Police Stop and Frisk and the Impact of Race: A Focal Concerns Theory Approach

Abstract

The findings of this study outline the racial differences in stop and frisk decisions by Illinois officers in consent searches and those based upon reasonable suspicion within the context of the elements of focal concerns theory. The analysis for this study was performed using propensity score matching (PSM) and allowed the researchers to create a quasi-experimental design to examine the race of the citizen and police decision making. According to our analysis of official Illinois law enforcement data, Black citizens, particularly males, were less likely to give their consent to a stop and frisk search. Black male citizens were also more likely to be stopped and searched due to an assessment of reasonable suspicion by the officer. Elements of focal concerns theory were also factors in pedestrian stops under conditions of consent and reasonable suspicion. Citizens judged as blameworthy were more likely to be stopped and frisked under conditions of consent and reasonable suspicion. The effect of a verbal threat and the officer’s prior knowledge about the citizen had even more significant impacts.

1. Introduction

The Terry v. Ohio (1968) decision instructed how police could lawfully make traffic and pedestrian stops. Probable cause that a person was engaging in crime was now supplemented by “reasonable suspicion” in an attempt to protect the public and the officers from weapons and danger as a justification for a stop. Here, we examine the outcome of stop and frisk pedestrian stops in Illinois. The research compares outcomes in light of the pedestrian’s race, and whether the search was conducted via consent or reasonable suspicion. Focal concerns theory was applied to consider the factors influencing the police decision to stop and frisk.

2. Stop and Frisk Literature Review

Studies on the effect of stop and frisk have focused on data from the New York City Police Department. NYPD made stop, question, and frisk (SQF) a centerpiece of their crime control strategy that drove the city’s crime rates to historic lows that were furthered under Mayor Michael Bloomberg and Police Commissioner Raymond Kelly. However, the practice led to citizen complaints and a legal decision (Floyd et al. v. City of New York 2013) that the NYPD was engaging in unconstitutional SQF practices that focused upon Blacks and Latinos (White and Fradella 2016, p. 104).

Accordingly, NYPD SQF studies focus upon the impact of the policy on race and crime control. For example, a comparative analysis of approximately 125,000 pedestrian stops made by the NYPD over 15 months (January 1998–March 1999) revealed that Blacks and Hispanics were stopped more often than Whites (OR = 1.5 and OR = 2.5, respectively)—even after controls for precinct crime variability and race-specific crime estimates (weapons offenses, violent and drug crimes) were considered (Gelman et al. 2007). Avdija (2014) analyzed NYPD data from 2006 to examine the factors involved in the decision to stop and frisk citizens. The top official reasons for making a stop included: location in a high crime area, time of day, type of crime incident, fugitive movements, casing a victim or location, and proximity to a crime scene (Avdija 2014, p. 30). Males (83.7%) were much more likely than females (7.1%) to be stopped (Avdija 2014, p. 32). Latino (OR = 1.73) and Black suspects (OR = 1.68) were more likely to be searched than Whites (Avdija 2014, p. 33). Fagan and Geller (2015) examined data from 4.7 million NYPD street stops (from 2004 to 2012) to consider police suspicion elements. The top reason for conducting a pedestrian stop was crime location (a high crime area)—nearly 75 percent of all cases and a “convenient script (i.e., stops were justified for being in high crime area) that is difficult to challenge” (Fagan and Geller 2015, p. 84).

Focusing upon racial discrimination, Ferrandino (2015) tested the minority threat hypothesis in examining NYPD street stops. In predominantly White neighborhoods, Blacks were more likely than other racial groups to be subjected to frisks and the use of force by police—supporting the view that they are “out of place” (Ferrandino 2015, pp. 225–26). Ferrandino (2015, 2018) also examined the NYPD stop and frisk policy’s effectiveness and equity from 2003 to 2014. SQF had mixed results. It was largely ineffective in reducing crime, but was also somewhat free of racial bias in frisk decisions. Specifically, Ferrandino determined that persons of color were “out of place” (Ferrandino 2015, p. 226) and subjected to more stops and were frisked more often, but fewer were searched and arrested as a result, while crime increased (Ferrandino 2018, p. 132). Using NYPD data from 2012, Zhao et al. (2019) examined the impact of racial segregation on the risk of four policing outcomes for Blacks and Hispanics (frisk, search, use of force, and arrest). The findings indicated that police stops of Blacks were more likely to involve all four outcomes in areas with high levels of Black–White segregation. However, police stops of Hispanics were less likely to result in the four outcomes in areas with high levels of Hispanic–White segregation (Zhao et al. 2019, p. 378).

