In recent years, a growing number legal scholars and social psychologists, including ourselves (Nier, Nier, Gaertner, & Dovidio, 2007), have explored the manner in which research on implicit bias may help inform the enforcement of antidiscrimination laws in the United States, most notably Title VII of the 1964 Civil Rights Act. This law, and many similar state antidiscrimination statutes, presumably forbids racial discrimination in many contexts, such as employment and housing. Despite the large and growing body of literature examining this topic, finding the appropriate linkages between psychological research and legal practice has proven to be quite challenging for a variety of different reasons. For example, the majority of antidiscrimination law is structured to combat overt bias and focuses on curbing intentional discrimination, whereas research has emphasized that some forms of discrimination can occur without conscious intent (Krieger & Fiske, 2006). Also, some have questioned whether the relevant psychological research is sufficiently well developed to inform legal practice (for a recent discussion of these debates, see Jost et al., 2009; Tetlock & Mitchell, 2009).
Ambiguity Promotes Subtle Bias
Social perception involves subjective processes; two individuals viewing the same person may perceive that person in different ways due to differing preconceptions, beliefs, and stereotypes. In the workplace, negative implicit attitudes and stereotypes may result in racially biased evaluations, as suggested by a significant body of literature (e.g., Dovidio & Gaertner, 2000; Hodson, Dovido, & Gaertner, 2002). This literature also indicates that negative attitudes and stereotypes are more likely to influence judgments when the information at hand is ambiguous or unstructured. Similarly, racial bias is more likely to occur when the standards for evaluating employees (or prospective employees) are ambiguous and subjective: the more subjective the procedures by which employees are evaluated, the greater the likelihood that subtle bias may result in biased employment decisions. Moreover, the potentially discriminatory nature of subjective employee evaluations has been acknowledged by the courts in a number of previous legal decisions (see Cook v. Billington, 1992; Grano v. Department of Dev, 1983; O'Connor v. Peru State College, 1986; Thornton v. Coffey, 1980).
We believe that SB 1070 will require law-enforcement officers to make judgments using ambiguous information as well as ambiguous standards, and as a result, the likelihood of racially biased enforcement of the statute will increase. In particular, Arizona law-enforcement officers have not yet been given clear guidance about what will constitute a “reasonable suspicion” of unlawful presence in the United States. This is not entirely surprising, because the standard itself—reasonable suspicion—is ambiguous. The instructional materials that law-enforcement officers have received, which have been issued by the Arizona Peace Officer Standards and Training (POST) Board, do indeed repeatedly caution against racial profiling, yet provide only ambiguous guidance as to what information will constitute a reasonable suspicion (Arizona Peace Office Training and Standards Board, 2010a).
Specifically, current instructional materials indicate that, under SB 1070, law-enforcement officers will be required to engage in a series of judgments when they encounter individuals. First, law-enforcement officers must make a judgment of “reasonable suspicion” of criminal activity in order to make a lawful stop, which was the case prior to the enactment of SB 1070. Following this judgment, which itself is likely vulnerable to the influence of subtle bias, a second layer of judgments must be made by law-enforcement officers; they must now also determine whether they have a reasonable suspicion that the individual is in the country illegally.
The instructional materials issued by the Arizona POST Board offer only limited guidance about how this judgment should be made (Arizona Peace Officer Standards and Training Board, 2010b). The instructional materials suggest that officers should first ask the individual for documentation (e.g., driver's license) that indicates the person is in the country legally. (This was already common practice prior to the enactment of SB 1070.) If the individual cannot produce such documentation, then any number of factors may be used to make a determination about whether the person is in the country illegally. For example, instructional materials indicate that “dress” can be considered a factor when making a determination of reasonable suspicion of being in the county illegally. Yet, the type of dress that constitutes a “reasonable suspicion” is not clearly specified, and the manner in which this factor should be considered relative to other factors, such as “demeanor” and “location,” is not clearly delineated. Instead, officers are asked to make this judgment based on the “totality of circumstances” (Arizona Peace Officer Standards and Training Board, 2010c), which is an ambiguous standard. In the absence of a more concrete standard, the judgments made by law-enforcement officials are likely to be quite subjective. This ambiguity in the specific information that should be sought and how this information should be used in making a determination of reasonable suspicion will likely result in greater reliance on implicit attitudes and stereotypes.
