In the previous chapter we discussed race as a social–cognitive construct that evolves over time and in which racial categories reflect one’s own or one’s ancestors’ physical features and associated characteristics that have acquired socially significant meaning. In this chapter we turn to the concept of racial discrimination, defining it from a social science perspective, which includes not only legal definitions of discrimination but also aspects that go beyond legal concepts. We provide examples of the large and persistent differential outcomes by race in various social and economic domains that make racial discrimination an important topic for social science analysis and motivate our examination of methods for measuring the role that race-based discrimination may play in those differences.
For completeness, we examine the legal definitions of discrimination. Although discrimination is often understood in legal terms because, once it has been identified, legal consequences ensue, our definition encompasses forms of discrimination that may not be explicitly unlawful or easily measured.
A DEFINITION OF RACIAL DISCRIMINATION
In this report, we use a social science definition of racial discrimination that includes two components: (1) differential treatment on the basis of race that disadvantages a racial group and (2) treatment on the basis of inadequately justified factors other than race that disadvantages a racial group (differential effect). Each component is based on behavior or treatment that disadvantages one racial group over another, yet the two components differ
on whether the treatment is based on an individual’s race or some other factor that results in a differential racial outcome. As we discuss further below, we are particularly interested in discrimination that disadvantages racial minorities.
The first component of our definition of racial discrimination occurs when a member of one racial group is treated less favorably than a similarly situated member of another racial group and suffers adverse or negative consequences. This definition of discrimination is used in many social science fields (e.g., economics, psychology, sociology) to refer to unequal treatment because of race. Intentional discrimination of this kind is frequently unlawful under either the Constitution or specific legislative prohibitions, such as those in employment, housing, and education. The second component of our definition of racial discrimination includes some instances in which treatment based on inadequately justified factors1 other than race results in adverse racial consequences, such as a promotion practice that generates differential racial effects. A process with adverse racial consequences may or may not be considered discrimination under the law, depending on whether there is a sufficiently compelling reason for its use and whether there are alternative processes that would not produce racial disparities.2 In the areas in which this type of discrimination is unlawful, the reason is to curtail the use of unintentional practices that can harm racial minorities, as well as to sanction intentional discrimination that might not be identified because of the difficulty in establishing intent in the legal setting.3
The two components of our definition—differential treatment and differential effect discrimination—are related to, but broader than, the standards applied in a large body of case law—disparate treatment and disparate impact discrimination (see the detailed discussion below in this chapter).4 Legally defined, disparate treatment racial discrimination occurs when an
individual is treated less favorably—for example, is not hired for a job—because of his or her race. Disparate impact racial discrimination occurs if a behavior or practice that does not involve race directly has an adverse impact on members of a disadvantaged racial group without a sufficiently compelling reason. An example is an employment practice or policy against hiring job applicants with a criminal arrest record when such a policy results in proportionately fewer hires for disadvantaged racial groups while not significantly advancing any legitimate employer interests. These kinds of practices and policies—whether intentionally or unintentionally harmful—are deemed unlawful unless a sufficiently compelling business reason can be supplied to justify them.
Although our definition encompasses the legal definitions of discrimination, we do not believe that a social science research agenda for measuring discrimination should be limited by those legal definitions. Although many of the issues that we discuss may be relevant to certain debates within the courts, our primary intention in this report is to provide guidance to social science researchers interested in measuring racial discrimination. Therefore, in our definition we allow both categories to include a range of behaviors and processes that are either not explicitly unlawful or not effectively prohibited because of difficulties in measurement or proof (see Chapter 4). For example, subtle forms of discrimination might not be susceptible to legal challenge but fall within the scope of our definition. An example of a subtle form of discrimination (perhaps unintentional) would be when interviewers of job applicants more frequently adopt behaviors (e.g., interrupting, asking fewer questions, using a hectoring tone) that result in poor communication and consequently poorer performance by disadvantaged minority applicants as compared with other applicants. Compared with overt discrimination, it is often more difficult to find proof that subtle discrimination has occurred and to address it legally, even if in theory such subtle discrimination constitutes actionable disparate treatment discrimination.