However, some evidence shows that the Floyd et al. v. City of New York (2013) decision has affected stop, question, and frisk internal and external outcomes. MacDonald and Braga (2019) reported that hit rates (i.e., the percentage of searches that turn up contraband) and other outcomes for minority groups were much lower after the decision. The NYPD adopted a new neighborhood policing plan that may have improved results.

The effectiveness of SQF as a crime control measure was also addressed in the literature. An analysis of NYPD data from the Bronx in 2006 (150 days) examined the impact of SQF practices on crime. The results indicated that SQFs produced a modest reduction in crime that extended over three days with a diffusion of benefits that extended 300 feet from the point of the SQF (Wooditch and Weisburd 2016, p. 191). Research on NYPD SQF stops over seven years found no relationship between the number of stops and burglary and robbery rates (Rosenfeld and Fornagno 2014, p. 116). However, a second study of NYPD SQF stops by the authors over the period 2006–2011 reported statistically significant reductions in violent and property crimes across all racial and ethnic groups (Rosenfeld and Fornango 2017, p. 946). A deterrent effect was confirmed in an NYPD SQF stop analysis for the same period within small areas and across short periods (Weisburd et al. 2016, p. 46). Weisburd et al. (2014) examined NYPD SQF stops conducted from 2009 to 2010 to determine their contribution to hotspot policing during a time of declining police resources. They determined that SQFs increased during this period in crime hotspots. However, they found “no evidence that NYPD emphasized procedural justice and legitimacy evaluations in its efforts to police crime hotspots” (Weisburd et al. 2014, p. 147).

The use of force during SQF was another salient issue. An analysis of over two million NYPD stops from 2007 to 2014 revealed that Blacks (especially youths) were more likely to experience a potential use of force when engaging in criminal activity. The stop’s productivity was not related to this disparity (Kramer and Remster 2018, p. 986). The police viewed Black youths as a potential threat. A similar study of over 500,000 NYPD stops in 2010 found that Blacks and Latinos were more likely to be frisked and be subjected to force—despite precinct-level crime differences (Levchak 2017, p. 400). A study of NYPD SQF stops in 2012 examined the relationship between race and police use of force. Although the use of force was a rare event (14% total, 0.01% weapon force), Black and Hispanic citizens were more likely to be the object of police non-weapon force than White citizens (Morrow et al. 2017, pp. 378–79).

Other studies examined SQF in different locations, also focusing upon its impact on crime and racial bias. A study of official data on stop and frisk practices in Boston determined that minority neighborhoods were subjected to higher field interrogation and surveillance levels—even after controlling for crime levels and other social factors. Despite controls for gang membership and criminal history, Black suspects were targeted more often than Whites. White police officers had higher rates of interrogations and were more likely to stop and frisk subjects of all races. Officers of all races were more likely to take action against a minority subject when the officer and subject’s race differed (Fagan et al. 2016). Ashby and Thompson (2017) compared police stop activity under routine activities theory in London and New York from 2006 to 2011. They determined that the number of offenders on the street did not drive the decision to stop and search. Instead, special events (such as Halloween) appeared to influence police activity (Ashby and Thompson 2017, p. 126). In their study of 144 drug-related police stops in St. Louis, Gaston and Brunson (2020) found evidence of “ecological contamination (i.e., when the police perceive persons encountered in high crime areas as likely offenders)” in pedestrian stops. Even in high crime areas, Blacks were more likely to be stopped regardless of a lack of evidence of their suspected involvement in criminal behavior. At the same time, Whites were detained only when they were engaged in drug dealing (Gaston and Brunson 2020, p. 215).

Attitudes toward SQF was also an analytical focus. Goff and Rau (2020) offer theoretical explanations for biased policing by combining perspectives from social psychology and routine activities theory. Police are typically engaged in risky situations and places with persons who often elicit disgust and feelings of out-group derogation. As a result, these perceptions put them at risk of engaging in aggressive or biased citizens’ responses.