Detecting Subtle Discrimination Is Extraordinarily Difficult
The experimental method is the primary tool used by social psychologists to establish the presence of subtle racial discrimination, both in the laboratory and in the field (Dasgupta, 2004; Pager, 2007). Only through systematic, controlled experiments can subtle discrimination be directly substantiated.4 In the domain of employment discrimination, field studies of this type—often referred to as “audit” or “testing” studies—are conducted by having a team of trained personnel (including members of different races with matched qualifications) contact prospective employers while posing as job applicants, or by submitting similar resumes with different race-type names to prospective employers. Then, the experimenters record whether the applicant receives any interest from the prospective employers, as measured by a job offer or a “call back.”
These studies typically reveal a pattern of racial bias in the hiring process (Pager, 2007), with White applicants receiving job offers or call backs more frequently than Blacks or Latinos. Yet, when the field notes of the testers are examined, they often reveal little, if any, evidence of overt differential treatment (Pager, Western, & Bonikowski, 2009), that is, even testers who were discriminated against often report that they were treated fairly by their prospective employers. Only when the results of many testers are aggregated does a pattern of bias emerge. Thus, at the individual level, racial bias in screening and interviewing prospective employees can be hidden from the applicants. Similarly, the subtle processes by which biased decisions are made can also be hidden from the offending employers.
We suggest that a similar pattern of discrimination may occur as a result of the Arizona statute. It will be difficult for individual Latinos who are stopped by police to determine whether racial bias played a role in their encounter with law-enforcement officers. Indeed, law-enforcement officers themselves may also have difficulty in determining whether their judgments of reasonable suspicion are truly unbiased. While one possible outcome of the highly publicized debate over the new law will be heightened awareness and sensitivity about the potential for racial profiling, it will likely be difficult for individuals (both Latinos who are stopped and law-enforcement officers themselves) to accurately determine which particular stops were due to subtle racial bias and which stops were unbiased.
An Ounce of Prevention Is Worth a Pound of Cure
Thus, the findings from subtle bias research conducted in laboratory contexts, as well as the results of field experiments of discrimination, suggest that hidden racial profiling is likely to occur under the Arizona law. At the same time, research also indicates that this discrimination is typically too subtle to be easily detected in the course of everyday interactions. The subtle and hidden nature of this form of bias makes it extraordinarily difficult to recognize or litigate such discrimination, that is, many of those who are targets of subtle discrimination are unlikely to have sufficient evidence to pursue a legally actionable claim of discrimination. Consequently, subtle bias research has not had a widespread effect on the litigation of employment discrimination complaints. While there have been some successful applications of the subtle bias research in litigation (Price Waterhouse v. Hopkins, 1989), these cases still remain the exception rather than the rule.
As a result, Kang (2011) has suggested that focusing exclusively on the litigation of discrimination complaints may not be the most productive approach to minimizing discrimination. He argues that the lessons of subtle bias research are more easily applied to the prevention of discriminatory conduct in employment decisions, rather than redressing the damage already done by such behavior. Again, we see a clear parallel with the Arizona statute. As with many who are the unknowing targets of unlawful treatment in employment contexts, Latinos who are targets of racial profiling will have extraordinary difficultly establishing that their treatment was racially biased, since the likely sources of the bias (either private racism or implicit bias) are, by their very nature, easily masked. While some of these potential litigants may be successful, it may prove more useful to establish policies and procedures that discourage racial profiling before it occurs.
