an analysis of police traffic stops and searches in kentucky: a mixed methods approach offering...

Post on 12-Mar-2023

0 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Springer is collaborating with JSTOR to digitize, preserve and extend access to Policy Sciences.

http://www.jstor.org

An Analysis of Police Traffic Stops and Searches in Kentucky: A Mixed Methods Approach Offering Heuristic and Practical Implications Author(s): Brian N. Williams and Michael Stahl Source: Policy Sciences, Vol. 41, No. 3 (Sep., 2008), pp. 221-243Published by: SpringerStable URL: http://www.jstor.org/stable/40270967Accessed: 30-04-2015 17:41 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/ info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 41:221-243 DOI 10.1007/sll077-008-9065-z

An analysis of police traffic stops and searches in Kentucky: a mixed methods approach offering heuristic and practical implications

Brian N. Williams • Michael Stahl

Published online: 23 July 2008 © Springer Science+Business Media, LLC. 2008

Abstract Is race a significant predictor of being searched by the police? Is race a sig- nificant predictor of having positive search results during traffic stop? We address these two questions by analyzing traffic stop data (n > 93,000) collected by two state and 24 local police agencies in a single state during the 2001 calendar year. Our findings show that race does correlate with a fruitful traffic stop but not in the manner that may be commonly thought. To supplement and better contextualize our quantitative findings, an exploratory study was then designed that used the focus group interviewing technique with groups of officers from five of the participating agencies to explore their perceptions of (1) traffic

stops and searches and (2) public allegations of racial profiling and bias-based policing. The findings from this qualitative phase of the study highlighted the officers' perceived role as community problem solvers "who profile problems and not people." In tandem, this mixed-method approach was instrumental in advancing our knowledge of both the patterns and results from related searches, in addition to better contextualizing the underlying perceptions of officers regarding the use of race in "solving" community problems. The results from this combination of methodological approaches offer important heuristic and

practical implications.

Keywords Racial profiling • Traffic stops • Bias-based policing •

Police-community relations

Negative criminal stereotyping of minorities by the majority population is a reality that has been shared by many countries and cultures (Goodey 2006). Scattered throughout the

pages of history are numerous examples of the global phenomenon of defining different or

B. N. Williams (El) Department of Public Administration & Policy, The University of Georgia, 204 Baldwin Hall, Athens, GA 30602, USA e-mail: bnwillia@uga.edu

M. Stahl Stanford Law School, Stanford University, Crown Quadrangle, 559 Nathan Abbott Way, Stanford, CA 94305-8610, USA

Ö Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

222 Policy Sci (2008) 41:221-243

difference as deviant (Lombroso 1911). Consequently, this view has been embedded within the various laws, structures, and institutions of government. The American experience of

criminalizing the "other" is not unlike other countries (Amnesty International USA 2004). From its more distant history of chattel slavery to the more recent, legally sanctioned, discrimination based on race during the Japanese Internment and the era of Jim Crow and de jure segregation, racial, ethnic, religious, or other types of threat have been the object of social control via one primary institution of government - law enforcement agencies (Harris 2006; Barnes 2005). The historical realities of American society, in particular, disparate service delivery to, and focused police attention on, certain racial, ethnic, and other marginalized populations, have gone against the philosophical values, principles and foundation of our democracy and continues to negatively effect public trust and confidence in government, in general, and in policing, in particular (Weitzer and Tuch 2002, 1999). Moreover, these historical realities set the backdrop for the more contemporary discussion and interpretation of bias-based policing.

Bias-based policing, a phrase that refers to any form of prejudice, whether religious, racial, ethnic, age, gender, or sexual preference, that one might encounter from law enforcement officers during the performance of their official duties, continues to be a hot

topic for scholarly debate and investigation. Of particular importance is the concept of racial profiling, especially in the context of traffic stops and searches.

Racial profiling was defined by the U.S. General Accounting Office (now the U.S. Government Accountability Office) as the use of "race as a key factor in deciding whether to make a traffic stop" (2000, pp. 1). Similarly, Ramirez et al. (2000) define racial profiling as

...any police-initiated action that relies on the race, ethnicity, or national origin rather than the behavior of an individual or information that leads the police to a

particular individual who has been identified as being, or having been, engaged in criminal activity (pp. 3).

This is a commonly perceived police practice that centers on police interactions with citizens principally during traffic stops and searches.

During the last decade, racial profiling has emerged as a pressing national issue. In a

span of 2 years, the number of media, research, legal and other related articles written on the subject grew in an exponential fashion - from a mere 290 in 1998 to an astonishing 19,145 articles in 2000 (Meehan and Ponder 2002). This increase is even more significant when realizing the exclusion of the impact of the terrorist attacks of September 11, 2001 and its implications for racial profiling related articles specific to Arabs and Arab- Americans.

As a response to consent decrees, mounting negative public perceptions of law enforcement, and the erosion of public trust and confidence in government and its policing agents, law enforcement agencies at the state and local levels have been mandated and/or volunteered to participate in the collection of traffic stop data to monitor the behavior of officers and to ascertain whether or not race was the determining factor in the decision to stop and search a vehicle. However, previous research and its accompanying statistical analyses have illustrated the difficulty in clarifying the muddy waters associated with traffic stops. Of particular note are the problems that emerge from confounding environ- mental variables, like a high neighborhood crime rate, and the absence of baseline data on who exactly is the driving public for a particular state or jurisdiction, that make ascer- taining whether or not racial profiling exists very complicated, and often unreliable (Ramirez et al. 2000; Walker 2001 ; Engel et al. 2002; Smith and Alpert 2002; Rojek et al. 2004; Ridgeway 2006). Consequently, verifying if race is the determinant reason for the

£} Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 41:221 -243 223

stop is increasingly problematic. Furthermore, it is our perspective that researchers have overlooked one of the major reasons why racial profiling on our roadways began and one of the purposes that continues to underlie the stop and search practices of today: drug interdiction through searching vehicles.

The theory and practice of traffic stops and searches: a review of selected literature

It has been argued by some that the common conception of racial profiling, as we know it

today in the context of traffic stops, began during the Reagan Administration's "War on

Drugs."1 A "drug courier" profile based on common characteristics thought to be shared

by drug couriers or "mules" was created by the Drug Enforcement Agency (DEA), cus- toms officials, and local and state law enforcement agencies to decrease the flow of drugs up and down the "pipelines" of the 1-95 corridor. This profile was the byproduct of what can be considered a technical-rational (Miller and O'Leary 1989; Adams and Balfour 1998) or scientific-management approach to public safety and public order - an attempt to address the problem of illegal drug trafficking via scientific methods and procedures. In

theory, the development and implementation of this profile would generate more stops, searches and arrests, ultimately serving as an efficient and effective means to stem the flow and resulting effects of illegal drugs.

A review of related literature, however, paints a picture that often reflects the not-so-

rosy results of when theory meets practice. At the local and statewide levels, recent studies and government surveys overwhelmingly reveal that police have disproportionately caught minority motorists in the traffic stop dragnet (Lamberth 1996; Cordner et al. 2000; Gross and Barnes 2002; Close and Mason 2003; Alpert et al. 2007). However, these stops and

subsequent searches have not been an effective tool in finding contraband and stopping the flow of drugs (Meehan and Ponder 2002; Petrocelli et al. 2003). Minorities (especially blacks and Hispanics) who would fit the profile of the drug courier are more likely to be

stopped and searched than white motorists, especially as they travel farther from "black communities" and into "whiter" areas (Meehan and Ponder 2002; Alpert et al. 2007), yet race has not been a significant predictor of the police having a positive search of the vehicle. This finding has been supported by the research efforts of Engel and Calnon

(2004) and Hernandez-Murillo and Knowles (2004). Despite the formulation and imple- mentation of the drug courier profile, the analyses of local and statewide data highlight the

fragile, faulty theoretical logic that supports racial profiling as an efficient and effective

attempt at drug interdiction. These empirical findings highlight the apparent ineffectiveness of stops and searches and converge with the anecdotal reflections of an Operation Pipeline instructor.