In terms of citizen opinions, a survey of New York City pedestrians found that youths and minorities held more negative attitudes toward the practice of stop, question, and frisk. At the same time, persons with a higher level of education, full-time employment, were married, personally knew a police officer, had never been and never knew anyone who had been stopped and frisked, and had more knowledge about the procedure expressed positive support for the practice (Evans and Williams 2017). A survey of young men stopped by the NYPD examined their perceptions of police legitimacy. The findings indicated that higher police legitimacy perceptions were related to lower levels of criminal activity and greater levels of cooperation with the police. Citizen perceptions of fair treatment by the police were directly related to beliefs that the police were legitimately conducting street stops. However, this perception changed with the number of stops experienced by the respondents—the greater the number of stops, the less likely the belief in police legitimacy (Tyler et al. 2014, p. 776).

3. Focal Concerns Theory

The application of a theoretical explanation for officer decision making is important because it can help provide an explanation for how an officer interacts with a citizen. Utilizing theory in the context of racial profiling could potentially explain officers’ decision making, especially when considering the potential for bias. Although theories have been applied in racial profiling for traffic stops and in the stop and frisk research, Engel et al. (2002) criticized the research in criminal justice/criminology as being largely atheoretical. One theory used in understanding prosecutor’s and judge’s decision making is focal concerns theory; Tillyer and Hartley (2010) made the argument that focal concerns theory could be applied to police officer decision making.

Steffensmeier (1980) developed focal concerns theory to explain the decision making of court actors specifically for sentencing. Focal concerns theory was then expanded by Steffensmeier et al. (1998) to emphasize three components that impact decision making for sentencing. The first is blameworthiness and focuses on the individual’s level of potential responsibility in committing the crime and making sure the punishment fits the crime. However, it is not the officer’s job to determine if the citizen is guilty; an officer will only establish probable cause. A citizen is blameworthy when the officer determines that probable cause is present. The second component is protecting the community. It assesses the offender’s future behavior and determining if they would be a danger to society (Steffensmeier et al. 1998). Practical constraints and consequences are the final components. In the sentencing literature, this measure is based on the criminal justice system’s incurred costs (Steffensmeier et al. 1998).

Due to the overload that judges may face, focal concerns theory was also created to deal with the uncertainty in sentencing (Albonetti 1991). Due to a potential in case overload, judges may develop a “perceptual shorthand” that may result in the judge connecting certain attributes or actions that would indicate criminality. Police officers may face a choice similar to judges when it comes to conducting a stop and frisk because they may not have all the pertinent information to conduct this type of stop. According to Smith and Alpert (2007), officers may develop a shorthand to help with simplifying the situation, including characteristics such as the citizens’ race, gender, and age. This shorthand and how prior interactions have occurred with certain citizens could create a profile of how an officer would potentially interact with a particular citizen.

Prior research in the racial profiling literature regarding police decision making and traffic stops has shown the utility of applying focal concerns theory (Higgins et al. 2012; Vito et al. 2017, 2018, 2019). This study seeks to advance upon prior work on stop and frisk using focal concerns theory to explain officers’ decision making and use propensity score matching for the statistical analysis.

4. Current Study

Previous research regarding studies examining racial bias has been critical of the methodologies or data collection practices used. Prior scholars acknowledge that police officers or police departments may falsify traffic stop data (Lundman 2010; Meehan and Ponder 2002; Skolnick and Caplovitz 2004). Citizens also have concerns about traffic stops, including whether the citizen’s demeanor when interacting with law enforcement impacts the stop outcome or if the citizen even views the stop as legitimate (Lundman and Kaufman 2003; Reisig et al. 2004). The third issue is that specific statistical techniques used in prior studies, including the benchmark test or hit-rate, is concerning because researchers are not able to provide an appropriate benchmark that would determine a comparison group of non-stopped drivers to stopped drivers (Engel and Calnon 2004; Neil and Winship 2019; Ridgeway and MacDonald 2010).

Researchers acknowledge methodological or data collection issues when studying racial profiling and recommend propensity score matching (PSM) as a statistical technique that would improve studying the issue of racial profiling (Fallik 2019; Neil and Winship 2019). By applying PSM, the researcher would better understand if the race of the citizen “matters”. Thus, this shows how race influences police officer decision making during pedestrian stops. Previous research has demonstrated the advantage of using PSM in racial profiling studies (Ridgeway 2006; Vito et al. 2018, 2020).

