Barriers to the Ballot Box: Implicit Bias and Voting Rights in the 21st Century

Arusha Gordon,Ezra D. Rosenberg October, 2015

While much has been written regarding unconscious or “implicit bias” in other areas of law, there is a scarcity of scholarship examining how implicit bias impacts voting rights and how advocates can move courts to recognize evidence of implicit bias within the context of a voting rights claim. This Article aims to address that scarcity. After reviewing research on implicit bias, this Article examines how implicit bias might impact different stages of the electoral process. It then argues that “results test” claims under Section 2 of the Voting Rights Act (VRA) present an opportunity for plaintiffs to introduce evidence regarding implicit bias in the electoral process. In addition, this Article explores policy solutions to reduce the impact of implicit bias in elections.

Introduction

This year marks the 50th anniversary of the passage of the Voting Rights Act (VRA),¹ a law that changed the voting rights landscape dramatically. Today the Jim Crow era of literacy tests and total denial of access to the ballot box for minorities is gone. Though participation of minority voters in the electoral process has increased, several states have recently implemented policies that threaten these gains.² For example, despite scarce evidence of in-person voter fraud, states continue to pass laws requiring voters to present forms of photo identification in order to vote, which some voters do not have.³ Many of these laws have a disproportionate impact on minorities, as minorities are less likely than Whites to have acceptable identification.⁴

What makes the protection of voting rights against racial discrimination particularly challenging is that, while Americans today might not intend to discriminate and are less likely than were previous generations to express racial motivations for an action, unconscious biases continue to influence individual and organizational decisions.⁵ This Article explores unconscious, or “implicit,” bias in the electoral process and examines how research on implicit bias might fit within the current legal framework of voting rights.

The next section of this Article provides a general overview of implicit bias. Part II examines the electoral process and highlights how implicit bias might impact each stage, from the allocation of resources for voting to decisions made by poll workers on Election Day regarding who may vote a regular ballot. Part III discusses the legal framework for voting rights advocacy today. Finally, Part IV argues that the “results test” in Section 2 of the VRA provides an opportunity for courts to consider evidence of implicit bias within the electoral context. In addition, Part IV provides examples of how evidence regarding implicit bias can be incorporated into a “results test” claim and discusses policy reforms to reduce the impact of implicit bias in the voting arena.

I. Implicit Bias Defined

Implicit bias describes the phenomenon by which decisions are impacted by unconscious prejudices held in the brain. Psychological research on implicit bias has gained significance in recent decades,⁶ following the 1998 introduction of the implicit association test (IAT).⁷ The IAT measures “the strengths of associations among concepts.”⁸ The IAT operates by showing participants a set of images and words and asking them to classify the images and words into groups as quickly as possible.⁹ The speed with which a participant is able to classify these words or images is tracked and used to measure how strongly one implicitly associates those words or images with the general categories (i.e. how strongly one associates African-Americans with negative attributes and European Americans with positive attributes).¹⁰ “The IAT operates on the supposition that when the two concepts are highly associated, the sorting task will be easier and thus require less time than when the concepts are not highly associated.”¹¹ Of the more than two million users who had visited implicit.harvard.edu to take the IAT, sixty- eight percent of them had some level of implicit bias based on race.¹² Furthermore, the time differentials in the IAT sorting tasks “have been found to be statistically significant,” i.e., likely not “due to random chance.”¹³

As the IAT shows, our brains are wired to take in information and form social “schemas,” mental theories about how we expect people and social situations to operate.¹⁴ Schemas are critical to our survival: our brains simply encounter too much information on a day-to-day basis to process it all. Our brains need short cuts to help process the onslaught of information efficiently and to anticipate and prepare us for contingencies.¹⁵ For instance, as children we learn that a snarling dog is a threat. We might learn this through experience (getting bitten by an angry dog) or through other informational sources (an adult telling us to be careful of dogs or a cartoon showing a character getting chased by a dog). Once we learn this, if we see a snarling dog, we automatically rely on a schema of “snarling dog as dangerous” and retreat. Therefore, there is no need to relearn that a snarling dog is dangerous every time we encounter one. Our ability to rely on shortcuts, such as “snarling dog is dangerous,” is critical to our survival (and our ability to avoid dog bites!).

Yet, while some schemas are necessary for our survival, other schemas may be based on inaccurate or incomplete information. For instance, the media and societal narratives have trained most people’s brains to react to individuals of other races in differing ways.¹⁶ While we might consciously reject racism, impressions learned when young and decades of watching the news, TV shows, or otherwise simply engaging in society engrain stereotypes in our brains and contribute to the development of schemas. The media help create stereotypes,¹⁷ for instance, by frequently portraying Whites as hardworking professionals, while often portraying Blacks as athletes and criminals, Hispanics as illegal immigrants and drug dealers, and Asians as scientific geniuses and classical musicians. Overriding these schemas and stereotypes is possible, but given the pervasiveness of the messages reinforcing racial stereotypes, overriding schemas and stereotypes about a social group requires consistent effort.

In addition to a number of other factors, research shows that implicit bias is most likely to taint our actions when we have more discretion or must make a quick decision. These are situations in which we may unwittingly rely on schematic shortcuts because we do not have guidelines by which to make a decision or time to process new information fully.¹⁸ The following sections take a closer look at how these factors (discretion, time pressures, etc.)—and implicit bias in general—can impact the electoral process.