In addition, many legislative and administrative actions that have a discriminatory impact are not legally prohibited because the constitutional mandate against racial discrimination does not recognize the disparate impact theory of discrimination. Social scientists, however, will still want to ascertain the possibly discriminatory effects of such legally permissible governmental actions. A final example of discrimination’s impact that we want to measure as social scientists, but which may not be unlawful, occurs when discriminatory effects cumulate across domains. Discrimination by real estate agents may result in housing segregation, which in turn affects educational quality (because of local tax financing of the schools) and long-term educational and labor market outcomes. Although discriminating real estate agents can be found liable for housing market discrimination, there is
no legal mechanism to allocate blame for educational or labor market differences that such discrimination might induce. Yet, as social scientists we want to identify and measure these cross-domain effects.
LIMITING THE DISCUSSION
The experience of discrimination and its consequences may vary with several factors, including the domain in which it occurs (e.g., the labor market, the health care system, the criminal justice system, the housing market); the actors involved (e.g., employers, insurance companies, police officers, mortgage lenders, neighbors); and the targets (e.g., African Americans, whites, Hispanics, American Indians, Asians). Within the scope of our broad definition of discrimination, we focus our analysis on specific aspects of racial discrimination in the United States. We are concerned primarily with discrimination that has adverse social and economic consequences for disadvantaged racial groups. We use the term disadvantaged racial groups interchangeably with minority groups and nonwhite groups and refer to non-Hispanic whites as the majority group. These terms describe the social stratification (rather than the numerical proportions) of different racial groups in the United States. We recognize that racial groups in different communities, institutions, and even countries (e.g., South Africa) can be in the numerical majority but still experience discrimination.
We acknowledge that non-Hispanic whites may face discrimination that results in adverse consequences (so-called reverse discrimination). However, members of disadvantaged groups have more often been discriminated against in various social and economic arenas (Council of Economic Advisors, 1998; National Research Council, 2001a), and ongoing discriminatory practices and policies can undermine efforts to overcome these disadvantages. Therefore, while we do not rule out the possibility of so-called reverse discrimination, we do not address discrimination against non-Hispanic whites in this report.
We refer more often to evidence of racial discrimination by whites against blacks, although we recognize that other racial groups, including whites, as well as some ethnic groups, face discrimination.5 Primarily, this is a result of the larger literatures on black–white disparities and research to
measure discrimination. In many data sets, sample sizes are too small for analysis of some groups (such as American Indians), or there is no separate identification of groups, such as Asian Americans or Native Hawaiians and other Pacific Islanders, or subgroups, such as Mexican Americans or Puerto Ricans.
Given limited time and space, we primarily discuss racial discrimination in general terms and do not discuss the differences in experiences of discrimination among racial groups, although we recognize that each group has a different historical experience. Furthermore, the broad categories used in most of the data reported here—such as African American or Hispanic—are very heterogeneous in terms of nativity, phenotype, culture, religion, and socioeconomic background. Although important to consider, nuanced attention to these differences is beyond the scope of this report.
We do not address policy issues regarding racial discrimination. For example, we do not discuss the implications or effectiveness or costs of policies intended to alleviate discrimination (e.g., affirmative action or diversity policies). Our charge is to assess social science research methods for measuring racial discrimination. One use of such methods is to assist in policy formulation and evaluation, but discussion of policies as such goes beyond our charge.
One aspect of differential behavior largely beyond the scope of this report is differences in associational choices made by members of different racial groups, such as whom one lives with and marries, whom one’s friends might be, and even whom one sits next to at lunch. Issues of associational choice do not fall into our definition of discrimination, although they may have large and adverse effects on differential racial outcomes. Most (though certainly not all) antidiscrimination efforts are focused on those arenas in which there are contracts or explicit markets for the exchange of goods and services. Ideally, equal access to those markets (be they in employment or in housing) would be available to all racial groups. There is neither a legal nor a social tradition of intervening in associational choices as long as those choices are based entirely on individual preferences and not on group-imposed exclusionary policies or practices. It is not always clear when an associational decision is freely chosen and when it is subject to such tight constraints that it might be considered discriminatory. Although important to the broad understanding of racial group differences in our society, these are issues that necessarily lie beyond the mandate of this panel and that we cannot adequately treat in this report.
Finally, our definition of discrimination is based on behaviors and practices, and as such it differs from a definition that also includes prejudiced attitudes and stereotypical beliefs. Discriminatory behaviors and practices may arise from prejudice and stereotyping, but prejudice need not result in differential treatment or differential effect. Similarly, whereas discrimina-
tory behavior in many domains is unlawful, prejudiced attitudes and stereotypical beliefs are not.
