DISPARATE IMPACT AND THE QUOTA DEBATES: Law, Labor Market Sociology, and Equal Employment Policies

Disparate Impact and the Quota Debates: Law, Labor Market Sociology, and Equal Employment Policies Author(s): Robin Stryker Source: The Sociological Quarterly, Vol. 42, No. 1 (Winter, 2001), pp. 13-46 Published by: Blackwell Publishing on behalf of the Midwest Sociological Society Stable URL: http://www.jstor.org/stable/4120924 Accessed: 18/09/2009 21:34 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=black. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Blackwell Publishing and Midwest Sociological Society are collaborating with JSTOR to digitize, preserve and extend access to The Sociological Quarterly.

http://www.jstor.org

IMPACT AND THEQUOTADEBATES: DISPARATE Law,LaborMarketSociology, and EqualEmploymentPolicies Robin Stryker

Universityof Minnesota Bringingsociologicaltheoryand researchto bear on the "quotadebates"dogging discussionof federalcivil rightslegislationin the early1990s,this articlehighlights sociology'srole in shapingemploymentlaw and shows how apparentlytechnical legal argumentsaboutallocatingburdensof proofaffectlabormarketresourceallocation amongthe classes,races,and genders.Contrastinginstitutional-sociological with liberal-legalconceptsof discrimination, the articleshowswhydisparateimpact theoryhas been the most sociologicalapproachto Title VII enforcement.It also showshow disparateimpact-a theoryand methodfor establishinglegallycognizable employmentdiscrimination injuriousto womenand minorities-is, and is not, relatedto affirmativeaction-a policy encompassinga broadrangeof procedures intended to providepositive considerationto membersof groups discriminated againstin the past.Finally,a competingincentiveframeworkis used to show that, althoughdisparateimpactcreatessome incentivesfor employersto adopt quota hiring,such incentivesare counter-balanced by majorincentivesworkingagainst stemfromdisparateimpact race-andgender-based quotas.Majorcounterincentives itself,from other aspectsof equal employmentlaw,and from organizationalgoals shapingbusinessresponseto the legalenvironment. Key turning points in contemporary equal employment law have been marked by political exploitation of quota rhetoric.What was ineffective in the late 1960s against a backdrop of the civil rights movement and the immediate history of legally enforced second-class citizenship for African Americans was much more effective in the late 1980s and early 1990s,when quota rhetoric helped undercut affirmative action policies (Pedriana and Stryker 1997; Stryker, Scarpellino, and Holtzman 1999). No sociologist could have failed to notice or appreciate implications of "the quota issue" as it became synonymous with political and legal debates over affirmative action. A related but different political battle that also invoked quota rhetoric has largely receded from view. This other battle was occasioned by a string of 1989 Supreme Court setbacks for civil rights enforcement and ultimately resulted in the 1991 Civil Rights Act. The 1991 legislative compromise undid several of the Court rulings that had narrowed the reach of federal civil rights law, but not until former President George Bush had vetoed the earDirect all correspondence to Robin Stryker,Department of Sociology, University of Minnesota, 909 Social Sciences Building, 267 19thAvenue South, Minneapolis,MN 55455-0412;e-mail:[email protected]

The SociologicalQuarterly,Volume42, Number1, pages13-46. Copyright? 2001by The MidwestSociologicalSociety. All rightsreserved.Sendrequestsfor permissionto reprintto: Rightsand Permissions,Universityof CaliforniaPress, JournalsDivision,2000CenterSt.,Ste.303,Berkeley,CA 94704-1223. ISSN:0038-0253

14

THE SOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

lier Civil Rights Restoration Act of 1990, calling it a "quota"bill (Stryker et al. 1999). The legal issue inspiring Bush's ire involved allocating and interpreting methods and burdens of proof in disparate impact discriminationunder Title VII of the Civil Rights Act of 1964.An issue that no doubt would elicit a great yawn from all but legal aficionados became a compromise-resistantpolitical hot potato when framed in terms of quotas. Where one side kept up a constant "drumbeat"of quotas, the other defensively denied the quota charge (Stryker et al. 1999). I have written elsewhere about the political framing of debate over the 1990-1991 civil rights acts and the impact of framing this debate as "the quota issue" (Strykeret al. 1999). Here I explore whether the equation of disparate impact theory with employment quotas has any basis in empirical reality.

MYTHAND REALITY IMPACT: DISPARATE Disparate impact theory was an aggressive enforcement strategy crafted in Title VII's early years by civil rights organizations in conjunction with Equal Employment Opportunity Commisssion (EEOC) attorneys (Blumrosen 1972; 1993; Graham 1990). From 1971, when the Supreme Court unanimously recognized disparate impact theory in Griggs v. Duke Power Company, until 1989, when a divided Supreme Court vitiated Griggs in WardsCove v.Atonio, disparate impact theory had seemingly had the "imprimatur of permanence" (Blumrosen 1972, p. 63). When the Supreme Court effectively undid disparate impact in Wards Cove, civil rights groups and their allies in Congress fought back hard.They complained that WardsCove dealt a "devastatingblow" to civil rights enforcement on behalf of women and minorities injured by discrimination (Norton 1990,p. 197;see also Biskupic 1989;Belton 1990,p. 251). As I have shown elsewhere (Stryker et al. 1999), the ensuing political battle over disparate impact theory focused heavily on alleged links between disparate impact and affirmative action employment quotas.The WardsCove Court invoked the undesirability of quotas when it argued that allowing a lower court ruling that plaintiffs had made a prima facie disparate impact case to stand "would mean that any employer who had a segment of his work force that was-for some reason-racially imbalanced, could be hauled into court and forced to engage in the expensive and time-consuming task of defending the 'business necessity' of the methods used to select the other members of his work force.The only practicable option for many employers will be to adopt racial quotas.... This is the result that Congress expressly rejected in draftingTitle VII" (57 Law Week4583, p. 4586). WardsCove did not expressly overrule disparate impact,but it did "establish doctrinal rules regarding statistical evidence, the business necessity defense, causation, and burden-shiftingrules that left little doubt that it intended to dismantle Griggs" (Belton 1990, p. 237). Much of the Court's zeal came from equating disparate impact with employment quotas-a move that echoed earlier Reagan administration rhetoric (see Stryker et al. 1999). Did the rhetoric mobilized by the WardsCove Court and by opponents of the 19901991 civil rights legislation have any basis in empirical reality? If it did, it is not because disparate impact and affirmative action are the same thing. Disparate impact is a method of proving discrimination injurious to women and minorities, whereas affirmative action involves special consideration for minorities and women because these groups have suffered discrimination in the past. In addition, affirmative action may or

DisparateImpactand the Quota Debates

15

may not include quotas.Affirmative action can involve anything from trying to broaden the diversity of an applicant pool by aggressively recruiting women and minorities, to hiring and promotion that consider race and gender along with many other factors, including standardhuman capital-related qualifications,to setting "soft"targets or goals for minority and female hires and promotions, to hiring and promotion based solely on race and gender hiring ratios (i.e., inflexible quotas) (Jencks 1992; Reskin 1998). For there to be any truth in the WardsCove equation of disparate impact with racial quotas, judicial application of disparate impact theory must cause employers not just to adopt affirmative action programs but to adopt programs that hire strictly "by the numbers" (Fisher 1985, p. 26). Sorting out whether this is likely requires examining legal theories of discriminationand especially how statistical evidence is mobilized to prove discrimination under Title VII. Once this is accomplished, we can see how Title VII relates to affirmativeaction. Likewise, we can separate out fact from myth in rhetorical linkage of disparate impact theory with employment quotas. In the remainder of this article, I first show how the legal theory of disparate impact intersects with sociological theory and research on labor markets and how legal resolution of the disparate impact question affects the allocation of labor market resources among the classes, races and genders. I then use a competing incentive framework to show that, although disparate impact theory creates some incentives for employers to use quota hiring,it also creates major counterincentives for them to avoid quotas. UNDERSTANDINGS SOCIOLOGICALVERSUSLIBERAL-LEGAL OF DISCRIMINATION Most sociologists are familiarwith at least the broad outlines of their discipline's theory and research on labor market discrimination.Nonetheless, these are worth underscoring briefly, to contrast institutional-sociological understandings with the liberal-legal concept of discriminationand to show how disparate impact is consistent with the latter but at odds with the former. For sociologists, discriminationis an inherently group-level phenomenon. It is behavior that disadvantages members of minority groups relative to members of majority groups because of their group membership.Sociologists have long understood that discrimination involves a "'system of social relations' not an isolated individual act" (Antonovky 1960; quoted in Simpson and Yinger 1985, p. 23). Similarly, they have understood that although discriminationmay be motivated by prejudice,discriminatory acts may stem not from racial animus but from action oriented to, or constrained by, institutionalized behavior patterns (Merton 1976).-Diverse social settings, practices,and institutions facilitate or inhibit the growth and expression of unintentional cognitive biases, racial and gender stereotypes, and race and gender prejudice and hostility (e.g., Simpson and Yinger 1985; Kirschenman and Neckerman 1991; Ridgeway 1991; Baron 1992; Reskin 1998; 2000). Institutionalized practices often perpetuate discriminatory patterns established in the past even when race or gender animus is absent (Baron 1992; Reskin 2000). One simple example involves an employer recruiting through referrals by current employees. If an employer's work force is all white due to past employment discrimination, and current employees' relatives, friends, and neighbors are likewise all white due to past and continuing trends toward endogamy and residential segregation, then