...the operating principle behind [Operation Pipeline is] volume, volume, volume. Its sheer numbers... Our guys make a lot of stops. You've got to kiss a lot of frogs before you find a prince." California Highway Patrol Units kissed nearly thirty-four thousand frogs in 1997. Only 2 percent of them were carrying drugs. In other states,

1 This perspective differs from the perspectives of others. In particular, Williams (2000), drawing from the perspectives of Williams and Murphy (1990), views contemporary racial profiling as a by-product and a lasting legacy of American chattel slavery. Others have noted that the tentacles of racial profiling in drug interdiction extend back to the 1950s for blacks, to the 1930s for Mexican Americans, and the turn of the 20th century for Chinese immigrants (Gray 1998). All arguments coalesce around the linkage of racial profiling being a social control tool used by law enforcement.

£} Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

224 Policy Sci (2008) 41 :221-243

up to 95 percent of all Pipeline searches have been found to be dry holes (Webb 1999, pp. 122).

The findings from national surveys mirror the findings that show racial profiling as an ineffective means of criminal investigation and drug interdiction at the local and state levels. An analysis of data generated by the 2002 Police-Public Contact Survey by Durose et al. (2005) revealed that approximately 5% of traffic stops resulted in searches of the driver or vehicle.2 Overall, blacks (10.2%) and Hispanics (11.4%) were more likely to be searched by police than whites (3.5%). Similarly, an analysis of data generated by the 1999 Police-Public Contact Survey by Schmitt et al. (2002) noted that blacks (8.5%) and

Hispanics (9.7%) were more likely to have their vehicle searched than whites (4.3%). Even

though black and Hispanic drivers represent 5.3% and 3.7%, respectively, of male drivers

age 25 and older stopped by police, they accounted for 12.5% and 6.1%, respectively of male drivers age 25 and older who had their vehicle and/or person searched. This compares to whites, in this same time period, accounting for 34% of male drivers age 25 and older who were stopped and 29.9% of those male drives age 25 and older that were searched. In the analyses of data generated from both the 1999 and 2002 Police-Public Contact Sur-

veys, younger and older male black and Hispanic drivers were more like to be searched than younger and older white male drivers. The analysis of Langan et al. (2001) revealed that Black (11.0%) and Hispanic (11.3%) motorists stopped by police were more likely than whites (5.4%) to be searched. Yet, searches of white drivers or their vehicles were more likely to yield criminal evidence (17%) than searches of blacks (8%), but not sig- nificantly more likely than Hispanics (10%) (Langan et al. 2001). The report analyzing search behavior on the New Jersey State Turnpike also revealed that minorities were being searched more than whites, yet the police searches were not yielding significantly more

positive results. Of all the searches conducted during the evaluation period on the New

Jersey State Turnpike, 86% were on non-white motorists. Despite the disproportionate search rate of non-white motorists, there was no significant difference between races or correlation with race and "hit rates" (searches in which police found contraband or reason to arrest the citizen). These findings reveal, at the local, statewide/regional, and national levels that race is a determining factor in who is being searched more frequently (minorities), but race has no significant correlation with the "hit rate" or who is actually found in possession of illegal substances. As noted by Dunham et al. (2005), these efforts seem to be effective only in transforming African-American and Hispanic citizens into suspects.

Previous studies reveal that researchers can still examine the equity and efficacy of police behavior by focusing on search patterns and search results. Focusing on search patterns addresses several of the major concerns in the racial profiling debate. First, it confronts the methodological problems of examining traffic stop data without a baseline population. Using search data allows the researcher to examine the data without a roadway population survey because the baseline population is the pool of motorists that have been pulled over. Analysts can control for the location and reason for the stop in order to examine if race is a predictive variable in whether or not police search a vehicle. In addition to assuaging methodological issues, focusing on the search allows researchers to return to and examine the original purpose of the profile: to stop drugs.

2 The 2002 Police-Public Contact Survey was administered to 76,910 individuals throughout the United States. This survey serves as a supplement to the National Crime Victimization Survey.

£) Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 41:221-243 225

However, there remains another problem in the racial profiling debate. The results of the cited studies are somewhat location specific. Aside from the national data collected by Langan et al. (2001) and Durose et al. (2005), researchers and states are interested in

verifying if racial profiling is indeed a problem within their jurisdictions. As a result, analysts and lawmakers must examine data collected locally. In response to such a need, we have conducted the following analysis to build on the current evidence that has examined search patterns and the results of these searches in traffic stops. Specifically, we examine whether race is a determining factor in search patterns on the interstates in the state of Kentucky, and, if so, whether or not the significant difference in ratio of searches coincides with significant differences between racial and ethnic groups having positive search results.

Methods

Hypotheses

The statistical analysis in this study will explore the search patterns and results of traffic

stops for the law enforcement agencies in Kentucky during the calendar year 2001. The data collection procedures and approach to racial profiling analysis in Kentucky mirrored the process conducted in other states such as Missouri and New Jersey. The state gov- ernment passed laws prohibiting racial profiling in police departments and issued orders to collect traffic stop and arrest data. The state of Kentucky began to collect traffic stop data in order to monitor police practices and to determine whether or not racial profiling was a

problem. In this analysis, we want to know: Who is being searched? Are all racial groups being searched at an equal rate, controlling for other variables? What are the results of these searches? Given these questions and our extensive data set, two clear research

questions emerge from this line of inquiry: (1) controlling for other variables, is race a

significant predictor of being searched by the police, and (2) controlling for other variables, is race a significant predictor of having positive search results?

We hypothesize that, despite state laws prohibiting racial profiling, traffic stops in

Kentucky will follow the same patterns of police in other states. That is, police will search

minority motorists more often than whites (in accordance with the drug courier profile). Also consistent with findings in other cases, we hypothesize that, despite their focus on

minorities, police will not have significantly more positive search results in the vehicles of minorities than in white vehicles.

Data collection

Data for this study come from traffic stops collected by participating law enforcement

departments and agencies within Kentucky during the 2001 calendar year. Participation in this data collection process was voluntary for local departments and mandated for state

police departments. The data collected came from the 26 law enforcement agencies that

participated in this study. These agencies represented 24 local law enforcement agencies, including local police departments and sheriff departments, as well as two state law enforcement agencies: the Kentucky State Police and Kentucky Motor Vehicle Enforce- ment. Police officers were required to complete a stop form every time they pulled a motorist over to the side of the road. Officers were then required to submit their forms to their respective departments. Information collected during the traffic stop and entered into

£) Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

226 Policy Sci (2008) 41:221-243

the data set was as follows: (1) race and gender of driver; (2) location of the stop; (3) reason for the stop; (4) number of passengers in the vehicle; (5) search conducted; (6) search results; (7) disposition; and (8) duration of stop.