DIFFERENTIAL OUTCOMES BY RACE
Evidence of large and persistent differentials in social, economic, and political outcomes among racial and ethnic groups in the United States characterizes virtually every social domain. Indeed, were there not such marked differences, there would be little reason to convene a panel of social scientists to study methods for measuring race-based discrimination. Even though prejudices and stereotyping might be present and individual cases of discrimination might occur, an absence of observable differences in outcomes among racial groups would almost preclude social science measurement of the role of racial discrimination in American society.
To motivate our report, we provide examples of differential outcomes among racial groups in five domains: education, the labor market, the criminal justice system, the housing market and mortgage lending, and health care.6 In these examples, we draw no conclusions about whether or to what extent differential outcomes by race are caused by discrimination. The magnitude of the differentials in these—and other—domains, however, is a primary reason to be concerned about our ability to identify and measure racial discrimination. Also, the greater the extent to which differential outcomes are the result of discriminatory behaviors or processes, the greater is the likelihood that antidiscriminatory efforts would be needed to reduce these differences.
Racial classification and many factors that are correlated with race (e.g., family structure, parental education, poverty, access to computers, and linguistic diversity) are associated with different educational experiences and levels of educational attainment (Choy, 2002; Lloyd et al., 2002; Mare, 1995). Research shows that blacks, Hispanics, American Indians, and Native Hawaiians and other Pacific Islanders—compared with whites and Asians—are more likely to attend lower-quality schools with fewer teachers and material resources and greater concentrations of poor, homeless, limited English-speaking, and immigrant students (Kahlenberg, 2001; Lee et al., 2001; Natriello et al., 1990; Van Hook, 2002). They are also more likely to have lower test scores, drop out of high school, not graduate from college, and attend lower-ranked programs in higher education (see Na-
tional Research Council, 2002a). For example, the U.S. Department of Education (2001a) reports that African American and Hispanic students are less likely to have completed advanced levels of math and science coursework compared with Asian and Pacific Islander and white students. However, overall educational attainment may vary substantially among Asian groups—for example, Japanese, Koreans, and Asian Indians versus Cambodians, Laotians, and Hmongs (U.S. Census Bureau, 1993).
Hispanics continue to face obstacles to educational achievement. Between the late 1970s and 1998, they had significantly lower educational attainment and higher dropout rates than both blacks and whites (Hauser et al., 2002). In 2000, 57 percent of Hispanics aged 25 and over had obtained at least a high school degree, compared with 79 percent of blacks and 85 percent of whites (U.S. Census Bureau, 2002). One factor influencing the education gap between whites and Hispanics is the increasing numbers of disadvantaged Hispanic immigrant groups entering the United States. Poor educational outcomes for many Hispanic groups may lead to subsequent disadvantages in social and economic opportunities (e.g., lifetime earnings or civic participation; see Blank, 2001).
Employment and Income
Black Americans are more likely to experience unemployment as teens and adults, to work at lower wages, to have lower wage growth over time, and to accumulate less wealth relative to whites (Altonji and Blank, 1999). Indeed, unemployment rates for blacks are generally twice those for whites. In 2002 the average annual unemployment rate for black workers aged 16 and over (10.3 percent) was nearly twice the overall unemployment rate (5.8 percent) and just over twice the rate for whites (5.1 percent). The unemployment rate for Hispanics was 7.6 percent that year (Bureau of Labor Statistics, 2003).
Median weekly earnings for blacks ($499 in 2002) and Hispanics ($424) are much lower than for whites ($627). The gap between Hispanics and whites has grown at a particularly rapid rate, a fact that may be attributable to differences in educational achievement (Bureau of Labor Statistics, 2003). These earnings differentials are reinforced by substantial differences in the occupational categories in which various racial groups are clustered, with disadvantaged racial groups generally having lower-status as well as lower-wage occupations. Empirical research on labor market outcomes for Asians and American Indians is more limited, reflecting the lack of data on these groups (Altonji and Blank, 1999).7
Perhaps the largest racial differences are observed with respect to wealth, which reflects not just current earnings but cumulative lifetime (and even cross-generational) differences. The average net worth of blacks is just a fifth that of whites (Conley, 1999; Oliver and Shapiro, 1995).