By applying PSM in the current study, researchers can create a quasi-experimental design that will allow for similarly situated pedestrian stops based on the citizen’s race. This application is essential because no prior study, to the best of the researchers’ knowledge of pedestrian stops, has applied PSM, which will enable the researchers to better consider the impact that a citizen’s race has on whether an officer conducts a stop and frisk based on the citizen’s consent or reasonable suspicion. This study seeks to answer five research questions:

Research Question 1: What effect does a citizen’s race have on whether the citizen will provide the officer consent to conduct a stop and frisk?

Using PSM, the researchers can better examine the observable racial differences amongst similarly situated citizens (i.e., Black citizens vs. White citizens) in pedestrian stops where a stop and frisk occurs because of citizen consent.

Research Question 2: What effect does being a male citizen and the citizen’s race have on the chances that a male citizen will provide the officer consent to conduct a stop and frisk?

In answering the second research question, the gender of the citizen is male in conjunction with the race of the citizen and using PSM.

Research Question 3: What effect does a citizen’s race have on the officer’s chances to conduct a stop and frisk based on reasonable suspicion?

By applying PSM, the researchers can better consider the observable racial differences amongst similarly situated citizens (i.e., Black citizens vs. White citizens) and the chances that the officer conducts a stop and frisk because of reasonable suspicion.

Research Question 4: What effect does being a male citizen and the citizen’s race have on the officer’s chances to conduct a stop and frisk based on reasonable suspicion?

To answer the fourth research question, the gender of the citizen is male in conjunction with the race of the citizen and using PSM.

Research Question 5: What is the relationship between the three components of focal concerns theory and a stop and frisk occurring based on citizen consent or if the officer conducts a stop and frisk because of reasonable suspicion?

All three components of focal concerns theory were measured to address this research question. The aim is to determine if there are theoretical components of the stop and frisk that explain officer decision making in consent cases or stop and frisks based on reasonable suspicion beyond the race of the citizen or the race and gender of the citizen.

5. Data

In 2004, the state of Illinois passed The Illinois Traffic and Pedestrian Stop Statistical Study. Due to this act, law enforcement must document and report their traffic stop data to the Illinois Department of Transportation. The act was expanded in 2017 to include pedestrian stops conducted by police. The data for this study came from the entire state of Illinois, including both state and local agencies. The data used in this study only came from those agencies that reported their data. This study’s data are a sample of pedestrian stops (n = 276,544) in 2017 and 2018.

6. Dependent Measure

The current study has two dependent measures. The first is whether the stop and frisk was conducted based on citizen consent (0 = no, 1 = yes). The second dependent measure is whether the officers’ stop and frisks were conducted based on reasonable suspicion (0 = no, 1 = yes).

7. Independent Measures

The independent variables for this study related to focal concerns theory. Focal concerns theory has three components (Steffensmeier et al. 1998). The first component is blameworthiness and was operationalized in the current study using four measures that are reasons for frisk: violent actions, violent crime, evasive actions, or suspicious bulge (0 = no, 1 = yes). These four measures were added together to create an index of blameworthiness ranging from 0 to 4, with higher scores indicating a greater degree of blameworthiness (a similar measure has been used in other studies: Higgins et al. 2012; Vito et al. 2019). The second component is protecting the community and was measured by whether the reason for a frisk was because of a verbal threat (0 = no, 1 = yes). The final component is practical constraints and consequences and was operationalized using one measure that asked if the reason for a frisk was because of prior knowledge (0 = no, 1 = yes) (a similar measure has been used in other studies: Higgins et al. 2012; Vito et al. 2019).

The demographics of the citizen were captured as control variables including citizen race (0 = White, 1 = Black) and citizen gender (0 = female, 1 = male).

8. Analysis Plan

The statistical analysis for this study was propensity score matching (PSM). The utility of using PSM is that it allows researchers to better consider the impact that a citizen’s race has on influencing the officer’s decision making (Bai and Clark 2018; Guo and Fraser 2014). The PSM analysis was performed in Stata 15. The PSM process was completed in four stages, with the first stage being the calculation of the descriptive statistics for all measures and using the mean value to interpret each measure.

Stage two was to match the passengers based on their race. In the current study, the matching was performed using the nearest neighbor technique (one-to-one matching without replacement) and allowed the researchers to assign one citizen in the treatment group to every citizen in the control group. A caliper (i.e., standard deviation) of 0.20 was also used (Rosenbaum and Rubin 1985).