II. Implicit Bias in the Electoral Process

Superficially, the administration of elections and the voting process seem strictly regulated with little room for election officials to exercise discretion. A closer look, however, reveals multiple points at which election officials make choices,¹⁹ from where to place new voting machines to how to assess a voter’s identification to how much assistance to offer a disabled, elderly, or minority voter. With each use of discretion, the risk of implicit bias infecting the electoral process increases.²⁰ This section analyzes how implicit bias might influence decisions made at various stages of the election process.²¹

A. Resource Allocation

Although social psychologists have not directly tested decision-making in the allocation of election resources, experiments examining how people make decisions when allocating resources in other areas are instructive. For instance, in a study by Laurie Rudman and Richard Ashmore, researchers asked participants, university students in an introductory psychology class, to make recommendations regarding how to implement a twenty percent budget cut to student groups on campus.²² The researchers also asked participants to complete a survey measuring their explicit attitudes about people from different racial groups.²³ In addition, the researchers gave the participants an IAT test measuring their implicit attitudes regarding Jews, Asians, and Blacks.²⁴ The researchers found that “implicit biases predicted economic discrimination toward Jews, Asians, and Blacks, and that the stereotype IAT was either an equal or superior predictor, compared with explicit attitudes.”²⁵ In other words, participants demonstrating higher implicit bias against Jews, Asians, and Blacks on the IAT were more likely to cut the budget of student groups associated with those social identities than their expressed attitudes would have indicated.²⁶

Researchers in a different study tested White participants’ willingness to place a chemical plant in predominantly minority or White neighborhoods.²⁷ The researchers surveyed half of the White participants regarding their receptivity to placing a chemical plant in a majority White neighborhood and half of the White participants regarding their receptivity to placing the chemical plant in a majority Black neighborhood.²⁸ The researchers also provided half of the participants with “local and national housing cost information,” and surveyed the participants’ perceptions of the neighborhoods, such as the degree to which “power plants and incinerators already existe[d] nearby,” the “perceived neighborhood house values,” and the “perceived socioeconomic status of residents.”²⁹ The researchers found that, “participants were less likely to oppose the plant when the neighborhood was black than when it was white – regardless of housing cost information – even when controlling for participants’ perceptions of house values and socioeconomic status, their concern for the environment, and their explicit feelings toward blacks.”³⁰ Participants who believed they treated people from all races equally were still more likely to be comfortable placing a chemical plant in a minority neighborhood rather than in a White neighborhood.³¹

These experiments illustrate how implicit biases can impact decisions regarding resource allocation and can implicate how election resources are distributed. County officials rely on their discretion when deciding how to disburse election resources in a range of situations, from determining where to create new polling locations to deciding which localities receive new voter machines. For example, up to 15,000 people left the polls without voting on Election Day 2004 in Columbus, Ohio,³² in large part due to decisions by officials about who was likely to vote. Franklin County officials allocated the machines “according to instinct and science.”³³ Officials “apparently . . . assumed that in a poor neighborhood, turnout would be low.”³⁴ Their failure to provide additional resources to certain neighborhoods led to long lines, from which many discouraged voters turned away before being able to cast their ballots.³⁵

Implicit bias may also impact which polling locations get newer voting machines. A 2003 lawsuit³⁶ brought by the Southwest Voter Registration Education Project and other groups noted that certain voting devices—which were prone to error at twice the rate of newer voting devices—were still in use in six largely minority counties in California. While eighty-one percent of African-American and sixty-seven percent Latino voters live in counties with “obsolete” machines in California, only fifty-nine percent of White voters live in counties with outdated machines.³⁷ Although intentional discrimination may still be a factor behind why polling places that serve majority-minority populations end up with older machines, when other, whiter counties receive newer voting machines, experiments regarding how implicit bias can impact resource allocation decisions indicate that unconscious biases could also be playing a role.³⁸

B. Pre-Election Day: Voter Registration and Voter Qualifications Election officials may also exercise discretion during voter registration and other steps in the election process requiring individuals to complete forms. Officials processing voter registration applications, absentee ballot requests, and other election related documents must ensure the forms are accurately completed.³⁹ Although this would appear to entail minimal discretion, these officials must make a number of decisions. For instance, they must ensure that the name and address on each form are legible, so the information can be added to the registrar’s database of voters.⁴⁰ Similarly, when considering an absentee ballot, the registrar or other staff member must ensure that the information sufficiently matches the data already on file for the voter and, if it does not, the election official may need to determine whether the mistake is cause to invalidate the ballot.⁴¹ Is listing a “street” instead of an “avenue” in one’s address a “material” difference and sufficient grounds upon which to reject the application? Is forgetting to include the suffix to one’s name—Junior or Senior—a material omission?⁴² Although many states provide guidance on some of these questions, officials processing applications must often rely on their own discretion when making more marginal decisions.⁴³ Research into implicit bias indicates that the risk is high that these seemingly ministerial decisions may be infected by unconscious biases.

Studies show that how an individual reviews a document can be influenced by unconscious biases, even when he or she is not directly interacting with the person who completed the document or form. A 2014 study found that, when grading the same memorandum, law firm partners gave lower grades when the author was a hypothetical Black law student than when the author was a hypothetical White law student.⁴⁴ The study found that law firm partners identified more “objective” mistakes (i.e. spelling and grammar mistakes) in the memorandum written by the fictional African-American summer associate and also had more “subjective” critiques (i.e. regarding the overall analysis) than in the memorandum written by the fictional White associate.⁴⁵

Similarly, election officials may unintentionally treat potential voters’ registration applications and requests for absentee ballots differently based on the perceived race of the voter. A registrar may identify more “objective” mistakes and give more weight to them when reviewing an electionrelated document submitted by a minority, rather than when reviewing one submitted by a White person, as was the case with law firm partners reviewing the hypothetical summer associate’s memorandum. There is evidence that this has, in fact, occurred. Testimony from Windy Boy v. Big Horn County ,⁴⁶ described how county officials subjected “Native Americans to a more technical and more difficult voter registration process than whites”⁴⁷ and that county officials “became hypertechnical . . . and looked for minor errors in [Native American] registration applications and used them as an excuse to refuse to allow registration.”⁴⁸

