Was the pope to blame? Statistical powerlessness and the predictive policing of micro‐scale randomized control trials

• DOI: http://doi.org/10.1111/1745-9133.12514
• Date: 01 August 2020
• Author: Jerry H. Ratcliffe,Ralph B. Taylor
• Published date: 01 August 2020

Received:  October  Revised:  June  Accepted:  June 

DOI: ./- .

SPECIAL ISSUE ARTICLE

CUTTING-EDGE RESEARCH IN POLICE POLICY AND PRACTICE

Was the pope to blame? Statistical

powerlessness and the predictive policing of

micro-scale randomized control trials

Ralph B. Taylor Jerry H. Ratcliffe

Temple University

Correspondence

RalphB. Taylor,Department of Criminal

Justice,Gladfelter Hall, Temple University,

Pollett Walk,Philadelphia, PA .

Email:ralph.taylor@temple.edu

Dataanalyzed here were collected under

Grant-R-CX- from the National

Instituteof Justice, Jerry H. Ratcliffe, Prin-

cipalInvestigator, and Ralph B. Taylor,

co-PrincipalInvestigator. Any views or

opinionsexpressed herein do not necessar-

ilyreflect the official policies of the Depart-

mentof Justice, the National Institute of

Justice,the Philadelphia Police Depart-

ment,the city of Philadelphia, or Temple

University.Authors appreciate helpful tips

onpc_simulate provided by Louis Freonas,

andinsightful comments provided by the

editors,reviewers, and Cory Haberman on

anearlier version of this article. Portions of

thisresearch were presented by the second

authorat the annual meetings of the Amer-

icanSociety of Criminology, November

,in San Francisco.

Fundinginformation

NationalInstitute of Justice, Grant/Award

Number:-R-CX-

Research Summary: Hinkle et al. () highlighted

a statistical powerlessness problem in hot-spots polic-

ing experiments in midsized cities with moderate prop-

erty crime rates. The current work demonstrates that

this problem is less readily resolved than previously sus-

pected. It reviews results from a predictive policing ran-

domized control trial in a large city with property crime

rates higher than Chicago or Los Angeles. It reports, for

the first time, a graphical analysis indicating the marked

car patrol intervention, practically effective at the ′

by ′ (mission) grid level, three grids per shift, likely

had a district-wide impact on reducing reported prop-

erty crime. In addition, it reviews results of a series of

thought experiments exploring statistical power impacts

of four modified experimental designs. Only one alterna-

tive design, with spatially up-scaled predictive policing

mission areas and concomitantly higher property crime

prevalence rates, produced acceptable statistical power

levels. Implications follow for current theoretical con-

fusion in community criminology about concentration

effects and units of analysis, and how models organize

impacts across those different units. Implications follow

for practice amid ongoing concerns about whether pre-

dictive policing works and, if it did, how to gauge its

impacts and social justice costs.

Policy Implications: The current work brings to

the fore important questions beyond “does predictive

Criminology & Public Policy. ;:–. ©  American Society of Criminology 965wileyonlinelibrary.com/journal/capp

966 TAYLOR RATCLIFFE

policing work?” Can we design predictive policing ran-

domized experiments capable of showing statistical

effectiveness? Furthermore, if we can, and if those stud-

ies include larger mission areas than the micro-scaled

geographic grids used so far, how do we integrate social

justice concerns into effectiveness metrics given the

broader segments of communities likely affected?

KEYWORDS

community criminology, effectiveness metrics and social justice

concerns, hot-spots policing, predictive policing, spatial and tem-

poral scaling

“There are at present insufficient rigorous empirical studies to draw any firm con-

clusions about either the efficacy of crime prediction software or the effectiveness

of associated police operational tactics. It also remains difficult to distinguish a pre-

dictive policing approach from hot spots policing” (National Academies of Sciences

Engineering and Medicine, ,p.S-)

This work builds on the results from a micro-scaled randomized control Predictive Policing

Experiment that worked practically, but not statistically, with graphically demonstrable district-

wide crime reduction impacts. Four thought experiments pose the following question: Can sta-

tistical powerlessness due to the rarity of Part I property crimes in micro-scaled locations dur-

ing specific time windows, in locations predicted to be the most property crime prone in their

respective districts, in a city with Part I property crime rates exceeding those of Chicago and Los

Angeles, when those rates are addressed with a demonstrably effective treatment, be surmounted

with alternative experimental research designs? Stated differently, what alternative hypothetical

experimental designs could have coped with the impaired statistical power associated with the

extremely low-property-crime prevalence rates at the micro-time-and-place-scale of a predictive

policing intervention?

Implications follow for theory, policy, and practice. In brief, they are as follows. For theory, the

current focus on extremely small crime intervention sites represents the culmination of half a

century of drilling down below the neighborhood level to examine, predict, and ultimately under-

stand variations in crime patterns and levels at micro-spatiotemporal scales. The current work on

near-repeat effects (Bernasco, ; Bowers & Johnson, ; Johnson & Bowers, ; Ratcliffe

& Rengert, ; Townsley, Homel, & Chaseling, ) and spatial concentration effects (Eck,

Lee, O, & Martinez, ; Lee, Eck, O, & Martinez, ; Weisburd, ) attests to some threads

of that drilling down. As yet, however, nobody has made a convincing case that, for a particular

crime type, a specific geo-scale or even a specific narrow range of geo-scales matches up to the

relevant micro-spatiotemporal dynamics driving crime occurrences. Therefore, from a theoreti-

cal perspective, predictive policing scholars, many of whom focus their research on small grids,

may wish to consider shifting to larger spatial frames. Results from specific hypotheticalscenarios

examined here support such a widening.