The third stage assesses the quality of the matching based on two pieces of information. The first piece is that the mean value for the matched groups should be identical or almost identical. The second piece is that the standardized bias after matching must be ≤10 percent for each variable, or the mean bias after matching must be ≤10 percent for the PSM to be acceptable (Rosenbaum and Rubin 1985). Within this stage’s application, the researchers considered the descriptive statistics for the treatment group (i.e., Black citizens) and the comparison group (i.e., White citizens). If PSM were not used, then the results would be biased because the sample of Black and White citizens is different in this study’s measures of interest. The current study’s use of PSM removes the bias among these measures, resulting in similarly situated Black and White citizens. Thus, this shows that the remaining differences should be attributed mainly to the citizen’s race. Weighted logistic regression analysis was used to examine this issue.

The final stage is to conduct a logistic regression based on the weighted matches from the PSM. When interpreting the current study’s findings, the odds ratios were used. They showed the likelihood that an officer conducts a stop and frisk of the citizen based on consent or reasonable suspicion. The treatment variable for this is the citizen’s race; the cases were matched on all other variables in the model to better show how the citizen’s race may impact a stop and frisk based on consent or reasonable suspicion.

9. Results

Before conducting the analysis, crosstab results are presented to show the racial and gender distribution of the data before using PSM. Black citizens (82.0%) said no more often to a stop and frisk conducted by consent, whereas White citizens (22.90%) more often said yes. Female pedestrians (89.30%) said no at a higher rate to stop and frisks conducted by consent versus male pedestrians (20.0%) who more often reported saying yes. Among male citizens, Black males (80.80%) reported a higher level of saying no to stop and frisks conducted by consent compared to White males saying yes (25.0%).

Table 1. Descriptive statistics and matching for race: stop and frisk based on consent.

	Before Propensity Score Matching			After Propensity Score Matching
Covariates	Treated	Control	SB (%)	Treated	Control	SB (%)
Male	0.87	0.77	26.60	0.77	0.77	−0.40
Blameworthiness	0.21	0.13	16.90	0.13	0.13	−0.70
Frisk Reason was a Verbal Threat	0.01	0.01	0.50	0.001	0.01	−7.00
Frisk Reason was Prior Knowledge	0.03	0.02	2.90	0.0001	0.02	−14.30
Sample	Mean Bias	Median Bias
Unmatched	11.70	9.90
Matched	5.60	3.80

shows the descriptive statistics for the sample of pedestrian stops for the model between Black and White citizens and a stop and frisk occurring based on citizen consent. Before using PSM, the mean standardized bias percentage was 11.70, and after utilizing PSM, the mean standardized bias percentage was 5.60.

Table 2. Weighted logistic regression: stop and frisk based on consent.

Measure	b	S.E.	Exp(b) (95% Confidence Interval)
Black Citizen	−0.51 ***	0.02	0.60 (0.58 to 0.62)
Male	0.49 ***	0.02	1.63 (1.55 to 1.71)
Blameworthiness	0.97 ***	0.02	2.65 (2.55 to 2.76)
Frisk Reason was a Verbal Threat	−0.10	0.14	0.90 (0.69 to 1.18)
Frisk Reason was Prior Knowledge	0.76 ***	0.07	2.14 (1.85 to 2.46)
2 Log-likelihood = −36,361.33
Chi-Squared = 3921.52 ***
McFadden’s R-Squared = 0.05
n = 78,816

*** p < 0.001.

presents the weighted logistic regression results based on matching for Black and White citizens and the police officer conducting a stop and frisk based on citizen consent. This analysis was performed to answer the first research question, “What effect does a citizen’s race have on whether the citizen will provide the officer consent to conduct a stop and frisk?” The findings show that Black citizens are 40 percent less likely to consent to a stop and frisk. The findings presented in Table 2 also address research question five, “What is the relationship between the three components of focal concerns theory and a stop and frisk occurring based on citizen consent or if the officer conducts a stop and frisk because of reasonable suspicion?” The results show partial support for focal concerns theory, providing a theoretical explanation for police officer decision making in stop and frisks occurring by consent. Citizens who are viewed as more blameworthy are more likely to have a stop and frisk occur based on consent (b = 0.97, S.E. = 0.02, Exp(b) = 2.65). A police officer who has prior knowledge of the citizen (i.e., practical constraints and consequences) makes it 114 percent more likely that a stop and frisk will occur because of citizen consent. The only other significant variable was that male citizens are 63 percent more likely to have a stop and frisk take place by giving consent to search.