Likewise, a 2012 report on treatment of Native American voters in Arizona found that staff in recorders’ offices in Arizona routinely placed Native voters on “suspense lists” (similar to inactive lists)⁴⁹ when the recorder was not satisfied that an applicant sufficiently clarified his or her address.⁵⁰ There are few guidelines on what should constitute an adequate address in Arizona;⁵¹ instead, it is left to the recorder’s discretion and may be influenced by implicit bias.⁵²

Implicit biases might also influence the degree of assistance that election officials, elected representatives, and political campaigns offer in response to voters’ questions in the weeks leading up to an election. A study from the University of Southern California tested the responses of state legislators to inquiries about voter identification laws.⁵³ The researchers sent emails to 1,871 state legislators in fourteen states in the two weeks before the 2012 elections.⁵⁴ Their emails read:

Hello (Representative/Senator NAME), My name is (voter NAME) and I have heard a lot in the news lately about identification being required at the polls. I do not have a driver’s license. Can I still vote in November? Thank you for your help. Sincerely, (voter NAME)⁵⁵

The experimenters changed the names in the emails they sent, so that one group of legislators received the email from a “Jacob Smith” while another group received the email from a “Santiago Rodriguez.”⁵⁶ Because no state actually required a driver’s license to vote, “legislators really could have simply responded with a ‘yes’” and explained that drivers’ licenses were not required to vote.⁵⁷ Instead, “[t]he researchers found that legislators who had supported voter ID laws were much more likely to respond to ‘Jacob Smith’ than to ‘Santiago Rodriguez.’ ”⁵⁸ As the researchers note, this provides strong evidence that voter identification laws were enacted with consciously held discriminatory intent.⁵⁹ But, the study also noted that even those legislators who did not back photo identification laws responded less frequently to correspondents with Hispanic names, although this difference in response rates was less significant.⁶⁰ This difference in response rates may be the result of an unconscious association of minority voters with “unlikely” or “ineligible” voters. Another study found that both elected and appointed election officials—the latter presumably having less direct political incentive to dissuade certain populations of voters from casting their ballot—responded less frequently to emails asking about voter identification requirements sent from a fictional Latino person rather than a fictional White person.⁶¹

These studies bear similarities with research in the education and employment settings showing a difference in responses based on nothing more than whether a name sounds “ethnic.” One study found that professors replied more often to requests by students asking to meet when the requests were from students with generically “White-sounding” names than names associated with minorities.⁶² Another study found that recruiters called job applicants with “White names” fifty percent more often than those with typical African-American names, even when their resumes were identical.⁶³ Given that law firm partners,⁶⁴ university professors,⁶⁵ and employment recruiters⁶⁶ have demonstrated bias based on the perceived ethnicity of an individual’s name, there is reason to question whether election officials are free from the influence of similar implicit biases.

C. Election Day Voting

Implicit biases may also affect Election Day decisions made by poll workers and election officials. For example, a survey of the 2008 elections found that in states where poll workers are not required or allowed to request identification, the race of both the poll worker and the voter affects the rate at which voters from different racial groups are asked for identification, with Black and Hispanic voters being asked to “show ‘picture ID’ more often than Whites.”⁶⁷

Psychological research concerning how we perceive and recognize individuals from an in-group versus an out-group raises important questions about how implicit biases might influence the election process in states with photo identification requirements.⁶⁸ The “other race effect” (ORE) or “cross race effect” (CRE) describes the phenomenon in which a member of one ethnic group is better able to recognize and “individuate”⁶⁹ between members of his or her own ethnic group than between members of other ethnic groups.⁷⁰ Researchers have proposed a number of theories to explain this phenomenon. Under the categorization-individuation and perceptual expertise models, an individual who grows up with consistent and sustained exposure to his or her own race is trained to individuate faces and remember names of people with the same racial background.⁷¹ These theories are supported by research showing that individuals “who grew up in diverse neighborhoods” are better able to individuate members of other races.⁷² Social-cognitive theories, on the other hand, hold that when an individual first encounters another person he or she automatically categorizes that person “as belonging to either our own social in-group or a social out-group.”⁷³ Because members of an ingroup are more “socially important,” our brains recognize and store information regarding in-group members’ individuating characteristics, while our memories struggle to store this information for out-group members.⁷⁴ Additional theories explaining the CRE have also been developed.⁷⁵

To explore the impact of the cross race effect and implicit bias on Election Day, it is a useful to hypothesize what might happen when “Jane,” a poll worker in a state with a new photo identification law, checks in her first voter. Jane is a White retiree, from a largely White, middle class neighborhood. Jane grew up attending largely White schools, and most of her friends are White. Jane would never call herself a racist and says the color of someone’s skin does not matter to her. Jane’s first voter is a Black man named Jamal Jones. What does psychological research on the subconscious suggest will happen when Jane attempts to check Jamal in to vote?

First, Jane will likely experience an increased rate of neurological activity, and her pupils will dilate.⁷⁶ In a study mapping the amygdala (the area of the brain that processes emotional information, such as fear, and helps make snap judgments), researchers found that when participants in a study were shown Black faces, but not White faces, they experienced an increase in amygdala activity.⁷⁷ The researchers concluded that, subconsciously at least, participants’ brains processed Black faces as possible threats.⁷⁸ When Jamal approaches Jane at the poll workers’ table, Jane’s amygdala may activate, sending fear signals and causing her eyes to dilate.

