.1016/j. aap.2018.09.027
*Lorenz, L., Meijer, A., & Schuppan, T. (2021). The algocracy as a new ideal type for government organizations: Predictive policing in Berlin as an empirical case. Information Polity: The International Journal of Government & Democracy in the Information Age, 26(1), 71–86. https://doi.org/10.3233/IP-200279
*Lowe, T., & Innes, M. (2012). Can we speak in confidence? Community intelligence and neighbourhood policing v2.0. Policing and Society, 22(3), 295–316. https://doi.org/10. 1080/10439463.2012.671823
Lum, C., & Koper, C. S. (2017). Evidence- based policing: Translating research into practice. Oxford University Press.
*Lyon, D. (2003). Technology vs ‘terrorism’: Circuits of city surveillance since September 11th. International Journal of Urban and Regional Research, 27(3), 666–678. https://doi. org/10.1111/1468-2427.00473
*Magers, J. S. (2004). Compstat: A new paradigm for policing or a repudiation of community policing? Journal of Contemporary Criminal Justice, 20(1), 70–79. https://doi. org/10.1177/1043986203262312
*Maguire, M. (2000). Policing by risks and targets: Some dimensions and implications of intelligence‐led crime con- trol. Policing and Society, 9(4), 315–336. https://doi.org/10. 1080/10439463.2000.9964821
Manning, P. K. (2008). The technology of policing: Crime mapping, information technology, and the rationality of crime control. New York University Press.
*Marciniak, D. (2023). Algorithmic policing: An exploratory study of the algorithmically mediated construction of indi- vidual risk in a UK police force. Policing and Society, 33(4), 449–463. https://doi.org/10.1080/10439463.2022.2144305
*Massarotti, M. (2012). Intelligence-led policing: The evalua- tion of the Denver Police department’s policy. Journal of Applied Security Research, 7(2), 268–283. https://doi.org/ 10.1080/19361610.2012.656259
*Mazerolle, L., Rombouts, S., McBroom, J., & Burke, R. (2007). The impact of COMPSTAT on reported crime in Queensland. Policing: An International Journal of Police Strategies & Management, 30(2), 237–256. https://doi.org/ 10.1108/13639510710753243
*McQuade, B. I. (2016). Police and the post-9/11 surveillance surge: “Technological dramas” in “the bureaucratic field”. Surveillance and Society, 14(1), 1–19. https://doi.org/10. 24908/ss.v14i1.5291
*Meijer, A., Lorenz, L., & Wessels, M. (2021). Algorithmization of bureaucratic organizations: Using a practice lens to study how context shapes predictive policing systems. Public Administration Review, 81(5), 837–846. https://doi.org/10.1111/puar.13391
Meijer, A., & Wessels, M. (2019). Predictive policing: Review of benefits and drawbacks. International Journal of Public Administration, 42(12), 1031–1039. https://doi.org/10. 1080/01900692.2019.1575664
Mergel, I., Rethemeyer, R. K., & Isett, K. (2016). Big data in public affairs. Public Administration Review, 76(6), 928–937. https://doi.org/10.1111/puar.12625
Millie, A. (2012). The policing task and the expansion (and contraction) of British policing. Criminology & Criminal Justice, 13(2), 143–160. https://doi.org/10.1177/ 1748895812466393
Moher, D., Liberati, A., Tetzlaff, J., Altman, D. G., & Group, P. (2009). Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. PLOS Medicine, 6 (7), 1–6. https://doi.org/10.1371/journal.pmed.1000097
*Mohler, G. O., Short, M. B., Malinowski, S., Johnson, M., Tita, G. E., Bertozzi, A. L., & Brantingham, P. J. (2015). Randomized controlled field trials of predictive policing. Journal of the American Statistical Association, 110(512), 1399–1411. https://doi.org/10.1080/01621459.2015.1077710
*Moses, L. B., & Chan, J. (2018). Algorithmic prediction in policing: Assumptions, evaluation, and accountability. Policing and Society, 28(7), 806–822. https://doi.org/10. 1080/10439463.2016.1253695
*Moy, L. M. (2021). A taxonomy of police technology’s racial inequity problems. U Ill L Rev, 139
*Nam, T. (2014). Assessing operational and collaborative per- formance management: A case study of phillystat. International Journal of Public Administration, 37(8), 514–527. https://doi.org/10.1080/01900692.2013.865648
*Northrop, A., Kraemer, K. L., & King, J. L. (1995). Police use of computers. Journal of Criminal Justice, 23(3), 259–275. https://doi.org/10.1016/0047-2352(95)00019-M
*Nunn, S. (2001). Police information technology: Assessing the effects of computerization on urban police functions. Public Administration Review, 61(2), 221–234. https://doi. org/10.1111/0033-3352.00024
*Nunn, S., & Quinet, K. (2002). Evaluating the effects of information technology on problem-oriented-policing: If it doesn’t fit, must we quit? Evaluation Review, 26(1), 81–108. https://doi.org/10.1177/0193841X02026001004
Oates, J. (2019). 237 UK police force staff punished for mis- using it systems in last 2 years. The Register.