Approach to data analysis

Because there was no roadway survey population conducted for these stops, we cannot establish a baseline population for the local roadways. Consequently, we will not try to establish whether or not race was the determining factor for the reason for the stop. Rather, as we have argued above, it serves the analysts to examine the search patterns and search results of the police. And, since racial profiling has been used in the effort to stop drug traffickers, we focus our analysis on search behavior of the police officers and the results of these searches. Because the baseline population for searches is the pool of motorists

stopped by the police, not the total number of drivers on the road, examining the search

patterns of the police allows researchers to effectively establish whether or not race is

being used as a determining factor to search a vehicle. In further efforts to conduct sound analyses and control for confounding factors, we

restricted the analysis to stops on the interstate. Only examining interstate stops provides a context for analyzing police-citizen interactions that removes many of the contextual variables that can complicate the reasons and results of police-citizen interactions in

neighborhoods. As Buerger and Farrell (2002) point out, the highway (or interstate):

...is relatively pure in terms of profile targeting: the interstate has no resident pop- ulation (although it does have regular users - commuters - along popular stretches); all citizens are engaged in a common activity, and all are relatively anonymous. None of the 'noise' that attends municipal policing is present, and law enforcement is entirely officer-initiated, (p. 302)

In addition to controlling for the location of stop, we have also controlled for reason for the traffic stop to remove the legal justification for police-initiated contact from the argument. All of the analyses include those motorists pulled over for compliance and courtesy stops3 since those motorists pulled over for criminal complaints would skew the results for search and search results.4

Data was only coded as missing if the category was not completed or an error was made

by the officer during the completion of the form so that the information was inaccurate.

Missing data could pose a severe problem in studies such as this one because, if data are missing in a non-random fashion, especially in demographic variables, it could skew the results. Missing data in this study, however, did not present a major hurdle. Because the

missing data composed less than one percent of any variable, missing cases were deleted listwise and did not have any substantive effect on the results.

3 A compliance stop is defined as any routine traffic violation where the officer believes the motorist has not complied with traffic laws. A courtesy stop consists of any stop that does not involve a traffic violation or any other suspected offense. For example, an officer might make a courtesy stop to advise the motorist that his or her gas cap was not properly fastened. For criminal complaints, an officer stops a motorist because the individual is suspected of committing a crime. 4 Because the search variable was bivariate, we could not distinguish between probable cause and consent searches. We did, however, focus on stops where officers would be the least likely to have probable cause for a search (compliance and courtesy stops) and eliminated stops where police officers would most likely have probable cause (stops involving criminal complaints).

£ì Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 41:221-243 227

Comparing results within groups

In addition to basic frequencies performed to examine the sample population, the analyses focused on comparing how different groups were treated on the roadways. While deter-

mining whether or not findings are statistically significant, verifying the substantive importance is crucial since large sample sizes tend to increase the likelihood of significant results (Agresti and Finlay 1997). In the original analysis, significant differences existed between groups (Stahl 2003).5 However, these are only effective to verify if global dif- ferences exist. Given our sample size, significant results do not necessarily mean that substantive differences exist between variables. To analyze these differences within groups, we chose to use logistic regression. Logistic regression analysis provides an

appropriate and sound statistical technique to analyze the substantive differences between

groups while employing regression techniques that offer predictive power (Pampel 2000). Utilizing logistic regression with this study allows the authors to examine how the different racial groups are searched and what the results are for these searches. In these regression models, a nested model was used to verify if gender also had an effect on search behavior since males are also thought to be targeted more than female motorists (as would be consistent with the drug courier profile) (Weitzer and Tuch 2002). Logistic regression allows the analyst to determine whether or not race, ethnicity or gender is a significant predictor of being searched and whether or not race, ethnicity or gender is a significant predictor of search results.

The analysis incorporated logistic regression and built models with the available vari- ables that go along with the "drug courier profile" described above. The included

independent variables were race, gender, and the reason for the stop and the dependent variables were search and search results coded into dichotomous variables for (1 = yes, 0 = no) and (1 = positive, 0 = negative) respectively. Logistic regression allowed the authors to examine how race and ethnicity would influence search behavior and search results while controlling for other variables such as reason for the stop and gender. Using logistic regression instead of regular OLS regression also enables researchers to use dichotomous dependent variables that would violate the assumptions of the more common OLS regression technique (Pampel 2000).

Because of the incredibly large sample size in the data (N > 93,000), the authors chose to set the significance level at 0.01 and to verify if the statistical differences were also

important substantively. We set the p-value at 0.01 in order to reduce the chance for a Type 1 error. In other words, by setting the p- value at 0.01, we have chosen to reduce the chance of rejecting the null hypothesis and accepting the research hypothesis from 0.05 to 0.01.

Agresti and Finlay (1997) support tougher significance levels in research with serious

implications. In the case of this research, we determined that the consequences of these

findings were serious enough to warrant tougher significance levels. Therefore, lowering the p- value to 0.01 reduces the chance of erroneously claiming that police use race as the

determining factor for searching the vehicle. The authors built two logistic regression models for each dependent variable (search and

search results) in order to verify the nested effect that gender had on the likelihood that an

5 All of the omnibus tests (F Tests) were significant at a p- value < 0.001 , except for the difference in means for search results (race was coded as a categorical variable using White, African -American, Hispanic and Other as categories). Thus, the initial tests indicate that police are searching groups at different rates but the results of searches are not significantly different between racial groups. For more information, please see Stahl 2003, pp. 21-29.

£} Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

228 Policy Sci (2008) 41 :221-243

individual was searched or that the search results were positive. The addition of gender had no significant effect on the other independent variables, thereby indicating that there was no serious interaction between gender and race. The gender variable lowered the beta coefficients somewhat; however, the change was not significant and did not substantively alter the nature of the results. In the following interpretations of the results, we will examine the full model that contains gender along with all of the other variables.

Interpretation of results

Search behavior

Table 1 lists the results for the logistic regression model predicting whether or not a search was conducted.

The regression model that tested police search patterns provided results that confirmed our hypotheses. Using the p~ value of 0.01 as the cut-off point, male, African- American, and Hispanic motorists had greater odds of being searched. These three variables were all

significant (each having a p- value < 0.001). The variable for white motorists was not

significant, indicating that being white did not increase the odds of being searched, con- sistent with the findings for African-Americans and Hispanics. What does this mean? As mentioned previously, especially with a big sample size, sometimes significant p- values do not mean that there is truly a substantive difference between the variables. In order to

verify the substantive value of these findings, we compared odds ratios and probabilities with these variables. Using the odds ratios based on techniques described in (DeMaris 1995), the results show that the significant findings for male motorists indicate that the odds of male motorists being searched are 1.8 times greater than the odds of police searching female motorists. This difference translates into a figure that reveals that the likelihood of police searching the vehicle of a male motorist is 1 1% greater than for female motorists. These figures reveal statistical and substantive differences between the search patterns of police for male and female motorists.

Table 1 Logistic regression predicting search9

B Exp(B)b

Independent variables0

White 0.698 2.010 African-American 1.05*** 2.859

Hispanic 2.40*** 11.067 Traffic stopc -1.03*** 0.358

Compliance stop - 1 .55*** 0.2 1 3

Courtesy stop -0.691*** 0.501 Male 0.623*** 1.864 Constant -3.977 0.019

* p < 0.005, ** p < 0.005, *** p < 0.001 a

x2 Goodness of fit (Hosmer and Lemeshow) = 2.67 (p = 0.45) b

Exp(B) represents the 'odds' in a logistic regression equation c

Stop variables are the reasons the officer pulled over the motorist d All variables coded as dummy variables (with "other" as reference category coded as 0)

£) Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 4 1 :22 1-243 229

The odds of black motorists being searched are 1 .44 times greater than the odds of white motorists being searched.6 This difference in odds is also statistically significant. The

probability of police searching an African- American 's vehicle is 1.5% greater than the

probability of police searching a white motorists' vehicle.7 Although this difference remains statistically significant, the difference does not translate into a substantive dif- ference between how police search white and African-American motorists.