In addition to experiencing heightened neurological activity, the cross race effect may create a situation in which Jane struggles to match Jamal to his picture. Jane wants to do her job properly, but she has always had difficulties with recognizing “Africans” because they “all look alike.” Jane would not be alone in experiencing difficulty recognizing and individuating between people of another race, as studies indicate that, when asked to distinguish individuals, participants are more likely to study faces of other races longer than faces belonging to their own race.⁷⁹

While Jane may ordinarily be able to override the cross race effect, circumstances on Election Day are ripe to exacerbate unintended reliance on implicit biases. Psychologists have established that people are more likely to rely on stereotypes and be influenced by unconscious assumptions when under stress or a tight deadline.⁸⁰ This makes sense because stereotypes are generalizations, and when one does not have time to investigate a situation fully before making a decision—whether that be a decision to hire someone or to give someone a provisional or regular ballot—our brains may automatically rely on generalizations or stereotypes.⁸¹ As discussed below, undue reliance on stereotypes and implicit biases may mean that Jamal and other minority voters are more likely to be handed provisional ballots.

D. Assisting Voters and Provisional Ballots

Implicit biases may also influence whether poll workers decide to assist certain individuals on Election Day. That poll workers’ decisions may be impacted by implicit biases on Election Day is particularly disconcerting for two reasons. First, the amount of effort a poll worker invests in reconciling an issue for a voter often determines whether that voter casts a regular or a provisional ballot.⁸² Second, Election Day is a voter’s last chance to cast his or her ballot. If implicit biases contribute to a situation in which a voter is denied a regular ballot, the harm is irreparable.

Information about the assistance offered to White voters as compared to minority voters is not readily available. We do not know if poll workers spend more time trying to reconcile differences between a name on an ID and a name in a poll book for some voters than for others. Nevertheless, research from other areas regarding assistance is instructive. A study by the U.S. Department of Housing and Urban Development in 2000 found that “approximately one in five Black or Hispanic home seekers did not receive assistance and information that were provided to equally qualified whites,” even though minority and White home seekers presented equal qualifications.⁸³ Minority home seekers were less likely to be offered the “chance to inspect units,” were more likely to be denied “information about available units,” and were more likely to be “steer[ed] towards [housing] in minority neighborhoods” than were Whites.⁸⁴ Minority home seekers were also more likely to be offered “inferior financial terms and inferior assistance and follow-up with housing transactions.”⁸⁵

Similarly, other studies examined the help given to a shopper who drops a bag of groceries.⁸⁶ In some cases, the “shopper” was White; in other cases, the “shopper” was Black.⁸⁷ In one study, researchers found that, while “whites and blacks were offered assistance in equal numbers, the amount of assistance offered was not equal.”⁸⁸ The study revealed that although “70% of the time white subjects helped black women, they gave only perfunctory help, picking up only a few packages.” However, “63% of the time that white subjects were aiding white women, the subjects gave complete help, picking up all of the groceries.”⁸⁹

Studies demonstrating that minorities frequently receive less assistance than Whites suggest that poll workers might unintentionally offer less assistance to minority voters than to White voters. A voter who attempts to cast a ballot on Election Day might run into any number of problems that could be resolved with help from a cooperative poll worker. If a voter lacks the proper form of identification, a poll worker could spend time explaining what is acceptable. If a voter’s name is not found on the voting roll, a poll worker could call other voting locations to determine if the voter’s name appears on the roll at a different polling location. If a voter needs additional assistance because he or she is not a native English speaker or is disabled, a poll worker’s actions might contribute to how comfortable the voter may feel when asking for help. If a poll worker unintentionally provides less assistance to traditionally marginalized voters as compared to others because of the poll worker’s implicit biases, there will be a disproportionate impact on minorities’ votes being counted.

There is evidence that this is the case, as a 2004 survey found that Latino jurisdictions had the highest rates of provisional ballot use.⁹⁰ Provisional ballots are frequently used when there is an issue regarding the voter’s eligibility and might not be counted because the voter must often take additional steps (i.e. go to a different polling location or submit additional identification).⁹¹ It is impossible to know whether the decisions made in the examples above might be influenced by implicit biases or by the conscious intent to discriminate. The next sections discuss the legal framework established to protect voters and posit that the “totality of the circumstances” test established by Section 2 of the Voting Rights Act provides an opportunity for courts to consider evidence of unconscious bias in a “results” claim brought under that provision.⁹²

III. Legal Frameworkof Voting Rights

In 1965, Congress responded to the systemic disenfranchisement of racial minorities by passing the Voting Rights Act (VRA).⁹³ The Voting Rights Act immediately changed the electoral landscape of the country. In the years after the VRA was implemented, voter registration of African- Americans increased dramatically. In Mississippi, for example, the number of African-Americans registered to vote rose from just 6.7 percent to sixty percent over the next three years.⁹⁴

The Act included a number of provisions that worked together to prevent discriminatory voting laws and practices, most important of which were Sections 2 and 5. Section 5 required certain jurisdictions to obtain preclearance from either the United States Attorney General or from a three-judge panel in the United States District Court for the District of Columbia before implementing any change in voting practices or procedures.⁹⁵ In 2013, in Shelby County v. Holder ,⁹⁶ the Supreme Court effectively gutted Section 5,⁹⁷ leaving Section 2 as the primary statutory means of challenging discriminatory voting practices.⁹⁸

Section 2, as amended in 1982, states:

(a) No voting . . . procedure shall be imposed . . . in a manner

which results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color . . . .