*Oatley, G., S, W., Barnes, G. C., Clare, J., & Chapman, B. (2019). Crime concentration in Perth CBD: A comparison of officer predicted hot spots, data derived hot spots and officer GPS patrol data. Australian Journal of Forensic Sciences, 51(sup1), S136–S140. https://doi.org/10.1080/ 00450618.2019.1569141
*O’Connor, C. D., Ng, J., Hill, D., & Frederick, T. (2022a). Police analysts on the job in Canada: Work experiences, data work, and the move towards evidence-based policing. Police Practice & Research, 23(4), 458–472. https://doi.org/ 10.1080/15614263.2021.2022483
*O’Connor, C. D., Ng, J., Hill, D., & Frederick, T. (2022b). Thinking about police data: Analysts’ perceptions of data quality in Canadian policing. Police Journal, 95(4), 637–656. https://doi.org/10.1177/0032258X211021461
O’Malley, M. (2014). Doing what works: Governing in the age of big data. Public Administration Review, 74(5), 555–556. https://doi.org/10.1111/puar.12260
O’Neil, C. (2016). Weapons of math destruction: How big data increases inequality and threatens democracy. Crown Publishers.
*Oswald, M., Grace, J., Urwin, S., & Barnes, G. C. (2018). Algorithmic risk assessment policing models: Lessons from the Durham HART model and ‘experimental’ pro- portionality. Information & Communications Technology Law, 27(2), 223–250. https://doi.org/10.1080/13600834. 2018.1458455
*Owen, S. (2017). Monitoring social media and protest move- ments: Ensuring political order through surveillance and
428 M. AFZAL AND P. PANAGIOTOPOULOS
surveillance discourse. Social Identities, 23(6), 688–700. https://doi.org/10.1080/13504630.2017.1291092
*Paasche, T. F. (2013). Coded police territories: ‘Detective software’ investigates. Area, 45(3), 314–320. https://doi. org/10.1111/area.12033
Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D. , & Moher, D. (2021). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. Systematic Reviews, 10, 1. https://doi.org/10.1186/s13643-021-01626-4
*Pasha, O. (2018). Can performance management best prac- tices help reduce crime? Public Administration Review, 78 (2), 217–227. https://doi.org/10.1111/puar.12856
*Pasha, O., Kroll, A., & Ash, M. (2023). Assessing the equity and effectiveness of PerformanceStat systems. International Public Management Journal, 26(1), 23–45. https://doi.org/ 10.1080/10967494.2021.1918300
*Paulsen, J. E. (2020). AI, trustworthiness, and the digital Dirty Harry problem. The Nordic Journal of Studies in Policing, 8(2), 1–19. https://doi.org/10.18261/issn.2703- 7045-2021-02-02
*Peeters, R., & Schuilenburg, M. (2018). Machine justice: Governing security through the bureaucracy of algorithms. Information Polity, 23(3), 267–280. https://doi. org/10.3233/IP-180074 .