16

THE SOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

recruitment through employee referrals will discriminate against minorities.This is so even though the recruitment practice is intended not to discriminate but to produce qualified employees (Kirschenman and Neckerman 1991; Holzer 1996). Without conscious societal intervention to alter such discriminatorypatterns,results of past discrimination in some institutional spheres (e.g., schools) continue in these spheres and also reverberate to other institutional spheres (e.g., the job market, because minorities lack educational credentials), contributing to what Peter Blau and Otis Dudley Duncan (1967, p. 209) over thirty years ago referred to as "cumulative disadvantage."This creates inequalities of reward for earlier generations that prevent later generations from achieving equality of opportunity.Similarly,discriminatorylabor market treatment for past generations may cause women and minorities to underinvest in their own human capital,perpetuating disadvantage across the generations (Lundberg 1989). From the idea that discriminationis a group-level phenomenon and the probability that discriminationexists absent racial animus on the part of some who practice it, it is not a long leap to the concept of institutional racism.This idea figured prominently in discourse critical of American society in the late 1960s to early 1970s (e.g., Carmichael and Hamilton 1967;Knowles and Prewitt 1969;Smith 1995). Black activists Stokely Carmichaeland Charles Hamilton (1967) defined institutional racism in contrast to what they termed individual racism. Individual racism consists of "overt acts by individuals,which cause death, injury,or the violent destruction of property."Institutional racism is "acts by the total white community against the black community" (Carmichael and Hamilton 1967, pp. 3-4). Institutional racism came from "the operation of established and respected forces in the society" and was "less overt [than individual racism but] far more subtle, less identifiable in terms of specific individuals committing the acts [but] no less destructive of human life" (ibid). Carmichael and Hamilton (1967, pp. 34-35) believed that any solution to institutional racism must involve revolutionary challenges to what they viewed as an oppressive, colonial society. Without embracing such an extreme view of the problem or solution, many sociologists found the concept of institutional racism, while sometimes imprecise and overreaching, nonetheless useful to signal discrimination so embedded in society's fabric due to historical legacies that racial animus was unnecessary to sustain it (e.g., Willie 1974; Bullock and Rogers 1976; Feagan 1977; Prager 1982). If this type of structural or systemic discrimination were not aggressively attacked, it would perpetuate inequalities stemming from the earlier, state-promoted second-class citizenship for African Americans that had been undergirded by an explicitly racist ideology of white supremacy. Similar to sociology's focus on perpetuation of racial inequality through institutionalized practices, sociologists have researched how organizations perpetuate gender inequality in the job market, absent an explicitly antifemale bias of individual actors (Bielby and Baron 1986; Steinberg 1990; Baron 1992; Reskin 2000). Their work illuminates how built-in institutional biases negatively affect contemporary job assignment and pay of women and minorities due to the powerful role played by both historical patterns of race and gender segregation and historical-culturalperceptions of race and gender in defining job classifications and evaluating job worth (Baron 1992). A sociological concept of discrimination as a group-level, institutionally patterned phenomenon is conducive to statistical models of possible or probable discriminationin schools, firms, and labor markets. As any graduate student in labor market sociology

DisparateImpactand the Quota Debates

17

knows, early regression models of race and gender inequalities in occupational and income attainments included as many nondiscriminatoryreasons for these differences as possible, so that remainingdifferences in direct effects of race and gender on rewards could be interpreted as possible discrimination (Blau and Duncan 1967, pp. 207-241; Featherman and Hauser 1976). In models recognizing interaction effects, differences in returns to individual-level human capital variables for blacks and whites, or men and women, were interpreted as signaling discrimination (Blau and Duncan 1967, pp. 211, 226; Featherman and Hauser 1976;Halaby 1979), as were differences in income returns to occupational attainments controlling for human capital (Featherman and Hauser 1976). Once sociologists recognized gender-segregated labor markets as a major cause of the gender gap in wages, they specified models including human capital and other pay-related job requirements,so that any remaining direct relationship between the sex composition of a job and its earnings could be interpreted as wage discrimination (Rosenbaum 1985;England 1992, p. 101). Sociologists also have found that a job's pay is inversely related to the percentage of minority job incumbents,controlling for job content and job-requisite human capital (Baron and Newman 1989;1990;Baron 1992). Thus, at the same time that sociologists tried to measure and model effects of those factors, such as ability or educational attainment, that they presumed would capture meritocratic employment practices,they tended to assume that remaining direct effects of ascriptive factors on attainments probably represented discrimination.Labor market sociologists also generally assumed that differences in returns to human capital by race or sex represented discrimination,as they assumed that differences in job pay associated with the race or gender composition of jobs, net of job-required human capital, constituted discrimination. In assessing economists' concept of statistical discrimination, in which employers allocate some jobs or pay levels to men or whites and others to women or minorities because of differences in skill or turnover costs, sociologists found that employers do systematically assign men and women to different jobs and do use race as a crude,inexpensive proxy for presumed differences in employee qualificationsand productivity (Bielby and Baron 1986; Kirschenman and Neckerman 1991; Bills 1992; Holzer 1996).1 Thus, both at Title VII's implementation and today, sociologists focused on institutional subordination of minority to majority groups. In contrast, the major legal interpretation of nondiscriminationas a constitutional and statutory right circa 1964 focused on deliberately chosen discriminatoryacts of some individuals against other individuals (Blumrosen 1972; Graham 1990). The Civil Rights Act of 1964 did not explicitly define employment discrimination,but both those who favored and those who abhorred later interpretation of Title VII agree that it saw discrimination as acts disadvantaging individuals on account of immutable factors such as race and gender (Blumrosen 1972;Belz 1991;Epstein 1992;Donahue 1994). Title VII's legislative history and its enactment revolved predominantly around understandings of antidiscriminationas the principle of equal treatmentof individuals. For example, section 703(a) made it an unlawful practice to "fail or refuse to hire or to discharge any individual, or to otherwise discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment because of such an individual's race, color, religion, sex, or national origin" (U.S. Code, vol. 42, Section 2000e-2[a], emphases added). If this provision were taken literally, violating Title VII required proof of discriminatoryintent, since the words "because of" made race and

18

THESOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

other stated characteristicsillegal motivating factors for behavior (Sovern 1966;Epstein 1992, pp. 184-186). Similarly,employer discrimination against union members under earlier federal labor law always had been interpreted to require proof of discriminatory motive (Gorman 1977, pp. 137-139). Congressional debates prior to Title VII's approval provide substantial evidence that Congress understood race discriminationin employment to be intentionalunequal treatmentof individualsbased on race.2For example,SenatorsJosephS. Clark(D-Pennsylvania) and Clifford P.Case (R-New Jersey),bipartisancosponsors of Title VII, explained: It has been suggestedthat the conceptof discrimination is vague.In fact it is clear and simpleand has no hiddenmeaning.To discriminate is to makea distinction,to make a difference in treatment ... based on any of the five forbidden criteria:race ...3

Other statements indicate that members of Congress understood discriminatory unequal treatment to involve bad motive. Representative Emanuel Celler (D-New York) stated: [TitleVII] involvesa questionas to whetheror not there has been discrimination based upon color .... If it can be said that she is qualified and that the employer

refusedto accepther becauseof the colorof her skin,then therewould deliberately be discrimination coveredby thisact.(Cong.Rec.,88thCong.,1stSess.,1964,vol. 110, pt. 2:2579,emphasisadded) That discrimination required intent was made explicit in the remedies section of Title VII. Section 706(g) states that if the court finds an employer to have "intentionally engaged" in an unlawful employment practice, the court could enjoin the action and order such other equitable relief as it found appropriate.4 Debate over Title VII's implications for employment testing and seniority systems shows that even when contemplating recognized institutional practices,Congress envisioned a legal concept of discrimination that required individual intent. When some members of Congress worried that employers without discriminatory motives would be prohibited from acting on employment tests if minorities did not test as well as whites, or that Title VII might disrupt existing workplace seniority systems, Title VII supporters assured them that employers would be permitted to use any selection device they wanted, so long as they did not design or intend the device to discriminate (e.g., Cong. Rec., 88th Cong., 1st Sess., 1964, vol. 110, pt. 10: 13492-13595, 13076, Graham 1990, pp. 139-141; Clark's Statement in Cong. Rec., 88th Cong., 1st Sess., 1964, vol. 110, pt. 7: 9107). An interpretive memorandum by bipartisancosponsors Clark and Case stated: in TitleVII thatemployersabandonbonafidequalification [T]hereis no requirement tests ... An employermay set his qualifications as high as he likes,he may test to determinewhichapplicantshave these qualifications, and he may hire,assign,and on basis of test the 88th Cong.,1st.Sess.,1964, Rec., promote performance. (Cong. vol. 110,pt.6:7213) The Clark-Case memorandum also assured that seniority systems would be fully pro-