(b) A violation . . . is established, if based on the totality of

circumstances, it is shown that the political processes leading to nomination or election . . . are not equally open to participation by [minorities] in that [minorities] have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice.⁹⁹

The “results” phrase and the “totality of the circumstances” standard were expressly added by Congress in 1982, in response to the Supreme Court’s decision in City of Mobile, Alabama v. Bolden requiring plaintiffs to show intentional discrimination in order to prove a voting rights violation under Section 2.¹⁰⁰ As a consequence of this amendment, there is no requirement that plaintiffs prove intentional discrimination to establish a Section 2 results violation (known as a “results test” claim).¹⁰¹

Courts have recognized two types of results test claims under Section 2: (1) vote access claims (otherwise known as “vote denial” or “ballot access” claims), which concern prerequisites to voting and practices which may limit a voter’s ability to access a ballot, cast that ballot, and have that ballot fairly counted, and (2) vote dilution claims, which challenge practices weakening minority voting strength.¹⁰² Although a plaintiff bringing a vote dilution results test claim must establish certain preconditions not applicable to a vote access claim,¹⁰³ both vote dilution and vote access claims require a showing of more than just the “disproportionate racial impact” of a practice.¹⁰⁴ This is where the “totality of the circumstances” standard comes into play. The “totality of the circumstances” analysis required by the Section 2 results standard “depends upon a searching practical evaluation of the ‘past and present reality,’ and on a ‘functional’ view of the political process.”¹⁰⁵ In conducting this analysis, a court “must assess the impact of the contested structure of or practice on minority electoral opportunities ‘on the basis of objective factors.’ ”¹⁰⁶

In connection with the amendments to the VRA in 1982, the Senate Committee on the Judiciary issued a report suggesting factors for courts to consider when assessing the “totality of the circumstances.”¹⁰⁷ The Senate factors include:

1. the “history of official voting-related discrimination in the state or political subdivision”;

2. “the extent to which voting in the elections of the state or political subdivision is racially polarized”;

3. the extent to which the “state or political subdivision” has

used “voting practices or procedures” that tend to “enhance the opportunity for discrimination against the minority group,” such as “unusually large election districts, majority-vote requirements,” and prohibitions against bullet voting;

4. the exclusion of members of the minority group from candidate slating processes; 5. the extent to which minority group members bear the “effects of discrimination in areas such as education, employment, and health, which hinder their ability to participate

effectively in the political process”;

6. the use of “overt or subtle racial appeals” in political campaigns; and 7. “the extent to which members of the minority group have

been elected to public office in the jurisdiction.”¹⁰⁸

The Senate Judiciary Committee made clear that the factors listed are not exclusive and that courts could consider additional factors in assessing Section 2 claims.¹⁰⁹ In Thornburg v. Gingles , the Court applied the Senate factors within the “totality of circumstances” and found that the overarching issue is whether the challenged practice “interacts with social and historical conditions to cause an inequality in the opportunities enjoyed by Black and White voters.”¹¹⁰ As discussed in the next section, the totality of the circumstances standard provides an opportunity for courts to consider social science research regarding implicit bias as part of a results test claim.

IV. Prescriptions for the Future

While much of the research discussed thus far illustrates the deeply engrained and subconscious nature of bias, the role of implicit bias in the election process can be reduced through litigation strategies and policy changes. This section first provides an overview of case law discussing implicit bias. Next, it discusses possible ways that Section 2 “results” plaintiffs may attempt to introduce implicit bias research into evidence as part of the “totality of the circumstances.” Finally, this section considers several policy changes that can be made at the state, county, and municipal levels to reduce the impact of implicit biases on the election process.

A. The Legal Framework for Implicit Bias

1. Overview of Case Law Discussing Implicit Bias and Challenges to Admitting Testimony on Implicit Bias

As previous literature has noted, courts have considered the potential implications of implicit bias in a wide range of cases, from warrantless searches to jury selection to employment matters.¹¹¹ For example, in the context of discussing the reasonableness of a warrantless stop of a person based on his “Hispanic appearance,” the Ninth Circuit recognized that “racial stereotypes often infect our decision making processes only subconsciously.”¹¹²

Courts have also discussed the impact of implicit bias in jury selection. In concurring with the Supreme Court’s decision in Miller-El v. Dretke , a case involving a prosecutor’s discretion in striking jurors, Justice Breyer cited social science articles discussing implicit bias and noted that “ ‘[s]ubtle forms of bias are automatic, unconscious, and unintentional’ and ‘escape notice, even the notice of those enacting the bias.’ ”¹¹³ In addition, the District Court for the Northern District of California discussed the effects of implicit bias on the jury process in Chin v. Runnels , finding that “a growing body of social science recognizes the pervasiveness of unconscious racial and ethnic stereotyping and group bias.”¹¹⁴ Ultimately, the court denied Chin’s petition, noting that “the problem of unconscious bias has not yet been directly addressed by the [Supreme Court].”¹¹⁵ Other courts have also discussed implicit bias in the jury selection process and jury deliberations.¹¹⁶

Courts have explored the implications of implicit bias in the employment arena more than in any other area of law and have relied on research on implicit bias and stereotyping to explain decisions made concerning hiring and firing, promotions, and pay. In EEOC v. Inland Marine , for example, the Ninth Circuit found that even when an employer’s practices were “harmless in appearance” and manifested discrimination “subtly,” rather than “through a ‘scheme or plan,’ ” they could still “hide subconscious attitudes, and perpetuate the effects of past discriminatory practices.”¹¹⁷ In Kimble v. Wisconsin Department of Workplace Development , the Eastern District of Wisconsin found that the defendant supervisor “behaved in a manner suggesting the presence of implicit bias.”¹¹⁸ In finding that implicit bias tainted the defendant’s actions, the court pointed to the defendant’s reluctance to interact with the plaintiff-employee, the speed with which the defendant blamed the plaintiff for mistakes but not other employees, and the defendant’s grudging acknowledgement of plaintiff’s achievements when the defendant “readily praised like accomplishments” of White employees, among other things.¹¹⁹ Although the legal standards governing employment cases require proof that discrimination was a motivating factor,¹²⁰ unlike those governing Section 2 “results” claims, these cases lend support to the proposition that evidence of subconscious bias should be admissible in discrimination claims generally.