For Hispanics, however, the findings are substantively meaningful as well as statisti-

cally significant. The odds of Hispanic motorists being searched are 5.51 times greater than the odds of white motorists being searched. The figures reveal that the probability of police searching the vehicle of a Hispanic motorist is 16% greater than the probability of

searching the vehicle of a white motorist. This first logit model testing search behavior shows that there are significant statistical

and substantive differences between how police search vehicles. Male motorists, African- American motorists, and especially Hispanic motorists are significantly more likely to have their vehicle searched than female motorist and white motorists. For males and Hispanics, these significant differences are also substantive. These outcomes align with the practices police follow using the "drug courier profile" and with the findings of other previously mentioned studies such as Petrocelli et al. (2003) and Smith and Petrocelli (2001). Some authors claim that the police have a good reason for this search behavior because they contend that minorities and males have higher rates of speeding and criminality than whites

(MacDonald 2002; Taylor and Whitney 2002). If their claim is justified, not legally but

empirically, then the disparate police searches of minorities should lead to more positive search results with minority motorists. We tested this claim to verify whether or not police are much more likely to find illegal substances or contraband in the vehicles driven by males, Hispanics and African-Americans. We examined the search results of the Kentucky law enforcement agencies in order to answer two questions. First, were police effective in their searches? Second, are the assumptions of the drug courier profile and the proponents of racial profiling correct? That is, are minority motorists much more likely to have

positive search results?

Search results

The results for the logistic regression model predicting search results are listed below in Table 2.

Once again, the results of the logistic regression models confirmed our hypothesis. Due to lack of statistical significance, the tests reveal that neither the race nor the ethnicity of the driver was a significant predictor of positive results of the vehicle search. The odds ratios reveal that odds of a positive search for white motorists were 1.04 times greater than for Hispanic motorists and 1.2 times greater than for African- American motorists8;

6 These numbers were reached by comparing the odds ratios between white and black motorists. Here 2.898/2.010 = 1.44. The same method was used to compare Hispanics with whites. 7 The probabilities calculated here reflect the probability of a male belonging to the specific racial category being searched during a courtesy stop, when an officer would be least likely to search a vehicle (e.g., the

probability of a White male versus African-American male versus Hispanic male during a courtesy stop). Although each stop category varies somewhat, the results remain consistent across all categories. Both

Hispanics and African- American motorists have statistically significant differences as compared to whites, but only Hispanics have substantive differences in search patterns as compared with whites. 8 The same process described in footnote 4 was also applied to calculations comparing search results for

whites, blacks and Hispanics.

£) Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

230 Policy Sci (2008) 41:221 -243

Table 2 Logistic regression predicting search results3

B Exp(B)b

Independent variables0

White 0.706 2.026

African-American 0.527 1 .693

Hispanic 0.669 1.952

Traffic stopc 1.07*** 2.908

Compliance stop - 1 .97*** 0. 1 40

Courtesy stop -0.561 0.571

Citation -4.26*** 0.014

Warning -2.63*** 0.072

No action _2.49*** 0.083

Male 0.63*** 1.878

Constant -4.25*** 0.014

* p < 0.001, ** p < 0.005, *** p < 0.001 a

x2 Goodness of fit (Hosmer and Lemeshow) = 890.9 (p < 0.001) b

Exp(B) represents the 'odds' in a logistic regression equation c

Stop variables are the reasons the officer pulled over the motorist d All variables coded as dummy variables (with "other" as reference category coded as 0)

however, none of these differences were statistically significant. The conversion of odds to

probabilities also confirms that there are no substantive differences between racial groups and positive search results. The probability of police having a positive search for Whites is 4% greater than African- Americans and 1% greater than Hispanics.9 The odds of police having positive search results are no different between racial or ethnic groups, statistically or substantively.

Male motorists, however, did have a statistically significant coefficient. This finding indicates that the odds of a positive search of the vehicles of male motorists are 1.88 times

greater than a positive search of a female motorist. The probability of having a positive search of a male motorist is 8% greater than for a female motorist.

Limitations

There are several limitations to this study. First, due to the lack of baseline roadway and neighborhood population, the analysis does not include the municipal stops and searches conducted in locations other than the interstate.10 We are still able to conduct a statistically sound analysis of what happened on the interstate but cannot include other locations in this

9 The probabilities calculated for search results follow the same method used in footnote 5. 10 This study also does not address how Kentucky might differ from other states in two areas: demographic data and legal differences that could affect traffic stops. While comparing Kentucky's demographics and legal nuances to other states might be informative, it is not crucial to the study for two key reasons. First, our study focuses on a methodological argument by asserting that searches can be used to investigate racial profiling practices. It does not argue that the findings in Kentucky can be generalized to all states. It merely asserts that the methodology (i.e. analyzing search patterns) can be used in any state. Second, this meth- odology should apply to all states despite any variance in state law, because the Fourth Amendment of the United States Constitution requires probable cause before conducting any search or seizure. Because all states must abide by this Amendment, analyzing search behavior could be used in any state.

4ü Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 4 1 :22 1-243 23 1

analysis. Another possible limitation is that researchers can never be 100% certain of the

validity of the collected data. Although departments in the state of Kentucky required all officers to record and submit the information in an accurate manner, analysts can never be

completely certain that the data was recorded and entered correctly. Also, the data set does not provide the option of separating state and local agencies.11 For state agencies that were

required to participate in this study, the possibility of intentional misrepresentation of traffic stop data could occur to create the impression that racial profiling is not an issue. Such a conscious misrepresentation of the truth actually has occurred when officers in New

Jersey were found to be "ghosting" their data, or recording race and ethnicity incorrectly to create the illusion of equitable stop and search procedures (State of New Jersey v Pedro Soto et al. 1996). While such a possibility should not be ignored, misrepresenting the data does not seem to be an issue in this study. Even if state agencies modified demographic variables to create the impression of equitable traffic stops, the results of our analysis clearly reveal that race is still a significant and substantive predictor of police searches.

As for local agencies, the voluntary nature of local police department participation could create the possibility of a self- selection bias. Such a bias would mean that the only departments willing to participate in this study may be those without a perceived racial

profiling problem. Once again, this limitation does not seem to be an issue since results from the analysis indicate that indeed race was a significant predictor of being searched by the police even if the participating departments felt like race was a non-factor. Despite these limitations, with a sample of over 93,000 motorists with less than 1% of the data

missing, the sample provides an excellent representation of who the police were stopping and searching on the interstates and what were the results of those searches. Moreover, when considering the real likelihood of a "Hawthorne Effect" ensuing from the fact that the actions of officers were being captured by their requirement to file a report that included the race of those drivers they stopped and searched, may actually lead to an underestimation of the extent of profiling in this study.12

Contextualization of quantitative findings

To supplement and better contextualize our quantitative findings, an exploratory study was then designed.13 A qualitative, non-experimental design using focus group interviewing was used to collect information about the perceptions of officers from five of the

1 ' Because the dataset does not allow for the separation of state and local agencies, we cannot perfectly match our quantitative and qualitative analyses. Despite this limitation, the focus groups still provide important contextual information concerning how police officers approach traffic stops within their juris- dictions. The focus groups identify drugs and drinking and driving as the major community issues. Because these concerns are fairly common across jurisdictions, state and local police officers will not likely vary in how they attempt to curb these problems during traffic stops. 12 The Hawthorne effect was first noted in the Hawthorne Studies which were performed on workers

employed by the Western Electric Plant in Cicero, Illinois in 1927. The research team, consisting of Elton

Mayo, Fritz Roethlisberger, and Martin Dickson, noticed an increase in worker productivity and deduced that this increase in productivity was a result of the psychological stimulus of being singled out and their actions being investigated and researched. 13 The primary objective of exploratory research is to gain a deeper understanding of a relatively unknown social phenomenon related to traffic stops and searches. Exploratory research efforts are often associated with phenomenological pursuits. Our particular exploratory study sought to describe or document those

exceptional characteristics or qualities that did not surface in our quantitative analysis to better explain what facilitated traffic stops and searches by officers.

£} Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

232 Policy Sci (2008) 41:221-243

participating 24 local agencies14 on (1) traffic stops and searches and (2) public allegations of racial profiling and bias-based policing. In particular, during the exploratory study a

sequenced set of open-ended questions was used to capture and examine officer responses to the following questions:

1 . What are the major problems in your community? 2. What types of activities are being done to address these problems? 3. Are traffic stops and searches effective in preventing or addressing some of these

pressing community problems?

Appendix A elucidates the methodology used, inclusive of the fundamental assumptions associated with focus group interviewing and the strengths and weaknesses of this

approach. Participating officers: A total of 24 officers participated in this phase of study. In terms

of demographic characteristics, all of these officers were male with only 3 of the 24 being racial minorities (African Americans). It is important to note that these officers were not a

representative sample of their respective agencies or of the aggregate number of officers who completed the traffic stop cards, however, they did serve as a purposeful population -

all were patrol level officers who routinely were involved in interstate traffic stops, and in some instances, searches.

Data breakdown and analysis

All focus group discussions took place in meeting or "training" rooms within each of the five participating departments and were audio-taped and transcribed. Considerable

emphasis was placed on getting in tune with the reality of the officers in the groups and

seeing the traffic stops and searches world through their eyes. This was reflected in the

approach to data analysis. In order to get a more intimate feel for the data collected, a manual approach to capturing

the content and context of group discussions was used instead of a computer assisted

approach. This approach followed the protocol for data breakdown advocated by Knodel (1993). As a result, data breakdown was facilitated by employing a two-step process where each group discussion was divided into two categories (see Fig. 1). These categories con- sisted of major incidents described by officers and their prevailing perceptions. From these experiences, incidents, and perceptions, certain themes emerged and were identified.

A combined ethnographic-content approach guided the analysis of emerging themes. This approach drew upon the strength of assertion analysis which couples designation analysis, or describing focus-group discussions by directly quoting participants with attribution analysis, or the counting and coding of the frequency of words, phrases, and statements in the identification of group perceptions or themes (Krippendorf 1980). Finally, a descriptive-interpretive approach was used to compare and contrast perceptions, atti- tudes, experiences, and opinions across the different officer/agency groups.

14 The focus group technique utilizes participants who are selected because they constitute a purposeful, but not necessarily representative, sample of a specific population (Stewart and Shamdasani 1990). All focus group participants in our exploratory study were affiliated with local police agencies. Even though there are unique differences between state and local police officers (i.e., more calls for service by local officers when compared to state officers and statewide jurisdiction for state officers when compared to local jurisdiction for municipal officers), these differences are subtle and not substantive in relation to how state or local police officers would perceive problems or concerns impacting the communities in which they live and jurisdictions in which they serve.

£} Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 41:221-243 233

Fig. 1 Visual depiction of focus group methodology

Results: emergent themes

Focus group interviewing is not designed to reach a consensus, yet little variation in

perceptions surfaced in the five focus group discussions across departmental/agency lines.

Consequently, four themes related to racial profiling and/or biased policing emerged. The

emergent themes were:

• Drugs and drinking and driving as major community problems in regards to traffic

stops. • Crack cocaine as a local problem associated with the African-American community. • Driving under the influence as a local problem associated with "Hispanic" immigrants

or migrant workers. • The intuitive nature of traffic stops and searches as an effective strategy of addressing

community problems.

The following paragraphs present, describe, and illustrate, via direct quotes from focus

group transcripts, these emergent themes.

Theme 1 : Drugs and drinking and driving as major community problems

In focus group discussions, officers were asked to identify major problems (i.e., those

pressing issues that they face on a daily basis) confronting their respective communities. Officers from the participating departments identified drugs and alcohol as major problems.

Drugs, theft. Yeah, drugs and alcohol related crime. Crack cocaine, methampheta- mines. Crack's sort of the big thing here (Local Agency 1 Focus Group participant).

I think drugs. Yeah... Assaults, drugs and alcohol. Assaults and attempted murders

usually stem back to something drug related or induced (Local Agency 3, Focus

Group participant).

Anything that happens, 99% of the crime can directly tie back to alcohol and drugs (Local Agency 4> Focus Group participant).

We've got a lot of flow through problems [interstate drug trafficking]... we get lots of the drug problem... (Local Agency 2, Focus Group participant).

4y Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

234 Policy Sci (2008) 41:221-243

...assaults, drugs, and alcohol. All sorts of violent assaults and attempted mur- ders... usually stem back to something... some type of drug related [problem], drug addiction, or something... (Local Agency 5, Focus Group participant).

Theme 2: Crack cocaine as a local problem associated with the black or African-American community

The content analysis of focus groups transcripts and individual interview notes also revealed the emergence of Theme 2 - crack cocaine as a local problem associated with the black or African-American community. According to participants, the drug problems in their communities were tied to specific locations or neighborhoods that seemed to correlate with a specific racial group. Of particular significance were the assertions made about African-Americans and crack cocaine:

... Here, being a small town, predominantly a lot of your drug trafficking, especially crack cocaine and stuff like that are the African- American ethnic group. They are the ones that are trafficking in cocaine (Local Agency 3, Focus Group participant).

And when you're out patrolling, as far as your drug enforcement, who are you after? You're looking for the ones that are outside hanging around on street corners and on streets. And that's the ones that you come into contact with more often. And that relates to some of this proportion of race too... the majority of the hot spots are

predominantly African- American hangouts for a lot of our community (Local Agency 2, Focus Group participant).

...a lot of our drug problems are random, but then we also have a hot spot. I know

everybody has a hot spot. But you know, if we ignore that hot spot out of fear of

saying they're profiling or they're profiling that group of people, we would see people killed. That's just the bottom line... if we didn't target that area, the rest of Kentucky would be in trouble... (Local Agency 5, Focus Group participant).

If you see a guy and you want to stop him... you've heard he's dealing dope and you know he's been in this car. You've got to fill out these forms. And you're proactively pursuing that thing and you turn in this sheet that you've stopped all these cars and

everybody's been African- American, everybody wants to send off these and say, 'Well you're profiling black people because that's all you're stopping." Well no, that's not the reason I'm stopping them. I'm stopping them because that's where the drug trade is here and I want to find out who they are and that's why I'm going to stop them (Local Agency 7, Focus Group participant).

I don't mean to say profile, but that's where it's at. I didn't make it up. I didn't make the rules for it. That's where it's at. And if you're going to target that particular problem, then you have to target those people (Local Agency 3, Focus Group participant).

It's no different than if you're actively patrolling the methamphetamine community. Those are lower-income whites and that's who you've kind of targeted... but they're not going to say, "You're racially profiling white people." Nobody's ever heard that because crack is in predominately the black community and if you attack that

£} Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 41:221-243 235

problem, then it may be interpreted as racial profiling the black community (Local Agency 4, Focus Group participant).

Theme 3: DUI as the major problem of the Hispanic community

In regards to the Hispanic community, officers made similar comments or generalizations about the perceived inclination of members of this particular racial/ethnic group to drive under the influence of alcohol. The following quotations illustrate the sentiments of par- ticipants in the various group discussions.

You get a lot of transitional, what you call temporary residents, or whatever, per- manent status and then illegals who come to the states... [their] culture, like he said, is where they're from DUI and traffic offenses are rarely enforced. So the perception is you know, driving and drinking is not a big issue. Although if you will look at it from our point of view, most of the Hispanics that get stopped usually end up with a DUI charge because to them it's normal... (Local Agency 2, Focus Group participant).