DisparateImpactand the Quota Debates

19

tected as long as they did not intentionally discriminate (Cong. Rec., 88th Cong., 1st Sess., 1964, vol. 110, pt. 6: 7213). Section 703(h) codified these assurances.5 Concern about employer selection criteria and seniority systems pointed to the broader issue of whether employers or unions would be found liable for discrimination if, as a result of their testing procedures or seniority rules, there continued to be significant racial disparities in the workforce. Planting seeds for our 1990s political debates, skeptics in Congress feared minority under-representationwould be equated with unlawful discrimination.Title VII opponents argued that to make up for racial imbalances, employers and unions would be forced to give preferential treatment to blacks by adopting employment quotas (see, e.g., Cong. Rec., 88th Cong., 1st Sess., 1964, vol. 110, pt. 2: 2571, pt. 6: 8500-8501). House supporter Charles E. Goodell (R-New York) responded: Thereis nothinghere as a matterof legislativehistorythatwouldrequireracialbalancing... We arenot talkingabout..,.an employerhavingto balancethe numberof employees.... Thereis no quotainvolved.It is a matterof an individual'srightshaving beenviolated.(Cong.Rec.,88thCong.,1st Sess.,1964,vol. 110,pt.2:2558) Similarly,Congressman Joseph G. Minish indicated that "no quota system will be set up.... [Title VII] is designed to ... permit every worker to hold the best job for which he is qualified" (Cong. Rec., 88th Cong., 1st Sess., 1964, vol. 110, pt. 2: 1600). Senator Hubert Humphrey insisted: Contraryto the allegationsof someopponents... thereis nothingin [TitleVII] that will giveanypowerto the [EEOC]or to anycourtto requirehiring..,.or promotion of employeesin orderto meeta racial"quota"or to achievea certainracialbalance." (Cong.Rec.,88thCong.,1stSess.,1964,vol. 110,pt. 5:6549) The Clark-Case memorandum maintained that "any deliberate attempt to maintain a racial balance ... would involve a violation of Title VII because maintaining a balance would require an employer to hire or to refuse to hire on the basis of race" (Cong. Rec., 88th Cong., 1st Sess., 1964, vol. 110, pt. 6: 7213). Title VII codified antiquota sentiment in Section 7030(j).6 This contrast between institutional-sociologicaland liberal-legal concepts of discrimination is not made to suggest that all sociologists embraced a group-oriented,institutional practice concept of discrimination,while all legal scholars embraced an individualistic, intent-oriented concept. Clearly, sociology made room for varied definitions, albeit stressing group-referenced behavioral results (e.g., Prager 1982; Simpson and Yinger 1985). Meanwhile, much of Title VII enforcement revolved around contesting legal definitions of discrimination.A definition that simultaneously focused on intent and individual perpetrators and victims marked only one of Title VII's major enforcement routes (e.g., Player 1988; Graham 1990;Pedriana and Stryker 1997). The following section describes how, in Title VII enforcement, the institutional-sociological and liberallegal definitions of discriminationcame together to forge three legally accepted causes of action for discrimination,and it summarizes the enforcement niche occupied by sociologists' statistical techniques. This paves the way for showing how disparate impact relates to affirmativeaction, as well as what incentives disparate impact creates toward and away from employment quotas.

20

THESOCIOLOGICAL Vol. 42/No. 1/2001 QUARTERLY WHAT IS DISPARATEIMPACT?HOW DOES IT RELYON NUMBERS? AND STATISTICAL DISPARATEIMPACT,DISPARATE TREATMENT, PROOF OF DISCRIMINATION

Congress initially structuredthe Equal Employment OpportunityCommission (EEOC) on an individual complaint processing model. It was to investigate and, upon a finding that "reasonable cause" existed to believe discriminationcharges were true, try to conciliate claims with employers on a case-by-case basis.If conciliation failed, the individual complainant (but until the 1972 amendments to Title VII, not the EEOC) could bring suit in federal court. Angered at an agency they believed hopelessly weak, civil rights groups, such as the Congress of Racial Equality (CORE) and National Association for the Advancement of Colored People (NAACP), complained about dealing with a systemic problem on a case-by-case basis.They argued that the EEOC should proactively prosecute patterns of discriminationin large companies, industries and unions (Graham 1990, pp. 189-190). An EEOC short on staff and resources agreed that it would be more efficient and effective to target widespread industrialpractices affecting large numbers of minorities, and it used the concept of discriminationas a group-oriented and institutionalizedphenomenon to promote aggressive Title VII enforcement (Rosen 1974; Blumrosen 1993, pp. 79-80). Agency attorneys sought a principle to address broad employment practices that created negative effects for minorities as a group (Blumrosen 1971; 1972; 1993; Graham 1990, p. 191). This was consistent with a view of legal interpretation that avoids reliance on exact statutory language or narrow construction of congressional intent in favor of broad leeway for expert agencies to act so as to best further the general policies enacted by Congress.The principle invented by EEOC attorneys and civil rights plaintiffs organizations became known as the disparate impact theory of discrimination.Disparate impact embodies the group-referenced and institutional features of sociologists' orientations, and many legal scholars explicitly consider it as a sociological approachto Title VII (Blumrosen 1972;Pedriana and Stryker 1997). Disparate impact theory required showing systematic practices that had a negative effect on minority employment, but it did not require showing that these practices stemmed from bad motives. Disparate impact developed alongside two intent-oriented causes of action for discrimination:one that captured liberal-legal emphasis on the individual and one that recognized systematic behavior patterns promoting inferior treatment of minorities as a group, while maintaining the requirement to show bad intent (Player 1988). The role for statistical evidence in substantiating the three types of discriminationclaims is different. Law can incorporate social science in ways that produce oxymoronic syntheses of formal-legal and scientific-technicalelements. For social scientists and statisticians,statistics do not prove anything. Rather, quantitative analyses are used to test whether hypotheses can be rejected according to a set of statistical conventions. In law,however, statistics do prove things. They do so by providing evidence that legally requires or allows judgments for one side or the other, assuming that the relevant criteria for type and weight of evidence are met. Currentdebates about disparate impact and the role of statistical evidence are rooted in the 1971 Supreme Court decision in Griggs v. Duke Power Company. In Griggs, a unanimous Supreme Court endorsed disparate impactalso known as adverse impact, adverse effects, disproportionate effects, and the effects

DisparateImpactand the Quota Debates

21

test of discrimination. In so doing, the Court provided plaintiffs with a results-based cause of action that was a potent weapon to change employment practices (Norton 1990).

DisparateTreatment By the late 1980s, it was well established that showing discrimination under Title VII other than by disparate impact theory required proving the employer had illegal, discriminatorymotives.7There have been two major types of illegal motive cases: the individual disparate treatment case and the pattern or practice disparate treatment case. Individual plaintiffs often pursue the former. Government suits on behalf of entire classes of individuals and private class-action suits often use the latter, which relies on systematic illegal motivation. Pattern suits also are available to individual plaintiffs. Ordinarily,an individual disparate treatment case will not get either the employee (or potential employee) or the employer enmeshed in statistical evidence, because the case will turn on a single instance of illegal intent. A job applicant establishes an inference of illegal intent and therefore a prima facie case of discriminationby showing four things:(1) he or she belongs to a protected class of people under Title VII (e.g., African Americans, women, etc.), (2) he or she applied for and was qualified for a job for which the employer sought applicants,(3) despite being qualified, he or she was rejected, and (4) the position stayed open and the employer continued to look for applicants with plaintiffs qualifications.If all this is established, the employer will lose unless able to provide evidence that a nondiscriminatoryreason led to plaintiffs rejection. Statistical evidence that the employer's work force is race- or gender-balancedis not considered very useful to rebutting the rejected job applicant's prima facie individual disparate treatment case.The employer's rebuttal burden is of medium difficulty,involving more than just stating a reason but less than providing enough evidence to persuade the court that the motive was legal. After both the plaintiff and defendant have provided evidence for their claims of illegal and legal motive, respectively, the issue of motive or intent is considered joined. The job applicant then bears the burden of persuading the court that the employer's motivation was in fact illegal. He or she can use statistics showing an imbalanced work force or an employer's rejection of a disproportionate number of minority applicants as circumstantialevidence to help establish that whatever nondiscriminatoryreason the employer has tried to establish is only a pretext for illegal discrimination.But, accordingto legal precedent, courts will find nonstatistical evidence,includingdirectevidence of prejudiceagainstminoritiesor evidence of employer inconsistency in applying the proffered legitimate reason, much more persuasive. In short, the individual disparate treatment case is the Title VII cause of action least imbued with an institutional-sociological orientation to discrimination. It is likewise least likely to involve statistical evidence. On the key issue of motive, statistical evidence cannot much help the employer's defense and contributes only marginally to the job applicant'scase. Notwithstanding a similar focus on intent, pattern or practice disparate treatment suits are more sociological. These suits can be very costly to employers. Many are filed by the government or as private class actions, and at least at the appellate level this increases the plaintiffs odds of winning the case (Burstein 1991).8In pattern or practice cases, once a systematic illegal motive is established, it creates an inference that each individual decision is illegally motivated. Statistical evidence cannot constitute the sole