In admitting expert testimony on implicit bias, in both voting rights and other civil rights cases, a major issue is whether experts will be required to connect their opinions to the specific facts of the case or whether they will be permitted to testify generally as to the phenomenon of implicit bias. In Jones v. National Council of Young Men’s Christian Association , the Northern District of Illinois blocked the introduction of expert testimony by Dr. Anthony Greenwald, one of the creators of the IAT, finding that even testimony just on the “general principles” of implicit bias must be “adequately tied to the facts” of a specific case in order to be admissible.¹²¹ The Supreme Court similarly rejected expert testimony regarding how stereotypes (but not “implicit bias” per se) might influence managers’ decisions in the Wal-Mart v. Dukes class action.¹²² The Court found that, because the plaintiffs’ expert could not “determine with any specificity how regularly stereotypes play a meaningful role in employment decisions at Wal–Mart”—whether that be “0.5 percent or 95 percent of the employment decisions”—his testimony could be “safely disregarded.”¹²³

The issue of whether such testimony is admissible will largely be governed by the Federal Rules of Evidence on relevancy (Rules 401 and 402) and the Federal Rule of Evidence regarding “helpfulness” or reliability (Rule 702). The Advisory Committee Notes to the 2000 Amendments to the Federal Rules of Evidence provide support for the admissibility of the general principles of the phenomenon of implicit bias:

[I]t might . . . be important in some cases for an expert to educate the factfinder about general principles, without ever attempting to apply these principles to the specific facts of the case. For example, experts might instruct the factfinder on the principles of thermodynamics, or bloodclotting, or on how financial markets respond to corporate reports, without ever knowing about or trying to tie their testimony into the facts of the case. The amendment does not alter the venerable practice of using expert testimony to educate the factfinder on general principles. For this kind of generalized testimony, Rule 702 simply requires that: (1) the expert be qualified; (2) the testimony address a subject matter on which the factfinder can be assisted by an expert; (3) the testimony be reliable; and (4) the testimony “fit” the facts of the case.¹²⁴ In Samaha v. Washington State Department of Transportation , the Eastern

District of Washington relied on the Advisory Committee Notes in admitting expert testimony by the afore-mentioned Dr. Greenwald.¹²⁵ The court found that Dr. Greenwald’s opinions were admissible as they were “based on reliable methodologies and consist of relevant subject matter” and “likely to provide the jury with information that it will be able to use to draw its own conclusions.”¹²⁶ Relying on the Supreme Court’s decision in Price Waterhouse , the court found that “[t]estimony that educates a jury on the concepts of implicit bias and stereotypes is relevant to the issue of whether an employer intentionally discriminated against an employee.”¹²⁷ Although Section 2 cases do not require proof of intent to discriminate, the decision to admit Dr. Greenwald’s opinions of “general principles” supports the admissibility of similar evidence in Section 2 cases.

Given that courts might consider the admissibility of expert testimony on the IAT and implicit bias as a close question, plaintiffs wishing to introduce evidence regarding unconscious bias in Section 2 cases must be careful to satisfy the “fit” test and show a valid connection, or “good grounds to extrapolate,” from psychological experiments to the situation at hand.¹²⁸ In addition, plaintiffs should be careful to present evidence regarding implicit bias as framed within the totality of the circumstances, as is discussed in more depth in the following section.

2. The Results Test and Implicit Bias

The “totality of the circumstances” test and application of the Senate factors present an opportunity for plaintiffs in a Section 2 results case to introduce evidence of implicit bias in the voting process. As Justice Scalia has noted, the “‘results’ criterion provides a powerful, albeit sometimes blunt, weapon with which to attack even the most subtle forms of discrimination.”¹²⁹

As discussed above, the ultimate inquiry is whether the challenged practice “interacts with social and historical conditions to cause an inequality in the opportunities enjoyed” by minority voters.¹³⁰ In this context, explanation of the principles of implicit bias may shed considerable light on the statistical analyses of disproportionate impact, demonstrating that the disproportionality was “on account of race,”¹³¹ particularly where subconscious bias might affect discretionary decisions like those dealing with the allocation of electoral or voting resources (i.e. voting machines, voter registration forms, electronic poll books, etc.) and the identities of voters.

This was precisely how the court treated evidence in what appears to be the only Section 2 case that admitted testimony of implicit bias.¹³² In Farrakhan v. Gregoire ,¹³³ prisoners challenged Washington’s practice of disenfranchising individuals convicted of felonies. Although, ultimately, Farrakhan was decided against the plaintiffs on a purely legal issue (i.e., that felony disenfranchisement laws may only be challenged under Section 2 of the VRA by meeting the high standard of showing “that the criminal justice system is infected by intentional discrimination” or that the legislature enacted the “felon disenfranchisement law . . . with such intent”),¹³⁴ its handling of expert testimony on implicit bias is instructive nevertheless. Specifically, the trial court considered the testimony of an expert to the effect that the racial disproportionality in Washington State’s criminal justice system could not be explained by higher levels of criminal involvement and, therefore, could not be explained other than by race.¹³⁵ This evidence—described by the trial court as consisting of reports regarding both conscious and subconscious bias—was tied specifically to the statistical evidence in the case and did not consist only of “general principles” of implicit bias.¹³⁶

It would appear that, at a minimum, evidence of studies into implicit bias, when tied to record evidence of disproportional impact or racially polarized voting,¹³⁷ should be readily admitted into evidence, as it would be highly relevant to any number of the Senate factors. For example, when considering the first Senate factor, “the history of official voting-related discrimination,” plaintiffs might connect evidence of implicit bias research to specific decisions concerning the allocation of election resources. Courts considering cases like the 2003 California lawsuit challenging the fact that minority communities were more likely to be provided “obsolete” voting machines than other communities¹³⁸ could benefit from testimony regarding the impact of implicit biases on resource allocation decisions.