...you know we probably have a higher percentage of Hispanics in the country that

go out, work all day, they come in, have a few drinks, eat dinner at some local restaurant... almost all the Hispanic restaurants in town are within the city limits. So, you know if they're drinking there, then they got to leave. That's when we wind up catching them (Local Agency 5, Focus Group participant).

I still believe it gets back to the drinking issue because where they're from. . . the DUI enforcement is non-existent... so if you grow up with no enforcement of drinking and driving, you're going to think drinking and driving is just like breathing. You do it (Local Agency 1, Focus Group participant).

Theme 4: the intuitive nature of traffic stops as an effective strategy to address

community problems

The final theme that emerged from the analysis of focus group and individual interview data was the projected effectiveness and intuitive nature of traffic stops as a strategy to address community problems. In particular, officers were confident that traffic stops helped deter the drug trade and improve the quality of life of their local communities. This theme cut across all departmental lines.

For us, traffic stops lead to a lot of investigative tips. That could be a good way when we're having problems... (Local Agency J, Focus Group participant).

[Traffic stops as a] quality of life issue that keep drunks off the road or people who are found with drugs and driving... that' s helpful as far as with major drug traf-

ficking... (Local Agency 2, Focus Group participant).

It [traffic stops] disrupts the day-to-day street business. Through searches or...

omnipresence... you're going to run across it, it's just a matter of time (Local Agency 5, Focus Group participant).

[Traffic stops]... look at it as a proactive means for stopping drug trafficking and stuff like that... (Local Agency 4, Focus Group participant).

4y Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

236 Policy Sci (2008) 41:221-243

Fig. 2 The role of the police: community problem solvers who profile problems not people

I know two officers. . . sat in the car one night and they could just look at one guy and go, there's something wrong there. That's that intuition thing that some good street cops have (Local Agency 3, Focus Group participant).

Summary: confluence and connectedness of emergent themes

These four emergent themes seem to create a four step cycle as depicted in Fig. 2. Theme 1 identifies their local problems, Themes 2 and 3 generalize about the roots and manifes- tations of these problems and Theme 4 serves as a rational action taken by local law enforcement to address these community problems. The pattern that emerges from the confluence and connectedness of these themes reveal an overarching, under-girding, and reinforcing perception of participants - the role of police as community problem-solvers and not profilers of people.

And we just get that stereotype that you've got a bunch of white officers out here harassing the black community. For us we don't see it that way. We see it as we're attacking the problem and that's what we want to get rid of is the problem (Local Agency 7, Focus Group participant).

Ô Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 41:221-243 237

I'm targeting the problem... You 're not profiling the person. You're profiling the

problem... (Local Agency 4, Focus Group participant).

Our qualitative findings seem to suggest that the actions of officers might be based on their

perception that they are profiling the problem of illegal activities and drugs and not the racial or ethnic make-up of the person. In particular, officers irrespective of jurisdictional boundaries identified Theme 1 , drugs and DUI, as the two prevalent community problems. This theme manifested itself in two ways (as symbolized by the solid unidirectional line) and with two populations: Theme 2 - crack cocaine and the black community and Theme 3 -

DUI and the Hispanic community. To combat the officers' perceived community problems and the manifestations of these problems, Theme 4 - traffic stops as an effective strategy -

emerged from Themes 2 and 3 (as symbolized by the solid, unidirectional line). In essence, the analysis of data from the focus group discussions yielded a shared perception among officers that their primary role was one of community problem solvers who profile problems and the manifestations of such problems, but not people (see Fig. 2). Moreover, this overall shared perception seems to be reinforced and supported by the resulting interplay between Theme 4 and Themes 2 and 3 and Themes 2 and 3 with Theme 1 (as symbolized by the dual

directional, segmented lines in Fig. 2). This finding might better explain the disparities found in our quantitative analysis of search patterns of drivers by officers.

Implications of quantitative and qualitative findings

The quantitative findings from this study highlight that race does correlate with a fruitful or

productive stop but not in the manner commonly thought.15 As such, these findings reaffirm Skolnick's (1975) notion of the symbolic assailant - the association of race and/or

ethnicity as a proxy for an increased likelihood of criminal behavior - while producing a new facsimile of the assailant to include Hispanic males that now seem to accompany Skolnick's African- American male original. The findings from the qualitative phase of the

study highlight that the actions of the officers who participated in the focus groups dis- cussions are rooted in their shared perceptions that they are community problem solvers who profile the problems and not a particular racial or ethnic group. Both sets of findings provide an opportunity to offer some heuristic implications and suggest additional

approaches to investigate and better understand police behavior in the context of traffic

stops and searches.

Considering our findings, it is evident that more systematic individual level research is

necessary to explore and better understand why more African-American and Hispanic motorists had greater odds of being searched, considering that probabilities of police having positive search results reveal no difference between racial or ethnic groups, sta-

tistically or substantively. The method of collecting and analyzing this individual level data should include both quantitative and qualitative approaches.

15 The data that we analyzed did not allow us to specify the quantum of drugs seized during subsequent searches. Nonetheless, we do not mean to imply or further perpetuate the myth that every vehicle where

drugs were found was involved in trafficking large, pre-market, Operation Pipeline-size quantities of drugs with our usage of the word fruitful. Fruitful in this case means a productive traffic stop where drugs were

found and seized as a result of the subsequent search. Based upon what has been found in previous studies, we can assume that the amounts found were primarily in the after-market, personal use range or quantities of

drugs.

£} Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

238 Policy Sci (2008)41:221-243

From a quantitative perspective, additional research could examine how race, gender, rank or tenure, and exposure to the type and extent of pre- and in-service training influ- ences officers' behavior to stop and search motorists. In particular, by examining the number of years that an officer has served on the force, we would be able to test if seniority had an effect on positive search results. Also, additional test could be done quantitatively to see if the race or ethnicity of the commanding or supervising officer and the chief exec- utive officer impacted positive search results, or, where possible, if the race of the partner had an impact on positive search results.

Qualitative approaches to collecting and analyzing individual or focus group level data could also be beneficial. In particular, in-depth individual interviews and focus group discussions of officers, their supervisors, and chief executive officers could be utilized to ascertain their respective perceptions of minority motorists, bias-based policing, inclusive of racial profiling, and the efficacy of stops and searches.

Conclusion

Previous studies have typically examined the issue of racial profiling by utilizing a

quantitative approach. However, in this study we employed a mixed methods approach to

investigate interstate traffic stops and searches. This coupling resulted in three significant contributions: (1) we avoided the common methodological pitfall of trying to determine the reason for a traffic stop by focusing on the search patterns of police and the results of these searches; (2) we highlight that current policing practices are ineffective in stopping drugs and finding illegal contraband; and (3) we contextualized our quantitative findings by utilizing the focus group method to explore the perceptions of officers on traffic stops and searches and public allegations of bias-based policing.

The quantitative phase of this study highlighted that examining whether or not racial

profiling occurs is a tricky process. Analysts must use baseline population data of the neighborhoods and/or roadways to determine if police are using race as the determining factor for the reason they stop a motorist. However, since most states do not collect this baseline data, other statistically sound procedures must be developed in order to effectively analyze the traffic stop data. As demonstrated in this study, researchers can still examine racial profiling in traffic stops using the population of motorists who have already been stopped on the interstate. Focusing on searches and search results by controlling for location and reason for the stop allows researchers to focus on police behavior and what happens once the motorists are stopped.

The confluence of the quantitative and qualitative phases of this study provides a nuanced perspective of the ineffectiveness of using this profile to guide policing behavior during traffic stops. In this study, more than 80% of the searches in vehicles for African- American motorists were negative and more than 92% of searches in vehicles driven by Hispanic motorist were negative. More than 80% of the total searches conducted in this study were negative, meaning that tremendous amounts of time and resources (human and financial) are being dedicated to searches that are not yielding positive results. If police want to be efficient and effective in their efforts to stop drugs, it seems that they need to disregard the "profile" as our analysis has found that it is an ineffective tool for turning up illegal substances in motor vehicles.