22

THESOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

basis for an inference of systematic illegal motivation, but it provides pattern or practice plaintiffs with resources to help establish their prima facie case. In general, the requisite statistical analysis contrasts the race or gender composition of an employer's workforce with the race or gender composition of a legally appropriate comparison group. Sometimes the race or gender composition of the entire population for the geographical area from which the employer can reasonably be expected to draw a work force is deemed legally adequate.However, normally the comparison data has to be drawn from the realistic labor supply,taking account of such factors as age, health, and needed skills and qualifications,as well as the geographical area constituting the relevant labor market. Sometimes "applicantflow" data comparing percentages of minorities and whites, or men and women, hired during some time period is considered an adjunct to, or substitute for, "snap-shot"comparisons of an employer's workforce and the labor pool from which it is presumed drawn (Player 1988, p. 354). Where the contrast yields a very large raw difference, statistically testing the null hypothesis that the difference occurs by chance may be unneeded. More often, an appropriate statistical test is required. In 1977, in InternationalBrotherhood of Teamstersv. United States and Hazelwood School District v. United States,defendants argued that Title VII's antipreferentialtreatment provision, Section 703(j), prohibited courts from finding discrimination based on contrasting applicant pool or flow data with any comparison group, no matter how the comparison was constructed. In both cases, the Supreme Court rejected such wholesale denial of a probative role for statistical evidence in helping a plaintiff create an inference of systematic illegal motivation. This set the stage for pattern or practice case battles between the social science experts called by plaintiffs and those called by defendants.For these battles, Hazelwood gave employers many statistical resources. For example, the employer's experts can reanalyze any statistics presented by the plaintiffs experts to cast doubt on plaintiffs case. They can argue with the plaintiffs experts about what would be the most probative data, quantitative analyses,and statistical tests, about whether the sample size is sufficiently large to produce reliable inferences, about whether the time frame and labor market delineated by the plaintiff are appropriate and about whether the central relevant factors to compute appropriate labor supply statistics have been considered. Hazelwood made clear that primary responsibility for analyzing statistical data rested with the trial courts.These must determine whether, in light of the plaintiffs data and analyses,those presented by the defendant are sufficiently convincing to prevent the plaintiff from making a prima facie case. Even if the employer's data and analyses do not prevent a prima facie showing of discrimination,they may be sufficient to keep the plaintiff from winning the case, once the issue of motive is joined. For example, if the plaintiff'sexperts used z scores to show that the employer underutilized minorities, the employer could introduce multiple regression to argue that factors other than race provided a more probable explanation for its hiring practices (Player 1988, pp. 352-353). Multiple regression often enters Title VII litigation when the possible disparate treatment involves pay. In battles between statistical experts in pattern or practice cases, employers had a good chance to win, because in these cases, as in individual disparate treatment cases, once the employerpresented statisticalor other evidence of motive, the plaintiff bore the ultimateburdenof persuading the trialcourt by a preponderanceof evidence that discrimination occurred.Since Hazelwood sanctioned a proliferation of legally relevant ways to

Disparate Impactand the Quota Debates

23

present and examine statistical data, it made possible inconsistencies of both social science results and ensuing legal inferences possible. Allocation of the burden of persuasion to the plaintiff, combined with the possibility of inconsistent results, gave employers a major resource to help them win pattern or practice suits.This is especially so because the trial court's inference of legal or illegal employer intent, based in part on statistical evidence, is ordinarily defined as an issue of "fact" rather than "law." In accord with the rules of evidence, it is subject to overturning only if "clearly erroneous" (Player 1988, p. 330). Whereas appellate courts are higher authorities on matters of law, trial courts hear evidence and are entitled to deference on matters of fact. In Bazemore v. Friday (1986), the Supreme Court indicated that while it did not expect the plaintiff to prove discriminatorymotive with "scientificcertainty,"it did expect the plaintiff to convince the trial court with "a preponderance of the evidence ... that it is more likely than not that the impermissible discriminationexists." In sum, if statistical evidence plays a negligible role in the nonsociological individual disparate treatment case, it plays an important role for both sides in the more sociological pattern or practice suits. Since the key issue in pattern or practice suits is one of systematic illegal motivation, statistical evidence provides key circumstantial evidence for a plaintiffs prima facie case. Still ordinarily,statistical evidence alone cannot make a plaintiff's prima facie's case, and statistics also give an employer a key resource to help it prevent a plaintiff from making a prima facie case or to rebut the prima facie case. Once the issue of motive is joined, allocation of burdens of persuasion to the plaintiff, combined with a legally sanctioned proliferation of ways that social science evidence can be presented and analyzed, means that the employer gets many statistical resources to help it mount a successful defense.

DisparateImpact Disparate impact cases involve much the same statistical evidence as do pattern or practice disparate treatment cases. However, prior to WardsCove, the resource value of statistical evidence in disparateimpact cases was substantiallymore skewed to the plaintiffs benefit than it was in pattern or practice cases. The Griggs disparate impact model focused on two elements: adverse impact and business necessity. The plaintiff bore the initial burden of proving that a particularselection device or system adversely affected the employment opportunities of African Americans, women, or other protected classes of people. Successful establishment of adverse impact made out a prima facie case under Griggs. Then the burden shifted to the employer to establish to a trial court'ssatisfaction that use of the at-issue selection system was a "businessnecessity"(Player 1988, p. 356). Since intent was not an element of the disparate impact case, employers could not justify their selection devices by showing proper motivation. By the late 1980s, disparate impact suits made up less than 5 percent of all employment discrimination cases (Donahue and Siegelman 1991, p. 998n57). This is because cases have shifted away from hiring to involuntary discharge.Even in 1972-1973, in the immediate aftermath of Griggs, disparate impact cases made up only about 9 percent of

all cases filed (Siegelman and Donahue 1990, p. 1164n58). During the 1970s to mid1980s,appellate courts decided a small number of disparate impact cases each year, and plaintiffs won more than 50 percent of these cases only prior to 1975 and in 1977 (Burstein and Edwards 1994).9

24

THESOCIOLOGICAL Vol. 42/No. 1/2001 QUARTERLY

In spite of all these figures,disparate impact is widely credited with promoting wholesale change in employment practices, especially in large manufacturing firms (Culp 1985; Belton 1990; Norton 1990). As John Donahue and James Heckman (1991) note, given diverse methods and the methodological shortcomings in studies by labor market economists and sociologists, sorting out how equal employment law has affected labor market outcomes is like detective work. The sociologist has to play Sherlock Holmes even before trying to isolate the effect of disparate impact from other aspects of EEO enforcement. Extending Donahue and Heckman's detective work to piece together evidence and make reasoned inferences about the likely employment effects of disparate impact yields the conclusion that disparate impact probably was an effective promoter of labor market change, for a combination of reasons. First, despite the relatively small numbers of disparate impact cases and the post1977 better-than-even chances of employer victory in appellate cases, disparate impact cases tended to be targeted to large, industry-leadingfirms,precisely to have maximum impact (Ornati and Giblin 1975;Smith and Welch 1984, p. 286; Pedriana 1999). Second, like pattern or practice suits,disparate impact cases often are filed by the government or as private class actions (Player 1988). These types of collective mobilization substantially increase the plaintiffs' odds of victory (Burstein and Edwards 1994). They also substantiallyincrease the likely costs of these victories to employers. Third,EEOC targeting of major industrialfirms heightened the visibility of disparate impact suits, as did disparate impact's innovativeness. By departing from emphasis on motivation, disparate impact was both a more sociological jurisprudence and a major legal innovation. It reverberated from employment to other areas of antidiscrimination law, including housing and voting rights (Donahue and Siegelman 1991, p. 998). Business press coverage of 1970s government enforcement strategies highlighted the innovativeness of disparate impact and promoted the perception that disparate impact cases were highly successful, creating enormous financial costs to employers (Rosen 1974; Ornati and Giblin 1975;Farrell 1978). Employers' perception of litigation threat, rather than litigation itself, may be the major mechanism through which Title VII enforcement improves the relative employment and earnings of minorities (Donahue and Heckman 1991).10This would seem especially likely for disparate impact enforcement, given its visibility,publicized success, and potentially large liability costs. For all these reasons, disparate impact created a perceived threat that probably did substantially reshape employers' practices,especially in industry-leadingmanufacturing firms,so as to have a positive effect on minority employment. Legal scholars thus have good reason to consider disparate impact the most important innovation in legal definitions of employment discrimination (Donahue and Siegelman 1991, p. 998; see also Matthies 1976, p. 29; Belton 1990, p. 249). As indicated, disparate impact cases often are government suits or private class actions, but individual plaintiffs can also make disparate impact claims and they can simultaneously pursue disparate impact and disparate treatment claims. Though the type of statistical evidence mobilized is similar in disparate impact and pattern or practice suits, for a disparate impact claim, the statistics must show one of two things.They must show the relative impact that the challenged selection device, such as an educational credential or employment test, would have on a legally appropriatepool of potential applicants. Or, even better, statistics must show the relative impact the device did