As with the other Senate factors, factor two—which deals with the existence of racial polarization in voting or the extent to which voters vote along racial lines—is not required for a successful results test vote access claim.¹³⁹ Senate factor two, however, is a prime example of in-group bias. “In-group bias designates favoritism toward groups to which one belongs.”¹⁴⁰ In the voting context, this contributes to racially polarized elections in which Whites vote for White candidates and minorities vote for candidates who share their race.¹⁴¹ Courts may consider evidence regarding implicit bias and in-group favoritism to provide context for their assessment of the second Senate factor and their consideration of whether and why a jurisdiction is racially polarized within the totality of the circumstances.

Senate factor five asks “the extent to which members of the minority group . . . bear the effects of discrimination in such areas as education, employment and health, which hinder their ability to participate effectively in the political process.”¹⁴² Advocates considering Senate factor five can draw on research showing how unconscious bias impacts minorities in terms of their education, employment, and health to establish how the effects of both explicit and unconscious discrimination “hinder” the “ability [of minorities] to participate in the political process.”¹⁴³ For example, studies showing how implicit bias contributes to patterns in which students of color are disproportionately punished, graded more harshly, and ultimately less likely to go on to higher education than White students¹⁴⁴ can be tied to other studies showing that those without a college degree are less likely to vote regularly.¹⁴⁵ Demonstrating such a link between implicit bias in education and other arenas (employment, health, etc.), and the “ability [of minorities] to participate effectively in the political process,” is one way advocates can frame a discussion of Senate factor five.¹⁴⁶

Depending on the racial appeals made in a particular campaign, evidence regarding subconscious “priming” might be especially relevant to plaintiffs considering Senate factor six, the use of “overt or subtle racial appeals” in political campaigns.¹⁴⁷ Research shows that priming—which “takes place when an idea, image, or association is activated or made salient”¹⁴⁸—can have a significant impact on one’s decisions and actions. For instance, researchers have found that men primed with a sexist television commercial later judged female job applicants more harshly.¹⁴⁹ Similarly, Whites primed by watching stereotypic portrayals of African-Americans on television were “more likely to [later] judge a Black defendant [as] guilty of an assault.”¹⁵⁰

In the context of an election, subtle priming could have notable results. Take, as an example, the Willie Horton advertisement run in 1988 by the George H.W. Bush presidential campaign against Democratic opponent Governor Michael Dukakis of Massachusetts.¹⁵¹ The advertisement featured intimidating photos of Horton, who committed robbery and rape while on a furlough from prison in Massachusetts.¹⁵² One Democratic strategist argued that the Horton advertisement “changed the course of that race . . . . [I]t made white Americans—especially white southerners—raise an eyebrow and think, ‘We can’t have a man from Massachusetts releasing quote black criminals all across the country and letting them rape our white women and children.’ ”¹⁵³ While it cannot be determined exactly what impact the ad had on voters, Dukakis’ standing in the polls dropped.¹⁵⁴ Although it is impossible to measure the exact effect of an ad appealing to racial stereotypes, social psychology research makes clear the significant impact of subtle priming, even when it consists of a brief television advertisement.¹⁵⁵ Research on priming equips courts to better understand the extent of the potential damage to a minority candidate when assessing Senate factor six in cases involving subtle racial appeals.

As an over-arching theme, plaintiffs might turn to the Court’s guidance in Gingles and introduce implicit bias evidence by showing how “social and historical conditions” interact with existing laws or legal structures “to cause an inequality in the opportunities enjoyed” by minority voters.¹⁵⁶ A court’s analysis of the social and historical conditions could include an examination of the effect implicit biases might have played in creating those conditions. Because the stereotypes may have played a role in decisions producing inequities and because stereotypes are a product of societal narratives and mass media, such evidence would fall within the Gingles guiding language and should be admitted to enable courts to consider the “social and historical conditions” of the situations at issue.

In addition to the seven factors listed in the Senate Report, “other factors may also be relevant.”¹⁵⁷ The non-exhaustive nature of the Senate Report opens the door to the consideration of research on implicit bias in its own right.¹⁵⁸ Although implicit bias does not necessarily fit within the framework of every Senate factor or may not be relevant in every case, the totality of the circumstances test nonetheless provides a means through which courts can consider implicit bias in the electoral process.

B. Policy Prescriptions

In addition to pushing courts to rely on a framing of the totality of circumstances test to include evidence of implicit biases, stakeholders, like local advocates, foundations funding voting rights work, and civil rights lawyers, can also encourage jurisdictions to adopt practices aimed at reducing the effects of implicit biases in the electoral process. For electoral policies to effectively reduce the impact of implicit biases, stakeholders must ensure that the factors that exacerbate reliance on implicit biases are minimized. As mentioned above, factors such as stress, tight deadlines, and reliance on discretion contribute to one’s reliance on unconscious biases.¹⁵⁹ Policy prescriptions should focus on eliminating or reducing these and other conditions that foster implicit biases.

1. Training of Election Officials

While most Americans believe that they are able to treat everyone equally, more than seventy percent of Americans possess some level of implicit bias based on race.¹⁶⁰ Without a basic understanding of implicit bias or recognition that their decisions might be impacted by implicit biases, election officials cannot take the necessary steps to address the problem.

