The Utility of an Internal Benchmarking Strategy in Racial Profiling Surveillance

• Author: Jeffery D. Dailey,Henry Jackson,Brian L. Withrow
• Date: 01 December 2008
• DOI: 10.3818/JRP.10.2.2008.19
• Published date: 01 December 2008
• Subject Matter: Article



*



 

 

Brian L. Withrow

Wichita State University

Jeffery D. Dailey

Texas A & M, Commerce

Henry Jackson

Kansas State University

* Abstract

Estimating the racial and ethnic proportions within the driving population at risk of being

stopped by the police is the most serious methodological threat to the viability of racial

prof‌iling research. Most strategies are externally based on the residential population,

f‌ield observations of drivers, and, more recently, accident records. Using data collected

in 2001 by the Wichita (Kansas) Police Department, the authors construct an internal

benchmark and are able to identify the off‌icers (at the individual level) that are actually

stopping substantially higher proportions of racial minorities when compared to their

similarly situated peers. Recommendations for developing an internal benchmarking

strategy are included.

JUSTICE RESEARCH AND POLICY, Vol. 10, No. 2, 2008

© 2008 Justice Research and Statistics Association



During the past 15 years the racial prof‌iling controversy has matured into one

of the most divisive issues facing American policing. In the early 1990s, the con-

troversy began primarily within the context of routine traff‌ic stops and focused

primarily on the overrepresentation of African-American and Hispanic drivers

in routine street-level enforcement activities. After September 11, 2001, the con-

troversy continued within the context of airport security checks focusing on the

overrepresentation of Middle Eastern and Muslim “looking” travelers in security

checks at the nation’s airports. Today, the controversy has emerged within the con-

text of local and state efforts to enforce federal immigration laws and the potential

discriminatory effect these efforts might have on the Hispanic community.

The most common response to the racial prof‌iling controversy is to collect data

on who (by race, ethnicity, age, and gender) gets stopped by the police and what

happens (e.g., searches, arrests, physical confrontation, etc.) to the people after they

are stopped. In most cases these data collection exercises are discrete events that

begin with some public controversy and end with formal reports, often containing

rather inconclusive f‌indings. A few jurisdictions collect stop data routinely. Some do

so voluntarily and others in response to state statutes requiring stop data collection.

These jurisdictions report their f‌indings at the end of each year, often focusing on

changes in the overall proportions (by race or ethnicity) of individuals stopped from

one year to the next. Some agencies respond to these reports by developing cultural

sensitivity training programs, promulgating policies that prohibit racial prof‌iling,

and initiating proactive citizen complaint procedures. For the most part, however,

police agencies remain unaffected and racial prof‌iling studies are seldom used as a

means to support signif‌icant administrative change (Schultz & Withrow, 2004).

In nearly all cases the stop data reporting the proportions of individuals

stopped by race or ethnicity are compared to some benchmark that ostensibly es-

timates the proportions of individuals available to be stopped by race or ethnicity.

During the past decade, the variety of benchmarking strategies has increased rather

dramatically, largely due to the immense creativity among the researchers active in

this agenda (Engel & Calnon, 2004; Withrow, 2004a). Unfortunately, no bench-

marking strategy has proven to be universally acceptable and there is considerable

disagreement amongst practitioners, scholars, and researchers on the reliability

and validity of the available benchmarking strategies (Withrow, 2007). In general,

benchmarking strategies fail to completely measure the population of individuals

that are actually observed by the police but not stopped. At best benchmarks esti-

mate the population of individuals that might or should be observed by the police,

and thus available to be stopped.

With some exception, the benchmarking strategies used by racial prof‌iling re-

searchers are based on calculations that are external to their routine enforcement

contexts. For example, most researchers use the residential population (by race

or ethnicity), or some derivative of it, as an estimate of the driving population.

Other researchers use information about the race or ethnicity of drivers involved

in two-

vehicle motor vehicle crashes, assuming these events are suff‌iciently random