DisparateImpactand the Quota Debates

25

have in determining relative percentages of men and women, or minorities and whites, selected for hire or promotion from the employer's own applicant flow." Prior to WardsCove,the Supreme Court said that impact analysis consisted of "statistical evidence showing that an employment practice has the effect of denying members of one race equal access to employment opportunities" (New York TransitAuthority v. Beazer, [1979] quoted in Player 1988, p. 363). The Court also had said that objective tests that "selected applicants for hire or promotion in a racial pattern significantlydifferent from that of the pool of applicants"constituted disparate impact (Albemarle Paper Co. v. Moody, [1975], quoted in Player 1988, p. 363). Beyond this, the Court had not defined how much impact was required for plaintiffs to make a prima facie case. But the 1978 Uniform Guidelines on Employee Selection Procedures adhered to by federal antidiscriminationagencies said that, when a plaintiff used applicantflow data (which the guidelines preferred), selection devices presumptively had an adverse impact if the selection rate for any race, gender, or ethnicity was less than four fifths of the selection rate for another group (Code of FederalRegulation,29 pt. 1607.4).This assumed that the applicant pool was large enough for a reliable and statisticallysignificantpattern to be established.12 The Supreme Court in Beazer made clear that employers could undermine plaintiffs' disparate impact cases with expert testimony of their own and showed that, at least in some cases, employers would be able to defeat prima facie showings of adverse impact by outlining multiple, cumulative defects in a plaintiffs data and analyses. Still, the power-resourceimport of statisticalproof of discriminationis different across disparate impact and pattern or practice cases, to the great benefitof the disparateimpact relative to the disparate treatmentplaintiffs. First, since statistical evidence in impact cases is used not as circumstantialevidence of motive but as direct evidence of effect, statistical evidence alone can make a successful prima facie showing of disparate impact, where it usually cannot alone make a prima facie showing of disparate treatment. Second, prior to WardsCove, assuming that a plaintiff succeeded in making a prima facie impact case, the employer's rebuttal burden was much harder to meet in disparate impact than in disparate treatment suits.Because a prima facie showing of adverse impact shifted the burden to the employer to prove the business necessity of the selection device, the employer had to do more than present rebuttal evidence as in the disparate treatment context. In disparateimpactcases,the employerhad to persuade trial courts thatthere was a "manifestrelationship"between a selection device shown to have adverse impact and a bona fide, significant business purpose (Griggs v. Duke Power [1971] p. 432). If the employer met this burden of persuasion, the plaintiff could present evidence that there were alternative selection devices that could have served the employer's purpose equally well without adverse impact. Evidence of such a "lesser discriminatoryalternative" shifted the burden back to the employer to show that the device used was "necessary"for the employer to conduct business safely and efficiently (Player 1988, p. 368). At the time of Wards Cove, the bulk of legal precedent made the employer'sburden to prove business necessity in disparateimpact cases very hard and potentially very costly to meet. To show the business necessity of any kind of objective test, Albemarle Paper Company v. Moody (1975) had required that the test be "validated according to the professionally developed test validation standards [that] require a statistical correlation between test scores and actual job performance" (Player 1988, pp. 367-368). The 1978 Uniform Guidelines set tough standardsfor employer validation studies, as had the earlier EEOC testing guidelines to which the Supreme Court deferred in Albemarle.

26

THESOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

Employers complying with the Uniform Guidelines could be certain their tests were legal, whereas failure to comply made it hard-but not impossible-to show legality. The Beazer case (1979) interpreted business necessity in a way that lessened the burden of persuasion on the employer in that particularcase, but the case did not involve testing and was one in which plaintiffs evidence was weak.13 In sum, if statistical evidence played a minor role in individual disparate treatment cases, law provided a much larger role for statistical evidence in the more sociological, more likely won, and more consequential pattern or practice disparate treatment suits. Yet even in pattern or practice cases, plaintiffs could not use statistical evidence alone to prove that employers had illegal intent. In addition, in pattern and practice cases, employers had many statistical resources to cast doubt on the sufficiency of plaintiffs analyses,and the burdensof proof and persuasionwere allocated to make the employer's task easier. In the most sociological type of case--disparate impact-intent was not an issue. Once plaintiffs met a reasonably easy to reach prima facie threshold, which they could do based on statistical evidence alone, the plaintiffs bore no continued burden of persuasion regarding impact. Unlike the pattern or practice disparate treatment case, inconsistencies in data or results showing impact could not benefit employers responding to disparate impact claims, once plaintiffs had made their prima facie case. Instead, the entirely new issue of business necessity arose;a new kind of scientificevidence, drawn from the psychometrics of testing, became relevant; and the burden of proof shifted to employers. The same business press coverage that publicized disparate impact as an aggressive,effective enforcement strategy also highlighted the role of burden shifting in promoting this success.For example, a 1978 article in PersonnelAdministratornoted: The shiftingof the burdenmadeit easy for the EEOCand privateplaintiffsto win class actionTitle VII suits.As a directresult,many employersfound themselves losingTitleVII cases and being saddledwith huge backpayawardsin cases based strictlyon statisticsPlaintiffs'lawyerswere findingit easy and lucrativeto file and settleTitleVII suits.Thesituationformanyemployerswasaptlydescribedby a companyattorneywho said:"It'snot a questionof whetheryou'regoingto win or lose 1978,p.56) yourTitleVII suit,the questionis howbadlyyou'regoingto lose."(Farrell Given the tough legal standardsfor validating tests and other selection devices set by the courts and EEOC guidelines, proving the business necessity of a selection system resulting in adverse impact was difficult and costly for employers. In conjunction with the potentially large liability costs of disparate impact suits, the difficulty and cost of proving business necessity supports the earlier conclusion that many employers probably did change their behavior to prevent future disparate impact lawsuits.But knowing whether employer behavioral changes involved adopting employment quotas requires directly examining the legal and practical relationship between disparate impact and other methods of proving discriminationagainst women and minorities and affirmative action programsintended to benefitthem. DISPARATEIMPACT,STATISTICAL PROOF OF DISCRIMINATION, AND AFFIRMATIVE ACTION Prior to WardsCove, affirmativeaction had been institutionalized in law and was widely practiced in some form, in both public and private sectors (e.g., Edelman 1992;Pedriana

DisparateImpactand the Quota Debates

27

and Stryker 1997). The predominant legal push for affirmative action was not Title VII at all but instead federal government contract compliance. Under Executive Order 11246, the Labor Department issued affirmativeaction regulations that required federal contractors to specify goals and timetables for minority and female hiring for any job category for which employer "utilization analysis" showed an "underutilization"of minorities and women compared to their availability in the workforce. Employers then had to make a "good faith effort" to meet these goals (Pedriana and Stryker 1997).14 Under Title VII, affirmativeaction plans appeared in two guises:as court-ordered remedies following adjudicativefindings of discriminationand as voluntary plans invoked as a defense to claims of reverse discrimination.Both are relevant to the WardsCove argument that disparate impact leads to employment quotas. With respect to court-ordered remedies, the Justice Department under Ronald Reagan argued that courts could not impose any kind of racial goal or timetable favoring those who had not been specific victims of illegal conduct (Days 1984). But the Supreme Court rebuffed this argument.The Court ruled that trial courts could order numerical goals and timetables where they found "widespread, systematic and egregious law violations," as long as the goals and timetables did not compel the hiring of unqualified employees or "unnecessarily"infringe upon interests of white employees (Player 1988, p. 31, discussing Sheet Metal Workersv. EEOC [1986] and United States v. Paradise [1987]). Court findings of "mere illegal conduct" or of statistical imbalance between the employer's workforce and the area population were not sufficient for courtordered remedial affirmative action to be justified under Title VII or equal protection (Player 1988, p. 312). Thus, pattern or practice disparate treatment would seem more likely to lead to court-imposed affirmative action than would findings of individual disparate treatment or even of disparate impact (Player 1988, p. 312). Similarly,to pass constitutional muster, public employer affirmative action had to be justified by past egregious, systematic discrimination. Because voluntary plans by private employers lacked the element of "state interest" making public employment and court-ordered remedies subject to constitutional protection, private employer voluntary affirmativeaction was not subject to the "systematic and egregious illegality" threshold. Instead, the EEOC's 1979 Affirmative Action Guidelines indicated that such voluntary plans would be a defense to a charge of otherwise illegal discrimination(Code of FederalRegulations,29 pt. 1608). In an environment in which some affirmativeaction plans adopted to conciliate Title VII claims or to satisfy the Office of Federal Contract Compliance had been challenged as reverse discrimination, the guidelines had the express purpose of carving out a secure niche for businesses to adopt affirmative action in response to prior institutionalized discrimination against racial minorities and women (Farrell1978; Code of FederalRegulations 29, pt. 1608.1). In the early reverse discrimination cases, both the reverse discrimination claim and the voluntary affirmative action defense were based on Title VII. Reverse discrimination plaintiffs argued that Title VII's antipreferential treatment language precluded plans with any sort of minority hiring goal or ratio. Affirmative action employers, the EEOC, and the OFCC argued that Congress intended Title VII to encourage voluntary affirmativeaction to make up for past discriminationwithout the necessity of litigation. Federal courts addressing contract compliance already had indicated that Title VII did not prohibit the government from requiringits contractors to make a good faith effort to meet specified minorityhiring goals and timetables,and the EEOC wanted to ensure that