Table 3. Descriptive statistics and matching for race and gender: stop and frisk based on consent.

	Before Propensity Score Matching			After Propensity Score Matching
Covariates	Treated	Control	SB (%)	Treated	Control	SB (%)
Blameworthiness	0.23	0.16	15.40	0.15	0.16	−1.00
Frisk Reason was a Verbal Threat	0.001	0.001	0.50	0.003	0.01	−5.20
Frisk Reason was Prior Knowledge	0.03	0.03	2.20	0.0001	0.03	−15.80
Sample	Mean Bias	Median Bias
Unmatched	6.00	2.20
Matched	7.40	5.20

provides the descriptive statistics for the sample of pedestrian stops where the sample was restricted to only include males because of a positive connection with stop and frisks based on citizen consent. Before using PSM, the mean standardized bias percentage was 6.00, and after utilizing PSM, the mean standardized bias was 7.40.

The weighted logistic regression results based on matching for Black male and White male citizens and the police officer conducting the stop and frisk based on consent are provided in

Table 4. Weighted logistic regression: stop and frisk based on consent for male citizens.

Measure	b	S.E.	Exp(b) (95% Confidence Interval)
Black Male Citizen	−0.49 ***	0.02	0.61 (0.59 to 0.64)
Blameworthiness	0.88 ***	0.02	2.40 (2.30 to 2.50)
Frisk Reason was a Verbal Threat	0.40 ***	0.12	1.49 (1.17 to 1.90)
Frisk Reason was Prior Knowledge	0.71 ***	0.08	2.04 (1.75 to 2.36)
2 Log-likelihood = −29,845.97
Chi-Squared = 2509.67 ***
McFadden’s R-Squared = 0.04
n = 60,714

*** p < 0.001.

. The second research question asked, “What effect does being a male citizen and the citizen’s race have on the chances that a male citizen will provide the officer consent to conduct a stop and frisk?” The current findings show that Black male citizens are 39 percent less likely to have a stop and frisk occur because Black male citizens do not give officers the consent to search. The findings in Table 4 lend support, showing that focal concerns theory does provide a theoretical explanation for stop and frisk cases in terms of officer decision making for male citizens. If the officer considers a male citizen blameworthy, it is 140 percent more likely that a stop and frisk will occur following consent. If a male citizen makes a verbal threat, then a police officer is 49 percent more likely to conduct a stop and frisk following consent because of the perceived need to protect the community. Police officers who have prior knowledge concerning a male citizen are 104 percent more likely to conduct a stop and frisk following consent—consistent with practical constraints and consequences.

Crosstabs were performed again when the dependent variable was a stop and frisk conducted based on reasonable suspicion before using PSM for the analysis. Police officers reported a lower number of stop and frisks conducted by reasonable suspicion for White citizens (89.90%), but a higher number of stop and frisks conducted by reasonable suspicion for Black citizens (17.70%). When considering the gender of the citizen, police officers reported a lower number of stop and frisks based on reasonable suspicion for females (93.60%) and a higher number of stop and frisks based on reasonable suspicion for males (17.60%). Considering the race and gender of the citizen, police officers reported a lower number of stop and frisks based on reasonable suspicion for White male citizens (88.40%), but a higher number of stop and frisks based on reasonable suspicion for Black male drivers (19.30%).

Table 5. Descriptive statistics and matching for race: stop and frisk based on reasonable suspicion.

	Before Propensity Score Matching			After Propensity Score Matching
Covariates	Treated	Control	SB (%)	Treated	Control	SB (%)
Male	0.87	0.77	26.60	0.77	0.77	−0.40
Blameworthiness	0.21	0.13	16.90	0.13	0.13	−0.70
Frisk Reason was a Verbal Threat	0.01	0.01	0.50	0.01	0.01	−7.00
Frisk Reason was Prior Knowledge	0.03	0.02	2.90	0.0001	0.02	−14.30
Sample	Mean Bias	Median Bias
Unmatched	11.70	9.90
Matched	5.60	3.80

shows the descriptive statistics for the sample of pedestrian stops for the model between Black and White citizens and a stop and frisk occurring because of reasonable suspicion. Before utilizing PSM, the mean standardized bias was 11.70, and after utilizing PSM, the mean standardized bias percentage was 5.60.