One of the first steps to reducing implicit biases in the electoral system is to educate election officials about implicit bias. Trainings aimed at raising awareness about the effects of implicit bias and at offering strategies through which individuals can reduce that bias have been proven to dramatically lower participants’ racial biases.¹⁶¹ Many counties and states already provide poll workers with training before Election Day,¹⁶² so including a segment on implicit bias would serve to start the process of rooting implicit bias out of the electoral system. In addition to including a segment on implicit bias in training programs, jurisdictions can also reduce the impact of implicit biases in electoral decisions by diversifying poll workers and other election officials. While “the majority of White respondents [to the Implicit Association Test] show a preference for White over Black, the responses from Black respondents are more varied.”¹⁶³ As Project Implicit explains, “[p]art of this might be understood as Black respondents experiencing the similar negative associations about their group from experience in their cultural environments, and also experiencing competing positive associations about their group based on their own group membership and that of close relations.”¹⁶⁴ Regardless of the reasons for the more varied rates of implicit racial association amongst Black participants taking the Implicit Association Test, that Black participants have more varied rates of implicit race associations and that IAT scores are strong predictors of real world behavior¹⁶⁵ are reasons to encourage diversity amongst poll workers. Because other minority group members are also likely to have more varied implicit racial associations, due to their “own group membership” and “close relations,” ensuring diversity of poll workers across all races may lower the risk that implicit biases impact the voting process.¹⁶⁶

2. Blind Review of Voter Registration Applications

Jurisdictions might reduce the role of implicit bias in the registration process by adopting blind review practices, in which the name and race (if such information is collected) of an applicant is temporarily redacted until other criteria for eligibility (i.e. criminal record, residency requirements, etc.) are confirmed. Blind screening has proven extremely effective in other areas. Major symphony orchestras, for example, have significantly increased their percentage of female musicians simply by investing in a cloth screen blocking the musician from the audition committee.¹⁶⁷ Although most orchestras previously required musicians to audition directly in front of an audition committee, by putting up a screen audition committees can no longer tell the musician’s race, age, gender, or other social identities.¹⁶⁸ The screens effectively break the audition committee’s unconscious reliance on the stereotype of a virtuoso musician as male. Similarly, temporarily redacting or otherwise blocking a voter’s name and race, can help prevent election workers from unconsciously relying on stereotypes about who is a “responsible citizen” and from unintentionally rejecting a registration form for a minor reason.

3. Legislative Responses

A comprehensive response to implicit bias in the electoral system must also provide for monitoring mechanisms similar to the Section 5 preclearance regime. In their study, What Do I Need to Vote?: Bureaucratic Discretion and Discrimination by Election Officials , Ariel White and her colleagues sent emails to more than 7,000 election officials in forty-eight states, including states and jurisdictions covered by Section 5 and states and jurisdictions not covered by Section 5.¹⁶⁹ Although the emails were identical and included an inquiry about voter identification requirements, the researchers sent some under a Latino alias and some under a White alias.¹⁷⁰ The researchers discovered that, overall, election officials were less likely to respond to an email sent by a Latino than to those sent by a White.¹⁷¹ Importantly, however, they found no bias in response rates or quality of responses in counties and municipalities covered under Section 5.¹⁷² White and her colleagues stated that this finding is consistent with findings in other areas, such as employment and housing, which also identify monitoring as a key factor in reducing discriminatory decision-making.¹⁷³ Legislation providing for a monitoring system, like the one previously required by Section 5, can be introduced at the state and national levels to reduce implicit bias in the electoral system.¹⁷⁴

4. Additional Research and Studies

Finally, but perhaps most important, it is time for individuals, organizations, jurisdictions, and other stakeholders to invest in research on the role of implicit bias in voting. Social psychologists and others have extensively studied the impact of implicit bias in other areas of law, from criminal justice to health care to education, yet little attention has been given to the effects of implicit biases in voting. Research is needed to examine how implicit biases may impact the election system at all stages, from planning for an election and allocating resources, to implementing new voting laws, to Election Day practices. Experiments could test participants’ decisionmaking when allocating polling machines to minority or White neighborhoods, the impact of the cross race effect when poll workers implement a photo identification requirement, or the amount of time a poll worker gives to resolving an issue for a White voter as compared to a minority voter.

In designing these experiments, researchers can use studies examining the effects of implicit biases in other areas as prototypes. An experiment examining how election officials allocate resources might replicate Laurie Rudman and Richard Ashmore’s experiment, in which they asked participants to determine how to administer a mandatory twenty percent budget cut to student groups.¹⁷⁵ Similarly, experiments regarding assistance offered during the voter registration process might replicate the University of Southern California’s study¹⁷⁶ in which researchers tested the response rates of state legislators to emails from hypothetical voters of different races asking about identification requirements.

Conclusion

When the right to vote is undermined through explicit or implicit biases, other rights are also at risk of becoming “illusory.”¹⁷⁷ The harm of denying a person the opportunity to participate in the political process on account of his or her race or ethnic background, even if not “intentional,” goes beyond the harm to the individual. As with cognate equal protection claims, the injury is also “to the law as an institution, to the community at large, and to the democratic ideal reflected in the processes of our courts.”¹⁷⁸ Steps can be taken to limit the effect of implicit bias in the electoral system. Section 2 results test claims provide a key opportunity for courts to consider evidence regarding how implicit biases might impact the electoral process. In addition, policy reforms at a more local level—holding implicit bias trainings for poll workers and redacting race on voter registration forms—provide a means through which stakeholders can reduce the detrimental impact of implicit biases on elections. Pursuing both litigation and policy strategies allows for a practical shift, in which our legal frameworks and daily electoral practices better account for the unconscious nature of much bias today. Only when all forms of discrimination are rooted out of our electoral system will we truly be able to ensure the health of our democracy.