28

THESOCIOLOGICAL Vol. 42/No. 1/2001 QUARTERLY

any employer, whether contractor or not, could be certain that Title VII supported voluntary affirmativeaction. So the EEOC guidelines said that an employer whose affirmative action plan was challenged as reverse discriminationwould have what was termed a "good faith reliance defense" under Section 713(b) of Title VII.15The EEOC guidelines encouraged voluntary affirmative action programswhen an employer's mandated "self analyses"showed practices tending to result in disparate treatment or disparate impact and also when it appeared that the effects of prior discrimination had not been corrected. Affirmative action employers were not required to admit past Title VII violations, but they had to document underrepresentationdue to past discrimination(Code of FederalRegulations 29 pts. 1608.3, 1608.4). In 1980, the Supreme Court went so far as to make clear that a private employer adopting race-based affirmative action quotas in training programs designed to reach percentage hiring goals according to specific timetables did not violate Title VII, as long as these quotas were voluntarily imposed, temporary, and enacted to remedy the employer's self-recognized past exclusion of blacks (Weber v. Kaiser Aluminum). In so doing, the Court held that Section 706(j), stating that Title VII is not intended to require preferential treatment to counter racial imbalance, nonetheless permits employers to make voluntary adjustments to remedy their own past discrimination.'6Weber also essentially adopted the EEOC guidelines stating that affirmativeaction programscomplying with Executive Order 11246 do not violate Title VII, and the guidelines themselves allowed plans that incorporated OFCC standards,including goals and timetables. Then, in 1987, in Johnson v. TransportationAgency, the Supreme Court relied on Weber and the EEOC guidelines to hold that a valid affirmativeaction defense to reverse discrimination need not show even "arguable"past disparate impact or disparate treatment (Player 1988, p. 318). The defense would only have to point to "conspicuous" imbalances of males and females or minorities and whites in traditionallysegregatedjob categories (Johnson v. TransportationAgency [1987] p. 630). This could be done by comparing the race and gender of the employer's workforce with (1) the race and gender of the general population if no special expertise or training were needed for the job or (2) the race and gender of the area workforce with needed qualifications,if special expertise or training were required.As of the mid-1980s,the only major Court restrictions on voluntary affirmativeaction in employment were that affirmativeaction could not use gender or race to determine the order of layoffs or to discharge white males to make room for females or minorities (Player 1988). In sum, by the time of WardsCove, the legal relationship between Title VII suits and affirmativeaction was as follows. On the one hand, Court-imposedremedial affirmative action required that the adjudicatedTitle VII violation be systematic and egregious,suggesting that pattern or practice disparate treatment cases were the likeliest candidates for court-imposed affirmativeaction plans,including numericalgoals and timetables.On the other hand, Executive Order 11246 affirmative action, as well as any affirmative action plan that a business entered into pursuantto a court-endorsedEEOC conciliation agreement and consent decree required only that there be prior exclusion or conspicuous imbalance of women or minorities in particularjob categories.This meant that utilization analyses alone could provide a basis for employers' voluntary affirmativeaction, not just when statistics suggested past disparate impact or disparate treatment, but also-and even-when discovered statistical imbalances did not meet the threshold of

DisparateImpactand the Quota Debates

29

arguableTitle VII violation. Nowhere, however, did courts endorse hiring or promoting unqualified employees to implement affirmative action. Indeed, the Johnson case suggested this was prohibited. DOES DISPARATEIMPACTMEAN EMPLOYMENT QUOTAS? So what is the upshot of our foray into equal employment law for the question with which we began: Is disparate impact tantamount to mandating employment quotas? We now have seen how and why disparate impact provides greater legal resources for minorities and women to secure positive labor market outcomes than do other methods of proving they have suffered discrimination. We know how and why the positive resource value of statistical proof for women and minorities bringing disparate impact suits is especially high relative to other methods of establishing illegal discrimination. We also know that, although court-ordered remedial affirmative action has little or nothing to do with disparate impact, perceived threat of disparate impact lawsuits creates some incentives for employers voluntarily to adopt affirmativeaction plans to help insulate themselves from Title VII liability for discriminatingagainst women and minorities. At the same time, not only did employers with affirmative action plans not run much risk of liability for reverse discrimination,they could actually point to the affirmative action plans they adopted to achieve contract compliance or to remedy prior race or gender imbalances in their workforces to insulate themselves from liability for reverse discrimination.The personnel literature publicized this information,touting affirmative action, including goals and timetables, as protection from Title VII liability for any kind of systematic discrimination,including disparate impact (Chayes 1974). In short, it is clear that disparate impact law, combined with Court deference to the EEOC guidelines, provided legal and practical incentives encouraging employers to do statistical analyses of the race and gender composition of their employees and of their applicant flows relative to the race and gender composition of relevant comparison groups.Disparate impact and pattern or practice law,combined with the guidelines, also provided incentives for employers to adopt voluntary affirmativeaction, including goals and timetables (Seligman 1973; Chayes 1974; Ornati and Giblin 1975; Fisher 1985; Vernon-Gerstenfeld and Burke 1985;Noble 1986; Kahlenberg 1996). This is so whether employers were federal contractors or not.17 A mid-1980s study of affirmative action at nine companies with the reputation for aggressive, effective programs found that most of these companies engaged in major outreach in education, training, and recruitment.They also set specific goals and timetables for minority hiring and monitored their performance in reaching them. Some of the companies provided financial bonuses to managers who achieved positive affirmative action performance (Vernon-Gerstenfeld and Burke 1985). Where employers in the Susan Vernon-Gerstenfeld and Edmund Burke (1985) study refused to provide information on the demographic composition of their workforce, J. Edward Kellough's (1989) study of employment outcomes in two government agencies showed that the increased presence of goals and timetables in agency affirmative action programs positively affected agency equal employment performance. Barbara Reskin's (1998) review of the literature on what makes affirmative action programs effective likewise stresses goals and timetables, combined with monitoring of, and rewards for, their achievement.

30

THESOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

But does all this suggest that disparate impact promoted widespread use of rigid race and gender employment quotas as suggested by the Wards Cove Court and by opponents of the 1990-1991 civil rights acts? We lack evidence from systematic qualitative studies of how affirmative action programs operate for representative samples of employers, and we lack quantitative studies, on representativesamples, that directly relate variation in employers' workforce composition to variation in their hiring and promotion responses to the EEO legal environment. So we are back to playing Sherlock Holmes, piecing together evidence that is relevant albeit imperfect.There is no smoking gun one way or the other. But we can piece together relevant empiricalresults and filter them through a framework that views Title VII enforcement as part of an overarching logic of competing incentivesto which employers in the 1970s and 1980s had to respond. If we do this, we see that disparate impact does not equate with employment quotas.

RelevantEmpiricalResults There are two broad types of studies that bear on the issue of whether disparate impact is tantamount to race and gender quotas in employment: (1) research on the incidence of preferential treatment for minorities or women and (2) research on employer responses to changing legal environments. First, a few studies find evidence of preferential treatment in wages for subsets of highly educated blacks, such as college professors or well-educated young African Americans, for a short time span in the late 1960s and early 1970s (Sowell 1976;Smith and Welch 1989; Jencks 1992). Likewise, Reskin's (1998) report on affirmative action signals a small number of cases in which courts have found employers guilty of reverse discrimination. A report by George Stephanopoulous and Christopher Edley (1995) found that a few federal contractors also had engaged in reverse discrimination. But as Paul Burstein (1993) points out, any widespread preferential treatment of minorities would be likely to show up in large aggregate benefits in minority employment and earnings attributable to Title VII and Executive Order 11246 affirmative action. The positive effects of Executive Order 11246 and Title VII on minority labor market outcomes are sufficient to attribute effectiveness to EEO law, but they are too modest to warrantany inference of pervasive preferential treatment,let alone pervasive quota hiring or promotion (Rosenbaum 1985; Leonard 1986; Kellough 1989; Donahue and Heckman 1991;Burstein 1993; Holzer 1996, pp. 152-155; Reskin 1998).18Similarly, Jonathan S. Leonard (1990, p. 54) argues that documented employment gains for women and minorities in firms subjected to federal contract compliance review would have been much greater if goals and timetables were merely "expedient and polite word[s] for inflexible quota[s]."These observations are consistent with Alfred W. Blumrosen's review of discrimination complaints filed with the EEOC. Blumrosen's report, discussed in Reskin (1998) suggested that reverse discriminationis rare. Complaints of reverse discriminationagainst whites are less likely than complaints of conventional discrimination to be judged credible by the EEOC (Reskin 1998). In addition, to the extent that methodologically sound studies have found discrimination against whites in employment, they have found discriminationagainst minorities to be still more prevalent. A 1990 Urban Institute field experiment matched black and white "testers"for entry-level jobs in Washington,D.C. and Chicago on hiring-relevant characteristics including education, experience, articulateness, physical size, age, and

DisparateImpactand the Quota Debates

31

energy level; the testers were trained to act similarly during the job application and interview process (Turner, Fix, and Struyk 1991). Black job applicants received job offers when their white counterparts did not 5 percent of the time, and they advanced farther in the application process than did their white counterparts7 percent of the time. But the white job applicantsreceived job offers when their black counterpart did not 15 percent of the time, and advanced farther than their black counterparts 25 percent of the time (ibid., p. 2). The Urban Institute study does not tell us about the equal employment procedures of the tested firms,so we do not know whether concerns about possible disparate impact liability figured into employers' preferential hiring of blacks in the five percent of cases where it occurred. However, since three-fourths of the preferential hiring favored whites in situations where the alternative was an equally welleducated and experienced minority candidate, disparate impact does not appear to force minority hiring even in those situations that arguably most favor it-that is, situations where minority and white candidates' qualifications are equal. Extrapolating from these results, it hardly seems reasonable to assume widespread quota hiring of underqualified minorities and women.19 Results of the controlled Urban Institute study square with the findings of Lawrence Bobo and Susan A. Suh's (1996) survey of Los Angeles employees in which a far greater percentage of African Americans and Latinos than of whites reported that they had experienced discrimination. Second, the accumulating results of studies of employer responses to their changed legal environment suggest that OFCC and Title VII pressures promoting affirmative action, including disparate impact, were internalized in ways that were likely to make strict race- and gender-based quotas the exception rather than the rule. Indeed, the purposely unrepresentative businesses studied by Vernon-Gerstenfeld and Burke (1985) and the government agencies studied by Kellough (1989) may indicate that the ordinary outer limits of aggressive affirmative action are closer to race- and gender-conscious hiring goals than to strict quotas. James N. Baron (1992) reminds us that, in general, antidiscriminationreforms will be limited, because employers are likely to respond to external pressure for change in ways that minimize disruptions to their preestablished patterns.Thus, organizational studies show more evidence of cosmetic reform. One example is discontinuing explicit restrictions on women's employment in physically demanding jobs rather than discontinuing the practiceof excludingwomen from such jobs (Baron 1992).JamesP.Smith and Finis R. Welch's (1984) suggestion that employers gained in representation of blacks and women among managers and professionals in part by reclassifyingjobs rather than by enhancing female and minority hiring in the jobs they traditionally saw as managerial and professional fits this "limited response" pattern, though it is not purely cosmetic. After finding widespread institutionalizationof affirmativeaction rules and offices in a systematic sample of public and private sector organizations,Lauren Edelman (1990; 1992) undertook qualitative investigation of a small number of employers to identify whether affirmativeaction's institutionalizationwas primarilysymbolic or was a prelude to major shifts in workplace demographics (see also Edelman, Petterson, Chambliss, and Erlanger 1991;Edelman, Erlanger,and Lande 1993).20 She found four ideal-types of affirmativeaction implementation styles, but two were minimalist and only one involved the affirmative action officer aggressively promoting enhanced employment outcomes for minorities and women. The latter strategy was extremely unstable, because it fos-