Sound research can be an effective tool in safeguarding democracy because it not only looks at the consequences of our public policy, but it also examines the underlying assumptions that guided the formation of that policy. Examining assumptions is very

Ô Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 41:221-243 239

important because they have profound effects on our paradigms and institutionalized

practices based on our beliefs (Etzioni 1996; Adams et al. 1997). In the case of racial

profiling, the historical assumption by the DEA and more contemporary assumptions by officers who were focus group participants was that minority motorists are more likely to

carry drugs and engage in an additional illegal activity - driving under the influence. The evidence presented in this study and in others reveals that this assumption is false and

operating based upon this assumption is ineffective (Barlow and Barlow 2002; Meehan and Ponder 2002). The consequences of these assumptions becoming institutionalized through police practices have been devastating. The American public, particularly minority com- munities, have lost their faith that the criminal justice system, the gatekeeper of

democracy, will treat all citizens equally, without regard to race or ethnicity (Weitzer and Tuch 2002, 1999). Researchers can help examine the underlying assumptions and conse-

quences of our public policies to ensure that they uphold the principles of democracy. Staying committed to equal and effective policies will be an important first step in cor-

recting the mistrust that eroded the relationship between the government and its citizens. In order to continue examining the relationship our communities have with our criminal

justice system, we believe that more research and analysis from a mixed-methods per- spective is needed both of citizen-police interaction and of the effects these interactions have on the attitudes of citizens. Much of the research focuses on the relationships between African- Americans and police; however, as this study reveals, Hispanics are also victims of

disparate policing practices. More research is needed to explore how police view the

Hispanic community, how the police are treating Hispanic citizens, and how the Hispanic community views their relationship with the police. This research can only help to foster

greater understanding and communication between these groups, something needed in order to have a healthy, equitable democracy. Now that Hispanics have become this nation's largest minority according to the U.S. Bureau of Census (2003), it is imperative that we foster good relationships between police departments and this growing community, not only to ensure a just and healthy democracy, but also to ensure that we do not repeat our past mistakes - defining different or difference as deviant.

Acknowledgments The authors dedicate this article to the memory of the late Susette M. Talarico -

scholar, mentor, and friend. Susette read the initial draft of our manuscript and offered insightful comments and constructive suggestions. We are also grateful to Barry Bozeman, Andrew Whitford, Tony Brown, Ellen Rubin, Paul Speer, Heather Davidson and the anonymous reviewers for their suggestions for improvement.

Appendix A

Frequencies

Frequency Percent

Gender

Valid

Female 17,603 19.0

Male 75,141 81.0

Total 92,744 100.0

Missing 866

Total 93,610

â Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

240 Policy Sci (2008) 4 1 :22 1-243

Appendix A continued

Frequency Percent

Driver's race

Valid

Asian American 946 1.0

African American 1 1 ,079 1 1 .9

Hispanic 2,051 2.2

Native American 53 0. 1

White 79,331 84.9

Total 93,460 100.0

Missing 150

Total 93,610 Search

Valid

No 90,953 97.8

Yes 2,063 2.2

Total 93,016 100.0

Missing 594

Total 93,610 Search results*

Valid

Negative 2,020 80.4

Positive 492 19.6

Total 2,512 100.0

Total 93,610 a The reader will note that the total number of cases for search results exceeds the total for search by 449 cases. This discrepancy is due to a human, clerical error. For these additional cases, police officers would conduct a search and complete the information only for search results rather than completing both columns. The additional cases do not change the conclusions for two reasons. First, entering information in search results necessarily conveys that a search was conducted, making a double entry essentially redundant. Second, the cases do not affect the quantitative results. Additionally, possible concerns about officers intentionally "ghosting data," as mentioned on page 18, do not seem problematic since race was still a significant predictor of search results. If anything, this problem might underestimate the extent of profiling in this study

Appendix B: Overview of focus group interviewing

Focus groups are a non-directive interviewing technique directed towards the concerns, needs, and feelings that underlie people's opinions and preferences. With origins in market research, the focus group method has been an effective tool in social scientific research efforts - especially those that seek to get in touch with lived and perceived realities of its participants (Byars and Wilcox 1991; Krueger 1988; Jarrett 1993; Jenkins 1995; Williams 1998).

Six fundamental assumptions provide the basis for focus groups research: (1) people are a valuable source of information; (2) people can report on and about themselves; (3) the facilitator can help people retrieve forgotten information; (4) people who share a common problem will be more willing to talk amid the security of others with the same problem;

£} Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 41:221-243 241

(5) group dynamics can generate genuine information rather than establish a "group think"

phenomenon; and (6) interviewing a group is better than interviewing an individual (Stewart and Shamdasani 1990; Krueger 1988). Previous research that utilized the focus

group interviewing method in the areas of community studies, police studies, education, and health sciences have highlighted its flexibility in gathering data from a range of

participants and effectiveness in providing a holistic understanding of the perceptions, experiences, and attitudes of service providers and service recipients. Other advantages are also associated with this approach. Focus group interviewing is more efficient, allows the researcher(s) to interact directly with respondents, provides an opportunity to capture real life data in large and rich amounts, allows respondents to react and build upon the

responses of other group members, and the results are easy to understand. There are also a number of limitations to the focus group method. Primarily, these

limitations include difficulty in assembling groups, lack of generalization to a larger population due to the small number of participants, and 'group think' and/or domination by an opinionated member (Stewart and Shamdasani 1990; Morgan 1988).

References

Adams, G., & Balfour, D. (1998). Unmasking administrative evil. Thousand Oaks, CA: Sage. Adams, B., Bell, L., & Griffin, P. (1997). Teaching diversity and social justice. New York, NY: Routledge. Alpert, G. P., Dunham, R. G., & Smith, M. R. (2007). Investigating profiling by the Miami-Dade Police

Department: A multimethod approach. Criminology and Public Policy, 6, 25-55. doi: 10.1 1 1 1/J.1745-9133.2007.00420.X.

Amnesty International USA. (2004). Threat and humiliation: Racial profiling, domestic security, and human rights in the United States. New York, NY: Amnesty International.

Agresti, A., & Finlay, B. (1997). Statistical methods for the social sciences (3rd ed.). Upper Saddle River, NJ: Prentice Hall.

Barlow, D., & Barlow, M. H. (2002). Racial profiling: A survey of African-American police officers. Police Quarterly, 5, 334-358. doi:10.1 177/109861 102129198183.

Barnes, K. (2005). Assessing the counterf actual: The efficacy of drug interdiction absent racial profiling. Duke Law Review, 54, 1089-1 141.

Buerger, M. E., & Farrell, A. (2002). The evidence of racial profiling: Interpreting documented and unofficial sources. Police Quarterly, 5, 275-305.

Byars, P., & Wilcox, J. (1991). Focus groups: A qualitative opportunity for researchers. Journal of Business Communication, 28, 63-77. doi: 10.1 177/002 194369 102800 1Ö5.

Close, B. R., & Mason, P. L. (2003). Beyond the traffic stop: Intersections of race, ethnicity, gender, and the decision to search. Tallahassee, FL: Florida Department of Transportation.

Cordner, G., Williams, B., & Zuniga, M. (2000). Vehicle stops for the year 2000: Executive summary. San Diego, CA: San Diego Police Department.

DeMaris, A. (1995). A tutorial in logistic regression. Journal of Marriage and the Family, 57, 956-968. doi: 10.2307/353415.