*Pereira-Kohatsu, J. C., Quijano-Sanchez, L., Liberatore, F., & Camacho-Collados, M. (2019). Detecting and monitoring hate speech in Twitter. Sensors, 19(21), 4654. https://doi. org/10.3390/s19214654
*Ratcliffe, J. H. (2000). Implementing and integrating crime mapping into a police intelligence environment. International Journal of Police Science & Management, 2 (4), 313–323. https://doi.org/10.1177/146135570000200404
*Ratcliffe, J. H., & Kikuchi, G. (2019). Harm-focused offender triage and prioritization: A Philadelphia case study. Policing an International Journal, 42(1), 59–73. https://doi.org/10. 1108/PIJPSM-08-2018-0118
*Ratcliffe, J. H., & McCullagh, M. J. (2001). Chasing ghosts? Police perception of high crime areas. The British Journal of Criminology, 41(2), 330–341. https://doi.org/10.1093/bjc/ 41.2.330
*Ratcliffe, J. H., Taylor, R. B., Askey, A. P., Thomas, K., Grasso, J., Bethel, K. J. . . . Koehnlein, J. (2021). The Philadelphia predictive policing experiment. Journal of Experimental Criminology, 17(1), 15–41. https://doi.org/ 10.1007/s11292-019-09400-2
*Ratcliffe, J. H., Taylor, R. B., & Fisher, R. (2020). Conflicts and congruencies between predictive policing and the patrol officer’s craft. Policing and Society, 30(6), 639–655. https://doi.org/10.1080/10439463.2019.1577844
*Richardson, R., Schultz, J. M., & Crawford, K. (2019). Dirty data, bad predictions: How civil rights violations impact police data, predictive policing systems, and justice. NYUL Rev Online, 94, 15
*Rønn, K. V., Rasmussen, B. K., Skou Roer, T., & Meng, C. (2021). On the perception and use of information from social media in investigative police work: Findings from a Scandinavian study. Policing: A Journal of Policy and Practice, 15(2), 1262–1273. https://doi.org/10.1093/police/ paaa028
Ruijer, E., Porumbescu, G., Porter, R., & Piotrowski, S. (2023). Social equity in the data era: A systematic literature review of
data-driven public service research. Public Administration Review, 83(2), 316–332. https://doi.org/10.1111/puar.13585
*Rydberg, J., McGarrell, E. F., Norris, A., & Circo, G. (2018). A quasi-experimental synthetic control evaluation of a place-based police-directed patrol intervention on violent crime. Journal of Experimental Criminology, 14(1), 83–109. https://doi.org/10.1007/s11292-018-9324-8
*Sanders, C. B., & Condon, C. (2017). Crime analysis and cognitive effects: The practice of policing through flows of data. Global Crime, 18(3), 237–255. https://doi.org/10. 1080/17440572.2017.1323637
*Sanders, C. B., & Hannem, S. (2012). Policing “The Risky”: Technology and surveillance in everyday patrol work. Canadian Review of Sociology, 49(4), 389–410. https://doi. org/10.1111/j.1755-618X.2012.01300.x
*Sanders, C. B., Weston, C., & Schott, N. (2015). Police inno- vations, ‘secret squirrels’ and accountability: Empirically studying intelligence-led policing in Canada. British Journal of Criminology, 55(4), 711–729. https://doi.org/10. 1093/bjc/azv008
*Sandhu, A., & Fussey, P. (2021). The ‘uberization of poli- cing’? How police negotiate and operationalise predictive policing technology. Policing and Society, 31(1), 66–81. https://doi.org/10.1080/10439463.2020.1803315
*Santos, R. B. (2014). The effectiveness of crime analysis for crime reduction: Cure or diagnosis? Journal of Contemporary Criminal Justice, 30(2), 147–168. https:// doi.org/10.1177/1043986214525080
*Saunders, J., Hunt, P., & Hollywood, J. S. (2016). Predictions put into practice: A quasi-experimental evaluation of Chicago’s predictive policing pilot. Journal of Experimental Criminology, 12(3), 347–371. https://doi.org/ 10.1007/s11292-016-9272-0
*Schroeder, J., Xu, J., Chen, H., & Chau, M. (2007). Automated criminal link analysis based on domain knowledge. Journal of the American Society for Information Science and Technology, 58(6), 842–855. https://doi.org/10.1002/asi.20552
*Sheptycki, J. (2004). Organizational pathologies in police intelligence systems: Some contributions to the lexicon of intelligence-led policing. European Journal of Criminology, 1(3), 307–332. https://doi.org/10.1177/1477370804044005
Sherman, L. W. (2013). The rise of evidence-based policing: Targeting, testing, and tracking. Crime and Justice, 42(1), 377–451. https://doi.org/10.1086/670819
*Shoub, K. (2022). Comparing systemic and individual sources of racially disparate traffic stop outcomes. Journal of Public Administration Research and Theory, 32(2), 236–251. https://doi.org/10.1093/jopart/muab029
*Silverman, E. B., & O’Connell, P. E. (1999). Organizational change and decision making in the New York city police department: A case study. International Journal of Public Administration, 22(2), 217–259. https://doi.org/10.1080/ 01900699908525382
Skogan, W. (2015). Disorder and decline: The state of research. Journal of Research in Crime and Delinquency, 52(4), 464–485. https://doi.org/10.1177/0022427815577836
*Spivak, B., McEwan, T., Luebbers, S., & Ogloff, J. (2021). Implementing evidence-based practice in policing family violence: The reliability, validity and feasibility of a risk assessment instrument for prioritising police response.