32

Vol. 42/No. 1/2001 THESOCIOLOGICAL QUARTERLY

tered employee concerns about reverse discrimination and led to conflict with higher management (see also Chambliss 1989). The most stable strategy for implementing affirmative action was an "in-between" strategy highly touted by personnel journals.Stressing formalized procedures and constructing equal employment as an element of a more general concept of good personnel policy,the "professional"strategyemphasizedaffirmativeaction officer neutralitybetween employers and employees, minorities and whites, and males and females. The professional strategy is far from hiring and promoting strictly according to race and gender numbers.It is, instead, consistent with subsuming legal rights under the broader interest of maintaininggood relations with all employees (Edelman et al. 1993, pp. 497, 511).

CompetingIncentivesfor Business Is disparate impact tantamount to employment quotas? First, we have seen how both Title VII enforcement and the federal contract compliance program created incentives for employers to focus on their race and gender numbers and to create voluntary affirmative action, including specification of goals and timetables, to improve the race and gender balance of their workforce.Second, of all methods of proving discrimination against women and minorities under Title VII, disparate impact is the one that simultaneously created the most legal threat to employers based on plaintiff mobilization of their race and gender numbers, while it also provided minority and female employees with their greatest legal opportunitysuccessfullyto exploit workplacenumbersto improve employment outcomes. Yet the empirical evidence on employer response to their EEO legal environment suggests that widespread hiring and promotion strictly by the race and gender numbers is very unlikely. This apparentlycontradictorystate of affairs makes perfect sense because business is subject to competingincentives.Where some of what pushes employers to adopt affirmative action also creates incentives for them to engage in more inflexible quota hiring and promotion, there are powerful incentives working in the opposite direction, that is, to keep business institutionalization of affirmative action away from hiring strictly by the numbers.21 Pre-WardsCove disparate impact, as it operated in its broader business and legal environment, created competing incentives both toward and away from employment quotas. Creating some pressure toward quotas were the cost and difficulty of validating employee selection procedures found to have disparate impact, while working against quotas were key elements of business values, concerns about employee productivity, turnover and company competitiveness, the role played by validation studies in improving company productivity,competitiveness and profits,and the role of company equal employment policies, including firms'response to disparate impact,in maintaining good employee relations and thereby contributing to lowered employee turnover and increased company productivity,competitiveness, and profits. Costly validation procedures for employee selection devices probably provided the greatest incentive for race- and gender-based quota hiring in the 1970s and 1980s.Meeting the validation requirements that protected against the Griggs disparate impact method of proving discrimination clearly was difficult, time-consuming, and costly for employers (Ruch 1971;Malbin 1973;Pearson 1973;Seligman 1973;Dhanens 1979;Belton 1990). Similarly stringent validation requirements were adopted to implement Executive Order 11246 (Pearson 1973). Some trade publications suggested that because vali-

DisparateImpactand the Quota Debates

33

dation studies conducted on successful employees only were improper, employers who wanted properly to assess their selection devices' ability to predict successful on-the-job performance would end up hiring some marginallyqualified and unqualified employees, at least in the short term. Only then could they show how use of their selection procedures would have validly screened these employees out (Dhanens 1979). Since validated employmentprocedures were requiredunder TitleVII only when a company selection or promotion device adversely affected minorities or women (e.g., Dhanens 1979), it is not surprisingthat the business press suggested that some employers would use ascriptively based hiring quotas rather than engage in costly and time-consuming validation studies (Pearson 1973; Seligman 1973; Hoffman 1986; Brimelow and Spencer 1993). This was especially likely given that, as indicated above, tough validation standardswere coupled with the employer bearing the burden of proving the business necessity of standards found to have disparate impact (Ruch 1971;Abrams 1972;Malbin 1973;Seligman 1973; Farrell1978). Some business press reports cited government discretion in contract compliance enforcement as another impetus to quota hiring, since compliance procedures were sometimes confusing and compliance regulations ambiguous (Pearson 1973; Seligman 1973, 1982). In addition, trade journals advised employers that if they set unrealistic affirmativeaction goals and failed to meet them, it would not constitute a Title VII violation but it might well subject them to close scrutiny and sanctions through contract compliance enforcement (Seligman 1973, 1982; Chayes 1974). These incentives to quotas were not built into disparate impact itself. But they were a part of the broader legal environment that occasioned some business publications to predict that some administrators would "take the easy way out ... by playing the ratio numbers game of hiring more minorities and hoping not to be sanctioned" (Pearson 1973, pp. 24-25). Competing againstthese incentives toward quotas were business interests in employee qualifications and productivity and in good employee relations. While highlighting the costliness of validation studies, business publications were simultaneously stressing the importance of employee qualifications (Mayfield 1964; Ruch 1971; Dhanens 1979; Vernon-Gerstenfeld and Burke 1985; Hoffman 1986; Brimelow and Spencer 1993). Harold Mayfield (1964), for example, cautioned that companies could not overlook the fact that they were competing with other companies and would not do as well if they had less qualified employees. This suggests minimally that an employer tempted, by EEO enforcement including disparate impact, to engage in quota hiring would weigh clearlyperceived disadvantagesagainstthe presumed advantagesoffered by quota hiring. Add to this the 1960s and 1970s business and personnel publications that argued that then-current employment standards failed to ensure qualified employees, so that the key reason for doing validation studies was not to protect against any new legal liability but instead to meet the traditional business interest of ensuring employee productivity. Mayfield (1964) argued that most employment standards then in use amounted to little more than supervisors' personal notions of the kind of employee needed. This resulted in standardsof "social acceptability"rather than in "capabilityto do the job" (Mayfield 1964, p. 16). In the same year as the Griggs decision, Floyd L. Ruch (1971) provided examples and numbers showing that the business savings generated by proper scientific validation of employee selection procedures would likely be impressive. Reminding managers that poor employee selection resulted in lower productivity,greater employee turnover, and greater employee dissatisfaction, Ruch (1971) stressed that increasing

34

THESOCIOLOGICAL Vol. 42/No. 1/2001 QUARTERLY

employers' ability to identify in advance which employees would perform best would lead to huge savings associated with enhanced output and reduced turnover.Since savings associated with proper validation were predicted to outweigh greatly the cost of developing valid employee selection devices, the net result would be that the rational employer would opt for validation rather than for any hiring method, including quotas, which threatened to provide underqualified employees. Similarly, Thomas Dhanens (1979, p. 39) admonished managers not to forget that "before employers were required to find valid selection procedures,leading businesses wanted to do this voluntarilysince it benefited the company." The point is not to accept uncritically what was written in the personnel literature and business press but rather to point out that employers were being made aware,not merely of cost incentives on them to adopt race and gender quotas, but also of cost incentives for them to avoid quotas and engage in validation studies. Adding weight to incentives against using race- and gender-based employment quotas, employers and their personnel managers have as a "major goal" "to achieve and maintain good employee relations, which according to managerial lore, helps to assure efficiency and productivity"(Edelman et al. 1993, p. 511). On the one hand, the need to maintain the good will and commitment of minority and female employees creates a continuing internal incentive for affirmativeaction in a company that responds to its externallegal environment by creating an affirmative action program. On the other hand, firms need to maintain the good will and commitment of their majority group employees, and an affirmative action program too aggressively tilted toward minorities and women will threaten this (Edelman et al. 1991, 1993;Jencks 1992). This may help explain research findings showing that affirmative action goals and timetables ordinarily promise more jobs for all demographic groups, including white males, and that affirmative action is most effective in increasing minority and female employment in firms experiencing employment growth (Leonard 1990). Notwithstanding pre-1989 legal endorsement of most types of employment-related affirmativeaction programs,if and when aggressive affirmativeaction were perceived as reverse discriminationby white male employees, it would be likely to spark resentment and increase workplace conflict, just as neglecting the needs of women and minorities would be likely to do (Jencks 1992). Smooth business functioning requires gaining and maintainingcredibilitywith all employees, as well as balancingthe interests and demands of majority and minority group employees. Thus, the organizational goal of maintaining good employee relations, either for its own sake or because of its presumed contributions to employee productivity and business profits, should provide strong incentives against responding to disparate impact by implementing race- and gender-basedquotas. At the same time, it should work to maintain affirmative action programs,once these are instituted. As shown above, the empirical results of labor market and organizational research are consistent with the broad institutionalization of affirmative action, combined with no widespread use of quotas. In analyses based on 1990 General Social Survey data, Marylee Taylor (1995) found that whites who reported that their employers practiced affirmative action were more, rather than less, likely to support race-targeted social interventions."And they [were] more likely to hold beliefs that provide a rationale for race-targeted remedies-beliefs that discriminationexists, that black Americans do not have their share of societal influence,that black disadvantage derives from systemic fac-