Dunham, R., Alpert, G., Stroshire, M., & Bennett, K. (2005). Transforming citizens into suspects: Factors that influence police suspicion. Police Quarterly, 8, 366-393. doi: 10.1 177/109861 1 105274539.

Durose, M., Schmitt, E., & Lanagan, P. (2005). Contacts between the police and the public: Findings from the 2002 national survey. Washington, DC: Bureau of Justice Statistics/Department of Justice.

Engel, R. S., & Calnon, J. M. (2004). Comparing benchmark methodologies for police-citizen contacts: Traffic stop data collection for the Pennsylvania state police. Police Quarterly, 7, 97-125. doi: 10.1177/1098611103257686.

Engel, R. S., Calnon, J. M., & Bernard, T. J. (2002). Theory and racial profiling: Shortcomings and future directions in research. Justice Quarterly, 19, 250-273. doi: 10. 1080/074 18820200095 231.

Etzioni, A. (1996). The new golden rule: Community and morality in a democratic society. New York, NY: Basic Books.

Goodey, J. (2006). Ethnic profiling, criminal (injustice and minority populations. Critical Criminology, 14, 207-212. doi:10.1007/sl0612-006-9010-4.

Gray, M. (1998). Drug crazy. New York, NY: Random House.

£} Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

242 Policy Sci (2008) 41:221-243

Gross, S. R., & Barnes, K. Y. (2002). Road work: Racial profiling and drug interdiction on the highway. Michigan Law Review, 101, 651-754. doi: 10.2307/1290469.

Harris, D. A. (2006). U.S. experiences with racial and ethnic profiling: History, current issues, and the future. Critical Criminology, 14, 213-239. doi :10.1007/s 106 12-006-90 1 1-3.

Hernandez-Murillo, R., & Knowles, J. (2004). Racial profiling or racist policing: Bounds test for aggregate data. International Economic Review, 45, 959-989. doi:10.1111/j.0020-6598.2004.00293.x.

Jarrett, R. (1993). Interviewing with low-income minority populations. In D. L. Morgan (Ed.), Successful focus groups: Advancing the state of the art (pp. 184-201). Newbury Park, CA: Sage.

Jenkins, M. (1995). Fear of the Gangsta: African-American males and the criminal justice system. Unpublished paper presented at the Annual Meeting of the Academy of Criminal Justice Sciences, Boston, MA.

Knodel, J. (1993). The design and analysis of focus group studies: A practical approach. In D. L. Morgan (Ed.), Successful focus groups: Advancing the state of the art (pp. 35-50). Newbury Park, CA: Sage.

Krippendorf, K. (1980). Content analysis: An introduction to its methodology. Beverly Hills, CA: Sage. Krueger, R. (1988). Focus groups: A practical guide for applied research. Newbury Park, CA: Sage. Lamberth, J. (1996). A report to the ACLU. New York, NY: American Civil Liberties Union. Langan, P., Greenfield, L. A., Smith, S., Durose, M., & Levin, D. (2001). Contacts between police and the

public: Findings from the 1999 national survey, NCJ 184957. Lombroso, C. (191 1). Crime: Its causes and remedies. Boston, MA: Little, Brown, and Company. MacDonald, H. (2002). The racial profiling myth debunked. City Journal (New York, N.Y.), 12. Available at

http://www.city-journal.org/html/! 2_2_the_racial_profiling.html. Meehan, A. J., & Ponder, M. C. (2002). Race and place: The ecology of racial profiling and African

American motorists. Justice Quarterly, 19, 399-430. doi: 10. 1080/074 18820200095 291. Miller, P., & O'Leary, T. (1989). Hierarchies and American ideals, 1900-1940. Academy of Management

Review, 14, 250-256. doi: 10.2307/258419. Morgan, D. (1988). Focus groups as qualitative research. Newbury Park, CA: Sage. Pampel, F. C. (2000). 'Logistic regression: A primer', Sage University papers series on quantitative

applications in the social sciences, 07-132. Thousand Oaks, CA: Sage. Petrocelli, M., Piquero, A., & Smith, M. (2003). Conflict theory and racial profiling: An empirical analysis

of police traffic stop data. Journal of Criminal Justice, 31, 1-1 1 . doi: 10.1 01 6/S0047-2352(02)00 195-2. Ramirez, D., McDevitt, J., & Farrell, A. (2000). A resource guide on racial profiling data collection systems:

Promising practices and lessons learned. Washington, DC: United States Department of Justice. Ridgeway, G. (2006). Assessing the effect of race bias in post-traffic stop outcomes using propensity scores.

Journal of Quantitative Criminology, 22, 1-29. doi: 10. 1007/s 10940-005-9000-9. Rojek, J., Rosenfeld, R., & Decker, S. (2004). The influence of the driver's race on traffic stops in Missouri.

Police Quarterly, 7, 126-147. doi: 10.1 177/109861 1 103260695. Schmitt, E., Lanagan, P., & Durose, M. (2002). Characteristics of drivers stopped by police, 1999.

Washington, DC: Bureau of Justice Statistics/Department of Justice. Skolnick, J. (1975). Justice without trial. New York, NY: Wiley. Smith, M., & Alpert, G. (2002). Searching for direction: Courts, social science, and the adjudication of racial

profiling claims. Justice Quarterly, 19, 673-703. doi: 10. 1080/074 18820200095391. Smith, M., & Petrocelli, M. (2001). Racial profiling? A multi variate analysis of police traffic stop data.

Police Quarterly, 4, 4-27. Stahl, M. (2003). Unlawful entry: Examining racial profiling through police search practices. Unpublished

Master's of Science Thesis in Community Research and Action, Vanderbilt University. State of New Jersey v. Pedro Soto, et al. (1996). 324 N.J. Super. 66, 734 A.2d 350. Stewart, D., & Shamdasani, P. (1990). Focus groups: Theory and practice. Newbury Park, CA: Sage. Taylor, J., & Whitney, G. (2002). Racial profiling: Is there an empirical basis? The Mankind Quarterly, 42,

285-312. U.S. Bureau of the Census. (2003). Social and demographic statistics. Washington, DC. U.S. General Accounting Office. (2000). Racial profiling: Limited data available on motorist stops. Report

to the Honorable James E. Clyburn, Chairman, Congressional Black Caucus. G AO/GGD-00-4 1 . Washington, DC.

Walker, S. (2001). Searching for the denominator: Problems with police traffic stops data and an early warning system solution. Justice Research and Policy, 3, 63-95. doi:10.3818/JRP.3.l.2001.63.

Webb, G. (1999). DWB. Esquire (New York, NY), April, 1 18-127. Weitzer, R., & Tuch, S. (1999). Race, class and perceptions of discrimination by the police. Crime and

Delinquency, 45, 494-507. doi: 10.1 177/001 1 128799045004006. Weitzer, R., & Tuch, S. (2002). Perceptions of racial profiling: Race, class and personal experience.

Criminology, 40, 435-456. doi.10.1 1 1 1 /j.l 745-9 125.20O2.tb0O962.x.

Ô Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

Policy Sci (2008) 41:221-243 243

Williams, B. (1998). Citizen perspectives on community policing: A case study in Athens. Georgia, Albany, NY: State University of New York Press.

Williams, B. (2000). Racial profiling: The personal costs and societal consequences of driving while black: Implications for the minority academic community. In L. Jones (Ed.), Brothers of the academy: Earning our way (pp. 263-276). Herndon, VA: Stylus Publishers.

Williams, H., & Murphy, P. V. (1990). The evolving strategy of policing: A minority view. In Perspectives on policing. Washington, DC: National Institute of Justice.

£) Springer

This content downloaded from 198.137.20.62 on Thu, 30 Apr 2015 17:41:04 UTCAll use subject to JSTOR Terms and Conditions

top related