Disadvantaged racial groups (particularly blacks) are disproportionately represented in the criminal justice system compared with non-Hispanic whites. Racial differences are largest in the corrections system, in which the incarceration rate for blacks is about eight times that for whites (Blumstein, 1982, 1993). In large part, this differential reflects more frequent arrests of blacks for serious crimes (e.g., murder and robbery), for which the ratio of black to white arrest rates is about 7, relative to less serious crimes (e.g., burglary and drugs), for which the ratio is closer to 3.8
In some cases, the punishment for crimes committed by blacks is significantly different from that for similar crimes committed by whites. One reason is lower thresholds for mandatory minimum sanctions for crimes that are more likely to be committed by blacks. This difference is particularly striking in a provision of the federal Anti-Drug Abuse Act of 1986: A mandatory minimum sentence of 5 years is imposed for possession of as little as 5 grams of crack cocaine; in contrast, a possessor of powder cocaine must have at least 500 grams to receive a mandatory minimum sentence of 5 years. In 2000, 85 percent of sentenced crack cocaine offenders, who were sentenced for possessing very small amounts of cocaine, were black, but only 31 percent of sentenced powder cocaine offenders, who had to have large amounts of cocaine to be sentenced, were black (51 percent were Hispanic and 18 percent white). Although the disparate sentencing thresholds are associated with the crime rather than the race of the offender, they have a marked differential racial impact.
Blacks are disproportionately represented not only as offenders but also as victims of crime (Sampson and Lauritsen, 1997; Walker et al., 1996; Weich and Angulo, 2002). Sampson and Lauritsen report that crime victimization rates vary systematically across racial and ethnic groups. Compared with whites, blacks were six times more likely to be murdered in 2000 (see U.S. Department of Justice, 2001).
In criminal justice research, there is a lack of consistent data on crime
These ratios are based on arrest data from Table 43 of the FBI’s Uniform Crime Reports for 2000 (U.S. Department of Justice, 2000) and 2000 population data from Table 10-2 in Chapter 10 of the present report (with those of “Other Race,” who are predominantly Hispanics, being counted as “white” because the arrest reports do not have a separate count for Hispanics).
for Asian Americans and American Indians. However, Weich and Angulo (2002) point out that Asian American youths are far more likely than whites to be transferred to adult courts, convicted in adult courts, and incarcerated in youth and adult prisons. Also, although African Americans are overrepresented in federal and state prisons relative to their proportion in the population (Walker et al., 1996), American Indians actually have the highest incarceration rate for any race: In 1997, 1,083 of every 100,000 American Indians in the United States were incarcerated (Smelser and Baltes, 2001). Moreover, American Indian youths, who are subject to federal rather than state prosecution, often end up facing harsher sentences than if they were subject to state prosecution (Weich and Angulo, 2002). As a result, approximately 60 percent of youths in federal custody are American Indian.
Housing Markets and Mortgage Lending
Housing segregation among black Americans is far greater than among any other identifiable group. For example, blacks are much more likely to live in segregated neighborhoods, to rent rather than own a home, and to have a lower-valued home when they are homeowners (Charles and Hurst, 2002; Massey, 2001). Although legal segregation and exclusion ended in 1968 with the Fair Housing Act, racial disparities in certain neighborhoods and housing markets continue. In addition, disparities in aggregate lending to black and white neighborhoods continue to exist in many communities (for a review of the evidence, see Ladd, 1998; Munnell et al., 1992, 1996; Turner and Skidmore, 1999; Turner et al., 2002a; Wyly and Holloway, 1999). For example, Wyly and Holloway found that applicants were more likely to have their loans approved in Atlanta neighborhoods in which their race was predominant (i.e., blacks were approved more in black neighborhoods and whites in white neighborhoods). Other studies have shown significant differences in the probability of mortgage loan approval by race in Boston (Carr and Megbolugbe, 1993; Munnell et al., 1992) and Milwaukee (Squires and O’Connor, 2001).
Health Care and Health Outcomes
African Americans, Hispanics, Asian Americans, and American Indians and Alaska Natives face large barriers to health care services as compared with whites (for recent reviews, see Institute of Medicine, 2003; Mayberry et al., 2000). These groups tend to experience lower levels of access to care and to receive lower-quality health care (Institute of Medicine, 2003). For instance, African Americans and Hispanics compared with non-Hispanic whites are less likely to receive kidney dialysis or transplants (Epstein et al., 2000), are less likely to receive appropriate cancer diagnostic tests or treat-
ments (Imperato et al., 1996; McMahon et al., 1999), and are more likely to receive less-than-desirable procedures, such as limb amputation for diabetics (Chin et al., 1998; for additional references, see Institute of Medicine, 2003).