INTERNATIONAL JOURNAL OF PUBLIC ADMINISTRATION 429
Policing and Society, 31(4), 483–502. https://doi.org/10. 1080/10439463.2020.1757668
*Taylor, B., Kowalyk, A., & Boba, R. (2007). The integration of crime analysis into law enforcement agencies: An explora- tory study into the perceptions of crime analysts. Police Quarterly, 10(2), 154–169. https://doi.org/10.1177/ 1098611107299393
*Taylor, R. B., & Ratcliffe, J. H. (2020). Was the pope to blame? Statistical powerlessness and the predictive policing of micro-scale randomized control trials. Criminology and Public Policy, 19(3), 965–996. https://doi.org/10.1111/1745- 9133.12514
*Thomas, A. L., & Wolff, K. T. (2021). Crime distortion within the NYPD: A potential method for estimating crime mis- classification within CompStat statistics. Police Practice & Research, 22(4), 1390–1407. https://doi.org/10.1080/ 15614263.2020.1798236
*Trottier, D. (2015). Coming to terms with social media monitoring: Uptake and early assessment. Crime, Media, Culture: An International Journal, 11(3), 317–333. https:// doi.org/10.1177/1741659015593390
*Trottier, D. (2017). ‘Fear of contact’: Police surveillance through social networks. European Journal of Cultural and Political Sociology, 4(4), 457–477. https://doi.org/10. 1080/23254823.2017.1333442
*Uchida, C. D., & Swatt, M. L. (2013). Operation LASER and the effectiveness of hotspot patrol: A panel analysis. Police Quarterly, 16(3), 287–304. https://doi.org/10.1177/ 1098611113497044
*Urquhart, L., & Miranda, D. (2022). Policing faces: The present and future of intelligent facial surveillance. Information & Communications Technology Law, 31(2), 194–219. https://doi.org/10.1080/13600834.2021.1994220
Van Brakel, R. (2020). Rethinking predictive policing: Towards a holistic framework of democratic algorithmic surveillance. In M. Schuilenburg & R. Peeters (Eds.), The algorithmic society: Technology, power, and knowledge (pp. 104–118). Routledge.
*Versteegh, P., Van Der Plas, T., & Nieuwstraten, H. (2013). The best of three worlds: More effective policing by a problem-oriented approach of hot crimes, hot spots, hot shots, and hot groups. Police Practice & Research, 14(1), 66–81. https://doi.org/10.1080/15614263.2012.665244
*Vogl, T. M., Seidelin, C., Ganesh, B., & Bright, J. (2020). Smart technology and the emergence of algorithmic bureaucracy: Artificial intelligence in UK local authorities. Public Administration Review, 80(6), 946–961. https://doi. org/10.1111/puar.13286
Walsh, J. P., & O’Connor, C. (2019). Social media and poli- cing: A review of recent research. Sociology Compass, 13(1). https://doi.org/10.1111/soc4.12648
*Walsh, W. F., & Vito, G. F. (2004). The meaning of Compstat: Analysis and response. Journal of Contemporary Criminal Justice, 20(1), 51–69. https://doi. org/10.1177/1043986203262310
Webster, J., & Watson, R. T. (2002). Analyzing the past to prepare for the future: Writing a literature review. MIS Quarterly, 26(2), xiii–xxiii.