We suggest several preventative measures. First, law-enforcement officials should receive training that increases their awareness of the subtle nature of contemporary racial bias. Simply raising individuals’ awareness of subtle forms of racial discrimination may be effective in reducing discriminatory conduct and may be useful in limiting racial profiling. While the governor of Arizona, Jan Brewer, did issue a statement indicating that some training would be implemented (State of Arizona Office of the Governor, 2010), the resulting training materials released by the Arizona POST Board (2010b) do not squarely address the possibility of subtle bias resulting in discriminatory behavior. The materials do indeed speak strongly against making judgments based on race and state flatly that “racial profiling is police misconduct.” However, the materials also imply that racial bias is not likely to be a widespread problem. For example, at one point, the Executive Director for the Arizona Police Association states, “I urge you not to use race, not to be baited by the questions that may be posed to you by individuals whose purpose is to find Arizona Police Officers discriminatory in nature. I know you're not. You know you're not. Don't allow them to make you that” (Arizona Peace Officer Standards and Training Board, 2010c). Moreover, another law-enforcement official, the Chief of the Tucson Police Department, argues that while profiling has occurred in the past, it has been the result of “a few bad apples that have spoiled it and have had their actions racially motivated” (Arizona Peace Officer Standards and Training Board, 2010c).
These characterizations do not accurately reflect the empirical reality of contemporary discrimination. By failing to acknowledge that officers may have unintentional biases, and that they should be on guard against these biases, the materials demonstrate a lack of awareness of the subtle nature of covert and implicit racism. The materials are, in our view, inadequate because (1) unintentional biases can likely influence the judgments made by law-enforcement officers, and (2) the instructional materials do not clearly acknowledge this possibility.
Second, we propose that that training programs be rigorously studied in order to evaluate whether they actually have a meaningful impact on subtle bias and discriminatory conduct. While training programs for addressing racial profiling exist throughout the United States, usually offered through POST organizations, little is known about their efficacy. Their effectiveness is not typically subject to the scrutiny of rigorous independent review. Nonetheless, studying the impact of training is crucial to understanding what aspects of training seem to be effective and which elements are ineffective. For example, even if information about subtle bias were included in the training of police officers, it might not be sufficient to reduce discriminatory behavior toward Latinos or members of other groups. Moreover, in private industry, programs designed primarily to increase awareness of bias have only limited impact on producing equity within organizations (Kalev, Dobbins, & Kelly, 2006).
There are at least two reasons why training to reduce profiling of Latinos, stemming from SB 1070, may be limited in its effectiveness. One reason training may be limited in effectiveness is that awareness of subtle bias is not sufficient, by itself, to inhibit its operation. As noted earlier, implicit bias operates more strongly when people are under time pressure or have other cognitive demands on them, limiting their capacity to cognitively inhibit the influence of implicit bias. In addition, the effects of implicit bias are manifested subtly, in ways which people often have difficulty recognizing and, thus, controlling such bias. Therefore, we suggest that a component of training be directed at combating implicit bias directly. Specific and intensive training can undermine the association between a particular group and stereotypic characteristics (Kawakami, Dovidio, Moll, Hermsen, & Russin, 2000), reducing the activation of implicit biases that often motivate subtle discrimination (Son Hing, Chung-Yan, Hamilton, & Zanna, 2008). Plant and colleagues found, for example, that police officers (Plant & Peruche, 2005) and college students (Plant, Peruche, & Butz, 2005) displayed a systematic “shooter bias” in a computer simulation task in which they tended more often to erroneously “shoot” an unarmed Black suspect than a White suspect. However, with intensive training to inhibit this bias and to undermine stereotypic associations, this bias on the laboratory-based shooter task was significantly reduced for college students and eliminated among police officers.
Another reason why simply increasing awareness of subtle and implicit bias research may be limited in effectiveness is that it will likely have little influence on those with more blatant biases—those who knowingly harbor and endorse racial or ethnic bias. Thus, even some who receive training may nonetheless deliberately target Latinos. While this form of discrimination may be rarer than implicit bias (Wilson, 1996), there is evidence that people who aspire to be police officers tend to be higher than people in other professions in Social Dominance Orientation (Sidanius, Liu, Shaw, & Pratto, 1994), an ideological orientation that is related to overt, intentional bias against a range of minority groups (Sidanius & Pratto, 1999).