Disadvantaged racial groups are also more likely than whites to suffer from adverse health status and outcomes (Institute of Medicine, 2003; Keppel et al., 2002; National Research Council, 2001a). Thus, substantial racial differentials exist for rates of infant mortality, certain cancers, cardiovascular disease, and kidney disease (Keppel et al., 2002). For example, American Indians are more likely than other racial groups to die from diabetes, liver disease and cirrhosis, and unintentional injuries (Institute of Medicine, 2003). There is also considerable evidence that African Americans have disproportionately high levels of hypertension compared with other racial groups (see Anderson, 1989).
Interpreting Differential Outcomes
Differences in outcomes by race do not themselves provide direct evidence for the magnitude or even the presence of racial discrimination in any particular domain. These outcome differences are the result of any number of factors that may or may not include racial discrimination in that domain. For instance, racial disparities in the labor market (e.g., in hiring or wages) may reflect differences in school quality and achievement rather than any racial animus within the labor market per se. (We discuss these issues further in Chapter 11.)
Although racial disparities continue to exist in many domains, both social and legal changes have improved opportunities for many nonwhites in the United States. Recently, the Brookings Institution (2000) reported the federal government’s 50 most important achievements in the past 50 years, including expanding the right to vote (ranked 2), promoting equal access to public accommodations (3), reducing workplace discrimination (5), increasing access to postsecondary education (19), and increasing low-income families’ access to health care (34). Examples of legislative acts designed to promote equal opportunity and reduce discrimination include the Civil Rights Act of 1964 banning discrimination in employment and in public accommodations; the Voting Rights Act of 1965 (and its subsequent extensions and amendments), allowing full political participation of nonwhite groups once excluded from voting; Federal Executive Order 11246, requiring compliance by government contractors with federal antidiscrimination policies and the development of administrative systems to monitor compliance; and the Fair Housing Act of 1968, banning discrimination in housing.
Nonetheless, differential outcomes by race persist and motivate analysis to understand contributing factors, including the possible role of racial
discrimination. Black–white gaps in income, employment, higher education, test scores, housing segregation, health care, and treatment within the criminal justice system are large. Such sizable and persistent differences in outcomes, by themselves, are problematic and important to address. Even if differential outcomes do not in and of themselves prove that discrimination is occurring, they tell us where to look when seeking to assess whether discriminatory behavior occurs in various social arenas. In the example cited above, for instance, racial disparities in the labor market may reflect not only discrimination in that domain at that time (e.g., wage differentials) but also discrimination in earlier interactions (e.g., labor market experience) and in other domains (e.g., education).
Differential outcomes might be less informative if we believed that the groups involved were innately different. Yet, as noted in Chapter 2, scientists have not determined a genetic basis for the socially based racial and ethnic categories in American society—categories whose meaning has changed over time (e.g., the assimilation of previously “nonwhite” European immigrant groups into the “white” category). We can then infer that these differential outcomes reflect deep differences in the historical and current experiences and environment of disadvantaged racial groups versus non-Hispanic whites. For instance, surveys show that nonwhites perceive much greater discrimination toward nonwhite racial groups and experience much more discrimination themselves compared with whites (Bobo, 2001; Morin, 2001; Schuman et al., 1997). Cumulative disadvantage across generations—in access to nutritious food, decent housing, remunerative employment, and secure and stress-free environments—is a possible way to interpret the differences in current outcomes among nonwhite Americans (see discussion in Chapter 11).
THE LEGAL DEFINITION OF DISCRIMINATION
Thus far we have presented a definition of discrimination and examined racial disparities across several domains. As a point of comparison, in this section we look at the legal definitions of discrimination and identify the circumstances under which a legal finding of discriminatory behavior can be made. The law represents an important venue in which racial discrimination is often identified and measured. In a legal setting, once an act has been labeled as discriminatory, legal remedies, both monetary and injunctive, may be awarded.
An elaborate array of federal and state constitutional, statutory, and administrative provisions broadly prohibit discrimination on the basis of race in a vast range of public and private behaviors. A large body of law has developed to give content to this broad prohibition by defining specifically what constitutes impermissible discrimination. Because the foundations of
these laws emanate from different jurisdictions and legal authorities, there is no single definition of impermissible racial discrimination; standards depend on the particular jurisdiction or actor involved. Nonetheless, as noted above, two important doctrinal concepts—disparate treatment and disparate impact discrimination—are useful in defining the nature of the legal prohibition. Each is discussed in turn below.