Table 6. Weighted logistic regression: stop and frisk based on reasonable suspicion.

Measure	b	S.E.	Exp(b) (95% Confidence Interval)
Black Citizen	0.51 ***	0.03	1.66 (1.58 to 1.76)
Male	0.65 ***	0.04	1.92 (1.78 to 2.07)
Blameworthiness	2.50 ***	0.03	12.17 (11.57 to 12.80)
Frisk Reason was a Verbal Threat	1.57 ***	0.17	4.82 (3.47 to 6.71)
Frisk Reason was Prior Knowledge	2.12 ***	0.09	8.34 (7.05 to 9.86)
2 Log-likelihood = −21,445.74
Chi-Squared = 13,623.82 ***
McFadden’s R-Squared = 0.24
n = 78,816

*** p < 0.001.

presents the weighted logistic regression findings based on matching for Black and White citizens for the occurrence of stop and frisk due to the officers’ perception of reasonable suspicion. The third research question asked, “What effect does a citizen’s race have on the officer’s chances to conduct a stop and frisk based on reasonable suspicion?” The results show that Black citizens are 66 percent more likely to have a stop and frisk take place under reasonable suspicion. These results also support focal concerns theory (i.e., research question five) as a theoretical explanation for officer decision making. Citizens viewed as more blameworthy are 12.17 times more likely to have a stop and frisk occur because of reasonable suspicion. If a citizen makes a verbal threat, then an officer is 382 percent more likely to conduct a stop and frisk due to reasonable suspicion and the need to protect the community. An officer who has prior knowledge of a citizen (i.e., practical constraints and consequences) makes it 734 percent more likely that a stop and frisk will occur based on reasonable suspicion. Male citizens are 92 percent more likely to have a stop and frisk occur due to reasonable suspicion of the officer.

Table 7. Descriptive statistics and matching for race and gender: stop and frisk based on reasonable suspicion.

	Before Propensity score Matching			After Propensity Score Matching
Covariates	Treated	Control	SB (%)	Treated	Control	SB (%)
Blameworthiness	0.23	0.16	15.40	0.15	0.16	−1.00
Frisk Reason was a Verbal Threat	0.01	0.01	−0.50	0.003	0.01	−5.20
Frisk Reason was Prior Knowledge	0.03	0.03	2.20	0.0001	0.03	−15.80
Sample	Mean Bias	Median Bias
Unmatched	6.00	2.20
Matched	7.40	5.20

presents the descriptive statistics for the sample of pedestrian stops where the sample was restricted to include only male citizens because of the positive connection with stop and frisks based on reasonable suspicion. Before utilizing PSM, the mean standardized bias percentage was 6.00, and after utilizing PSM, the mean standardized bias percentage was 7.40.

Table 8. Weighted logistic regression: stop and frisk based on reasonable suspicion for male citizens.

Measure	b	S.E.	Exp(b) (95% Confidence Interval)
Black Male Citizen	0.55 ***	0.03	1.73 (1.63 to 1.83)
Blameworthiness	2.41 ***	0.03	11.18 (10.60 to 11.80)
Frisk Reason was a Verbal Threat	2.01 ***	0.14	7.48 (5.75 to 9.77)
Frisk Reason was Prior Knowledge	2.10 ***	0.09	8.13 (6.83 to 9.67)
2 Log-likelihood = −18,449.82
Chi-Squared = 11,012.84 ***
McFadden’s R-Squared = 0.23
n = 60,714

*** p < 0.001.

shows the logistic regression results based on matching for Black male and White male citizens and the police officer conducting the stop and frisk because of reasonable suspicion. Research question four asked, “What effect does being a male citizen and the citizen’s race have on the officer’s chances to conduct a stop and frisk based on reasonable suspicion?” The results in Table 8 show that Black male citizens are 73 percent more likely to have a stop and frisk take place based on reasonable suspicion. Once again, the present results show support for focal concerns theory (i.e., research question five) as a theoretical explanation for police decision making in male citizen stops that result in a stop and frisk because of reasonable suspicion. All three components were statistically significant with male citizens who are viewed as more blameworthy (b = 2.41, S.E. = 0.03, Exp(b) = 11.18), the frisk was conducted to protect the community based on a verbal threat (b = 2.01, S.E. = 0.14, Exp(b) = 7.48), or that the officer has prior knowledge of the male citizen (i.e., practical constraints and consequences; b = 2.10; S.E. = 0.09, Exp(b) = 8.13) all increased the likelihood that a stop and frisk would occur because of reasonable suspicion.