As a third preventative measure, we propose that robust mechanisms be instituted for investigating whether racial profiling is occurring in the field. Indeed, the executive director of training programs has indicated that procedures to detect discriminatory conduct will be in place (Arizona Police Officer Standards and Training Board, 2010b). This is certainly a promising sign, but the quality of discrimination “testing” programs varies widely, and given the great difficulty in detecting racial profiling in the course of everyday interactions between citizens and law-enforcement officers, this testing program must have the methodological rigor of other proven testing programs (Bendick & Nunes, 2011), and be deployed widely throughout the state in order for it to be effective. A haphazard testing program, or a more rigorous one that is not widely employed, will do little to curtail racial profiling and will not be an effective deterrent against discriminatory behavior. An effective and widely implemented monitoring system, particularly one that is independent of the police force, would also enhance the public's confidence in the fairness of the police force, which would increase their cooperation with the police and ultimately increase the effectiveness of policing (Tyler & Huo, 2002).
We propose that a truly effective testing program would be composed of two elements. First, the State of Arizona should continue to improve its analysis of traffic stop data in order to detect whether profiling appears to be occurring at the aggregate (e.g., county or state) level. In other words, after statistically controlling for other relevant variables (e.g., time and location of the stop), are Latinos more likely to be subject to a request of verification of being in the country legally, relative to Whites or other groups? Previous analyses of Arizona traffic stops have indeed found evidence of racial disparities. For example, Latinos, as well as Blacks and Native Americans, are more likely than Whites to be searched after being stopped by Arizona Department of Safety Officers, even after controlling for a number of other relevant variables (Engel, Cherkauska, Smith, Lytle, & Moore, 2009). These types of data need to be collected and analyzed on a consistent basis, since statistical analyses of traffic stop data can reveal the circumstances under which racial disparities are likely to be greatest. These results could then help inform the types of training that are appropriate and show where the need for training to minimize racial bias is greatest.
Second, Latino and White testers could be sent into the field in order to detect whether they face a racially biased pattern of traffic stops. Using procedures similar to those used in the domain of employment and housing discrimination, testers could be trained to present themselves to law enforcement officers in a similar fashion, in terms of behavior and appearance. Would Latino testers evoke “reasonable suspicion” among police officers with greater frequency than White testers, even when all other variables are tightly controlled? If so, then it could be determined that race, per se, seems to influence the likelihood that Latino-Americans are subject to a verification of their legal presence in the United States under SB 1070.
Who might administer such a monitoring system? Civilian oversight organizations represent one possibility. These organizations act as liaisons between police departments and the local community. However, they typically exist only in larger towns and municipalities (e.g., Phoenix, Tucson, and Tempe). Therefore, in order to ensure an effective statewide monitoring system, it seems that only a statewide organization would prove effective. The Arizona Office of the Attorney General, being the organization responsible for the enforcement of the state's laws, would seem to be the logical organization to oversee such a program. However, this is potentially problematic since this office is also required to rigorously enforce SB 1070. The Office of Attorney General would find itself in a position in which it must advocate for robust enforcement of the SB 1070 on the one hand, while being vigilant for evidence of racial profiling on the other hand. Given these competing concerns, it is not clear that the Office of Attorney General would be sufficiently independent of law enforcement to provide effective independent oversight of the enforcement of SB 1070.
A system that monitors the discrimination that may occur as SB 1070 is implemented could also help officers engage in corrective action as they perform their ongoing duties. In other words, the results of discrimination “testing” programs could be integrated with training programs, in order to increase the effectiveness of training. For example, people are generally less likely to detect bias in their own behavior (and the behavior of others) when decisions are made sequentially. That is, bias often remains hidden when information about cases is presented one at a time, rather than on the basis of aggregated information that a monitoring system (e.g., a review board) could provide (Crosby, Clayton, Alksnis, & Hemker, 1986). Moreover, when people with egalitarian values become aware of systematic deviations from their personal standards, they become motivated to engage in corrective action (Monteith & Flynn, 2010). With practice, they could better recognize and inhibit subtle bias in their behavior. In addition, a review procedure that presumes the possible operation of bias in ambiguous situations could help bring the issue to consciousness and provide a structure for addressing ambiguity. The opportunity to deliberate and reflect on decision-making processes increases the extent to which egalitarian social norms and personal values guide these processes, and thereby reduces the influence of implicit biases (Crosby & Dovidio, 2008).