Disparate Treatment Discrimination
The core concept of disparate treatment discrimination emanates from the constitutional requirement of equal protection under the law and is codified in the main federal statute prohibiting racial discrimination in employment—Title VII of the 1964 Civil Rights Act. This statute prohibits an employment practice that affects an individual’s employment “because of such individual’s race….” Thus, an employer who refuses to hire, fails to promote, or discharges a worker because of his or her race is guilty of disparate treatment discrimination. So, too, is an employer who decides to pay nonwhite workers less than white workers or to discipline the former more heavily for identical conduct.
The language “because of” is interpreted as requiring proof that race was a motivating factor for the employment practice. In theory, the requirement that the discrimination be intentional before it runs afoul of the law may protect an employer who acts without conscious awareness of having discriminated, a phenomenon that the research literature in psychology indicates is common. For this reason, some legal scholars have suggested that the legal theory of intentional discrimination is flawed and should be expanded to prohibit unconscious or negligent acts of discrimination (Allen, 1995; Oppenheimer, 1993).
In practice, however, a defense that the discrimination is “unconscious” is virtually never encountered in employment discrimination litigation, which typically focuses on two issues: (1) the plaintiff’s threshold demonstration of racial disparity in treatment and (2) the credibility of the nondiscriminatory reasons for this disparate treatment offered by the employer. Therefore, although in theory any nondiscriminatory reason will constitute a defense against a charge of disparate treatment discrimination, in practice an employer will be more likely to lose the case if the reason does not appear to be sufficiently linked to the plaintiff’s lack of ability to perform the job or demonstrated misconduct. Nonetheless, the courts have held that the burden of persuading the court that the employment decision was discriminatory remains with the plaintiff. Even if the plaintiff establishes that the employer’s proffered reason for the employment action is not truthful, the employer will prevail if the plaintiff cannot persuade the court that race was a motivating factor.
Once it has been established that an employer has intentionally discriminated on the basis of race, the reason for the differential treatment will ordinarily not be relevant (unless it is pursuant to the implementation of a valid affirmative action plan). Accordingly, intentional racial discrimination will be deemed unlawful whether the employer acted because he or she dislikes nonwhites (say, blacks), prefers a nonblack ethnic group that is consequently favored, or believes that blacks will be on average less productive. Similarly, an employer cannot engage in disparate treatment on the grounds that customers or other employees demand such racial exclusion or would otherwise prefer it.
On the other hand, a decision to locate a plant in a suburb or in a state with a low black (or other nonwhite) or Hispanic population may have serious adverse consequences for potential black (or other nonwhite) and Hispanic employees. But this locational decision, even if motivated by racial animus, will not be prohibited unless it is deemed an “employment practice.” Cases focusing on infrequent institutional behaviors (as opposed to regularly implemented procedures) are essentially unknown for both doctrinal and practical reasons, a fact that underscores how potentially significant choices that may be affected by discriminatory motives can impair the employment prospects of minority groups without generating any legal response.
The task of measuring racial discrimination in a legal case often begins with the documentation of various racial disparities in such areas as income, wealth, educational attainment, incarceration or involvement in the criminal justice system, and health. Of course, as noted above, the mere presence of large disparities in some of these measures does not necessarily mean that discrimination exists. For example, in the United States it is well documented that women live far longer than men, but it is rarely thought that discrimination against men explains their substantially higher rates of death. Similarly, men commit suicide and are incarcerated at vastly higher rates than women, yet again discrimination against men is unlikely to play a large explanatory role in these male–female disparities. Moreover, as discussed above, even when a racial or other group disparity is the product of discrimination, it is not necessarily the result of discrimination occurring at the point in time at which the disparity becomes manifest. For example, employers are generally not held liable under Title VII of the 1964 Civil Rights Act for disparities resulting from pre–labor market discrimination against blacks.