10. Discussion

The current study sought to answer five research questions about police officer decision making in the state of Illinois regarding stop and frisks conducted by consent or reasonable suspicion. In both models, where stop and frisk occurred based on the citizens’ consent, the results showed that Black citizens (Research Question 1) and Black male citizens (Research Question 2) were less likely to provide consent. When it comes to the relationship of focal concerns theory and stop and frisks occurring based on consent, all three components (Research Question 5) increased the odds of a stop and frisk based on consent. In both models, blameworthiness had the most significant probability of impacting the likelihood of a stop and frisk because of citizen consent.

Two additional models analyzed the issue of stop and frisk occurring because of reasonable suspicion and were performed to answer research questions two, four, and five. For both models, Black citizens and Black male citizens were more likely to have a stop and frisk occur due to reasonable suspicion. Similar to the consent search models, all three components of focal concerns theory provide a theoretical explanation for officer decision making regarding stop and frisk based on reasonable suspicion. The citizen’s blameworthiness had the greatest odds of impacting the officer’s decision to conduct a stop and frisk based on reasonable suspicion.

Although the current study does show evidence of both racial and gender bias, the current study also confirms the importance of PSM. By applying PSM, the researchers can better determine how the citizen’s race influences officer decision making while using focal concerns theory as a theoretical explanation for police officer decision making.

This study provides two findings concerning racial or gender bias. The first is that it appears that Black citizens are aware of their legal rights regarding offering consent to stop and frisk. Based on the data collection, when Black citizens were asked if they consented to a stop and frisk, they indicated no to the officer. It would seem that Black citizens are aware they have the right to refuse consent and are doing so. Secondly, the two models that considered stop and frisks because of reasonable suspicion show that racial bias and racial and gender bias (i.e., Black males) are still present even after applying focal concerns theory. The results indicate that considering the components of focal concerns theory fails to remove evidence of racial bias alone and the effect of racial and gender bias considered together. The race of the citizen alone, or in conjunction with the gender of the citizen shows that these demographics still “matter” when it comes to police decision making in stop and frisk situations based on reasonable suspicion.

This study has certain limitations based on the data collected which future studies should consider. In terms of data collection, there was no information on certain demographic factors. There was no information on the citizen’s age. The demographic factors that were not collected related to the officer included age, race, ethnicity, gender, or years of service. The current study did not consider the time of day, because this could impact police officer decision making based on stops and frisks occurring in daylight versus those made in darkness. Given that the data includes the entire state, the location of the stop was not considered. There were no data showing whether stops occurred in high crime areas, high minority populated areas, or the area’s population density. In addition, the current study only examined the stop and frisks for the racial groups of Black and White pedestrians, and future studies may wish to consider all other racial/ethnic groups collected or, more specifically, the other predominant ethnic group, which was Hispanic citizens.

11. Conclusions

The research results highlight the need to balance the ability of stop and frisk as a crime control measure with Constitutional requirements to guarantee citizen’s rights against unreasonable stops and seizures by the police. Evidence of racial bias is especially troubling, even though Blacks stopped by Illinois officers seem well aware of their rights to refuse to consent to a stop and frisk-type search. Focal concerns theory taps into the components of reasonable suspicion, but race was also a factor here. Black males were 73 percent more likely to be stopped as a result of reasonable suspicion. When consent was given to search, citizen blameworthiness was a critical factor in stop and frisk events, especially when citizens made a verbal threat.

However, in their review of the NYPD studies, White and Fradella (2016, p. 29) concluded that “SQF can and should be a valuable crime-control tool for twenty-first-century police departments—if the practice is retooled and managed in ways that rely on the proper exercise of discretion”. To meet these requirements, the authors recommended that SQF procedures meet constitutional standards regarding reasonable suspicion, control possible abuses of police discretion, and be employed with an eye toward citizen concerns about procedural justice and legitimacy (White and Fradella 2016, p. 8). Their guidelines include careful selection of personnel, training in constitutional stop and frisk practices, administrative procedures to prohibit racial profiling, proper supervision to hold officers accountable, and external oversight of police stop and frisk practices (Purdon et al. 2021, pp. 471–73).