*Weisburd, D., Davis, M., & Gill, C. (2015). Increasing collec- tive efficacy and social capital at crime hot spots: New crime control tools for police1. Policing: A Journal of
Policy and Practice, 9(3), 265–274. https://doi.org/10. 1093/police/pav019
*Weisburd, D., Mastrofski, S. D., McNally, A. M., Greenspan, R., & Willis, J. J. (2003). Reforming to preserve: CompStat and strategic problem solving in American policing. Criminology & Public Policy, 2(3), 421–456. https://doi.org/10.1111/j.1745-9133.2003.tb00006.x
*Wenzelburger, G., König, P. D., Felfeli, J., & Achtziger, A. (2022). Algorithms in the public sector. Why context matters. Public Administration
*Weston, C., Bennett-Moses, L., & Sanders, C. (2020). The changing role of the law enforcement analyst: Clarifying core competencies for analysts and supervisors through empirical research. Policing and Society, 30(5), 532–547. https://doi.org/10.1080/10439463.2018.1564751
*White, M. D., & Katz, C. M. (2013). Policing convenience store crime: Lessons from the Glendale, Arizona smart policing initiative. Police Quarterly, 16(3), 305–322. https://doi.org/10.1177/1098611113497045
*Wientjes, J., Delsing, M., Cillessen, A., Janssens, J., & Scholte, R. (2017). Identifying potential offenders on the basis of police records- development and validation of the ProKid risk assessment tool. Journal of Criminological Research, Policy and Practice, 3(4), 249–260. https://doi. org/10.1108/JCRPP-01-2017-0008
*Wiig, A. (2018). Secure the city, revitalize the zone: Smart urbanization in Camden, New Jersey. Environment & Planning C Politics & Space, 36(3), 403–422. https://doi. org/10.1177/2399654417743767
*Williams, M. L., Burnap, P., & Sloan, L. (2017). Crime sen- sing with big data: The affordances and limitations of using open source communications to estimate crime patterns. British Journal of Criminology, azw031. https://doi.org/10. 1093/bjc/azw031
*Williams, M. L., Edwards, A., Housley, W., Burnap, P., Rana, O., Avis, N. . . . Sloan, L. (2013). Policing cyber-neighbourhoods: Tension monitoring and social media networks. Policing and Society, 23(4), 461–481. https://doi.org/10.1080/10439463.2013.780225
*Willis, J. J., Mastrofski, S. D., & Weisburd, D. (2007). Making sense of COMPSTAT: A theory-based analysis of organiza- tional change in three police departments. Law & Society Review, 41(1), 147–188. https://doi.org/10.1111/j.1540- 5893.2007.00294.x
Wilson, J. Q. (1968). Varieties of police behaviour. Harvard University Press.
Wirtz, B. W., Weyerer, J. C., & Geyer, C. (2019). Artificial intelligence and the public sector—applications and chal- lenges. International Journal of Public Administration, 42 (7), 596–615. https://doi.org/10.1080/01900692.2018. 1498103
*Worrall, J. L. (2016). Smart policing in Frisco, Texas: Geographic and temporal displacement in a micro place. Policing: An International Journal of Police Strategies & Management, 39(1), 36–51. https://doi.org/10.1108/ PIJPSM-05-2015-0063
*Young, M. M., Bullock, J. B., & Lecy, J. D. (2019). Artificial discretion as a tool of governance: A framework for under- standing the impact of artificial intelligence on public administration. Perspectives on Public Management and Governance, 2(4), 301–313. https://doi.org/10.1093/ ppmgov/gvz014
430 M. AFZAL AND P. PANAGIOTOPOULOS
- Abstract
- Introduction
- Methodology
- Overview of the articles
- Evolution of data in policing
- A framework of data-driven policing
- Data types
- Data-driven processes
- Crime construction
- Crime sensing
- Surveillance automation
- Disorder automation
- Antecedents of data-driven policing
- Institutional
- Organizational
- Individual
- Ethical and privacy considerations of data-driven policing
- Implications and future research perspectives
- Conclusion
- Disclosure statement
- ORCID
- References