Disparate Impact Discrimination
Although disparate treatment was the original conception of unlawful discrimination, in 1971 the Supreme Court established a second, poten-
tially broader notion of discrimination—the disparate impact standard. Under this doctrine, which, like disparate treatment, was judicially crafted in the arena of employment discrimination, the court first asks if an employment practice, even though facially neutral, has an adverse impact on members of a protected group. Once a finding of disparate impact has been made, the court will rule the challenged practice unlawful unless a sufficiently compelling business justification can be supplied for retaining it. The precise legal standard for this justification defense, first legislatively articulated in the Civil Rights Act of 1991, is that the defendant must prove “that the challenged practice is job related for the position in question and consistent with business necessity.” Moreover, this justification will be dismissed if the employer’s proffered legitimate business interest could be satisfied by another equally effective employment practice having a less racially adverse impact.
Everything from minimum educational requirements to rules against hiring those with arrest records to grooming standards that prohibit beards to certain types of seniority systems has been deemed under certain circumstances to constitute disparate impact discrimination against disadvantaged racial groups in employment. The goal of the doctrine has been to remove artificial barriers that prevent the economic progress of members of protected groups. At the same time, the rationalization of employment processes that has followed in the wake of the development of the disparate impact doctrine may have brought greater fairness to the process of selection of all employees.
Discrimination Law Regarding Governmental Actions
Government actions fall under a somewhat different set of legal rules. The constitutional prohibition against violations of equal protection (directly applied to the states under the Fourteenth Amendment and indirectly applied to the federal government under the due process clause of the Fifth Amendment) prohibits racial classifications unless justified by a “compelling interest” and unless the policy is “narrowly tailored” to serve that interest. However, the definition of racial discrimination for purposes of evaluating constitutional violations is narrower than the two-part legislative standard that governs employment discrimination law.
Specifically, although the equal protection clause prohibits disparate treatment discrimination that fails to have the most compelling societal justification, the Constitution prohibits only intentional discrimination; evidence of disparate impact alone will not establish a violation. Thus, the Constitution does not restrict a government from engaging in acts that harm disadvantaged racial groups unless the harm is caused intentionally. Moreover, knowing that a certain practice will cause harm is not enough to
render it an intentional act of discrimination barred by the equal protection clause. As the court has emphasized, a government is not prohibited from acting in spite of harm to members of disadvantaged racial groups; it is banned only from causing harm because of race.
This constitutional interpretation reflects the fact that many neutral governmental actions have predictable effects that either benefit or harm certain racial groups and that allowing all these actions to be challenged on equal protection grounds would make the federal courts the arbiters of a vast array of legislative and executive conduct. For example, the mortgage interest deduction for residential housing disproportionately benefits whites because of their greater housing wealth and possibly dampens investments in other types of productive capital that might generate more jobs that could disproportionately advantage blacks. Similarly, the war on drugs is designed to identify and punish the tens or even hundreds of thousands of workers in the illegal drug trade, a disproportionate number of whom will inevitably be drawn from disadvantaged groups having less abundant opportunities in the legitimate economy. Yet no doctrine of law would permit either of these ostensibly neutral governmental programs to be challenged as racially discriminatory. Similarly, governmental social programs that disproportionately benefit a racial or ethnic group cannot be challenged on that basis alone on equal protection grounds.
Any form of racially preferential treatment by a government entity—whether in giving preference to minority contractors or to minority applicants to state universities—is subject to strict judicial scrutiny. A racially based treatment implemented by government, even if motivated by the desire to promote affirmative action, will violate the Constitution unless it is “narrowly tailored” to serve a “compelling government interest” (Adarand Constructors, Inc. v. Peña, 515 U.S. 220, 237-8 ).9
We adopt a broad definition of racial discrimination for use in social science research, which includes individual behaviors and institutional processes but not attitudes or beliefs as such. Our definition includes two components that are related to (but broader than) a large body of case law: differential treatment on the basis of race that disadvantages a racial group and treatment on the basis of inadequately justified factors other than race that disadvantages a racial group (differential effect). In defining discrimination for this report, we focus primarily on discrimination that has harmful consequences for disadvantaged racial minorities.
Our definition is not limited to those actions defined as discriminatory within a legal framework but also encompasses subtle behaviors and processes and cumulative discriminatory effects that may not be explicitly unlawful or easily measured. In the next chapter, we discuss in more detail the possible ways in which discrimination may manifest itself and return to a discussion of when these discriminatory behaviors may or may not be explicitly unlawful.
There is a history of racial exclusion in the United States and a persistence of large disparate outcomes for racial groups across many societal domains. Although such disparities may not in themselves signal the presence of discrimination in any particular domain or event, they are problematic and motivate our work to assess social science analytical methods for measuring the role of racial discrimination in American society today.