DisparateImpactand the Quota Debates

35

tors and not from personal deficiencies" (Taylor 1995, p. 1406). If race-based quotas were used widely to implement affirmative action in employment, this pattern of employee attitudes would have been unlikely,to put it mildly (Jencks 1992). Finally,like the attitudes expressed in public opinion data more generally during the time period in question, values expressed by managers in large firms showed support for affirmative action mixed with hostility toward quotas.22A 1969 probability sample of executives in the Fortune 750 largest companies found that 100 percent of top management (but only 51 percent of line managers) thought companies should make a special effort to hire and train blacks for skilled jobs (Goeke and Weymar 1969). Fifteen years later, a New York consulting firm's survey of CEOs of large corporations found that 95 percent of the 127 CEOs who responded to a question about "numerical objectives" said that they would continue to use goals and timetables regardless of government requirements (Fisher 1985, p. 28; Noble 1986). But a 1983 Ford Foundation survey of forty-nine government contractorsfound that all opposed quotas and none thought that their company's affirmative action goals and timetables required quotas (Fisher 1985). Spokespersons of large businesses pointed out that quantifying their affirmative action goals and assessing company performance in meeting them did not signify rigid quotas but merely made these goals the same as any other aspect of the employer's business, from human resources to material resources to profits (Seligman 1982, p. 160; Fisher 1985, p. 28). Where an opponent of quotas might suspect that the goal/quota distinction constituted meaningless lip service to broader American values, these suspicions do not square with the results of qualitative and quantitative research on employer behavior and labor market outcomes. Thus, the values of top managers in large firms,which both reflect and motivate their organization's behavior patterns, simultaneously provide incentives for affirmativeaction plans but away from race- and gender-based quotas. If Christopher Jencks (1992, p. 57) is correct that employers prefer their employees to have the smallest number of rights possible, this would create yet another incentive away from quota hiring since the EEO legal environment made it clear that once on the job, minorities had more rights than their white counterparts.Accompanying a general desire that employee rights be at a minimum are more specific concerns about the difficulties of defending againstTitle VII actions for discriminatorydischarge.The latter provide employers with very strong incentives against race or gender-based quota hiring, since they make it harder for employers to fire minority and female employees, than to fire white male employees, who already are on the job but are not performing sufficiently well in management's eyes (Smith and Welch 1984;Jencks 1992). In sum, there is a kernel of truth in rhetoric mobilized by the WardsCove Court and by the early 1990s opponents of disparate impact. But the argument that disparate impact inevitably caused employers to adopt rigid employment quotas is false.23The kernel of truth is that disparate impact did provide incentives for employers to adopt affirmative action programs,to monitor their race and gender numbers, and to be concerned about having to do costly validation studies if their race and gender numbers could be shown to result from selection and promotion criteria having disparate impact. In the absence of competing incentives, these pressures could have pushed toward employment quotas. But there clearly were such competing incentives. In the 1970s and 1980s, disparate impact-created incentives toward employment quotas were countered by powerful incentives working in the other direction. These counter-incentives stemmed in part

36

Vol. 42/No. 1/2001 THESOCIOLOGICAL QUARTERLY

from disparate impact hiring and promotion enforcement itself since personnel journal coverage of it highlighted ways that common employee selection procedures could be invalid and also emphasized the positive association between valid employee selection procedures and employee satisfaction and productivity,and employer profits.The counterincentives also stemmed from other aspects of Title VII law, including the legal impermissibilityof hiring unqualifiedemployees and the increasing prevalence of involuntary discharge suits, the values of top managers in large firms,and the general business concern about maintaining good employee relations, employee productivity,and business competitiveness.Thus,at the same time that disparate impact'stough burden of proof on employers combined with the high cost of validating employee selection procedures created some incentives toward quotas as an easy way out, disparate impact's highlighting of the costs involved in failing to validate coupled with other features of employers' legal and business environment worked strongly against quotas. Consistent with these competing incentives, the empirical evidence points away from any widespread use of employment quotas.

CONCLUSION Further theoretical and empirical work is needed to help establish the weights associated with the various competing incentives identified in this article,including the competing incentives that are inherent in disparate impact enforcement. We also must consider how variation in both the external and internal business environment affects these weights. But extant empirical research suggests that in most situations,incentives against employer reliance on race- and gender-based quotas will be stronger than the incentives promoting quotas. There should be no lingering concern that disparate impact caused employment quotas or that the 1991 civil rights compromise returningto the Griggs disparate impact standardcauses quotas.24 Given that disparate impact is the most institutionally rooted and sociologically informed part of Title VII jurisprudence,and given that statisticsare essential to assessing labor market discrimination,the reinstatement of disparate impact should have pleased sociologists.The 1991 legislative resurrectionof disparateimpact and its emphasis on statistical proof of discriminatoryeffects, rather than on intent to discriminate,helped to maintain sociologists' contribution to Title VII enforcement. The work is never done, however. Court interpretationof disparateimpact and its burdensof proof and persuasion is ongoing. So in spite of the 1991 civil rights legislation,there is plenty of opportunityfor judicial vitiation of disparate impact. This is especially true since the 1991 legislation retained WardsCove'srequirementsthat plaintiffsordinarilyisolate the specific employer practices that create disparate impact and the legislation did not resolve the pre-Wards Cove ambiguities in lower court rulings interpreting business necessity (Player 1999). Thus, sociologists must continue to stress institutional mechanisms of discriminationto the law and public policy communities,as well as to the court of public opinion. Finally,the logic of sociological research on discriminationsuggests that we may wish to go farther still, moving beyond the misplaced quota debates to acknowledge that creating some legal incentives favoring women and minorities in the labor market is a positive good rather than a necessary evil. Surveying the cumulative import of our research alerts us to the myriad ways past discriminationagainst women and minorities reverberates to their disadvantage today, even absent any continuing racial or gender prejudice

DisparateImpactand the Quota Debates

37

(e.g., Blau and Duncan 1967;Lundberg 1989;Baron 1992;Jencks 1992; Oliver and Shapiro 1995). It alerts us to intentional and unintentional employer misuse of qualifications, to the role of race and gender in shaping organizationaljob classificationsand pay, and to the "social construction"of skill and the social cognition and status expectation biases that if left to operate unattended usually lead to underestimating the competences and qualifications of minorities and women (Steinberg 1990; Kirschenman and Neckerman 1991; Ridgeway 1991; Baron 1992; Holzer 1996; Reskin 2000). Likewise, it alerts us to the ways that any legal incentives created to benefit minority and female employees often are intentionally or unintentionally subverted by managers who respond to their EEO legal environment by assimilating antidiscriminationpolicy into prior ideologies of good managerialpractice so as to minimize change in the status quo (Baron 1992;Edelman et al. 1993). Every situation comes with structuralbiases. One key implication of the long shadow cast on life chances by race and gender and of the deep interconnections among race, gender and organizational practices is that race and gender blindness is not a realistic way to get to race and gender neutrality in the workplace. Instead, we should selfconsciously mobilize law to work toward a set of competing institutional incentives that will balance each other out. This is especially so since, as has been the case since the advent of affirmativeaction, employers will continue to have very strong incentives to maintain and enhance the skills, qualifications,productivity,and job satisfaction of their workforces,including both minority and majoritygroup employees. Purposely designing institutional incentives so that some of today's biases work in favor of women and minorities does not mean that employee qualifications,skill, or human capital variables should or would be irrelevant to employee selection procedures.Instead it means that the organization and personnel who are in the best position to know what constitutes qualifications,and on whom good business practice creates incentives to learn what constitutes qualifications,should have to justify their judgments. It also means that those most responsible for the social construction of qualification inside the employing organization should be devoting organizational time, energy, and resources to this issue. Consistent with the cumulative import of labor market and organizational sociology, disparateimpact legal theory tries to accomplish precisely this. Disparate impact is no panacea. But it is absolutely essential to sociologically informed and sound labor market policies. ACKNOWLEDGMENTS I presented an earlier version of this article at meetings of the American Sociological Association, New York,August 16-19, 1996. I thank John Hagan, Sheldon Stryker,Jennifer Glass, Barbara Reskin, Nicholas Pedriana,and especially TSQ editor Kevin Leicht for their helpful comments.The National Science Foundation (SES-9209777), a University of Iowa Faculty Scholarship,and the Obermann Center for Advanced Studies at the University of Iowa provided invaluable support for this research. NOTES 1. Whensuchaggregatedifferencesexist,statisticaldiscrimination maynot be seen as group But the discrimination because,in the aggregate,grouprewardsarebasedon groupqualifications.

38

THESOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

individualwho is underemployed or underpaidrelative to his or her qualificationswill have experienced discrimination,and the explanation for discriminationagainst that individual will be statistical discrimination (Cain 1986; Lundberg 1989; Jencks 1992). Such discrimination may be viewed as economically rational,but it is illegal. 2. Title VII prohibited discriminationon the basis of religion, sex, or national origin as well as race, but early enforcement focused almost exclusively on race. Similarly,much of the legislative debate over Title VII focused on race. Promoting an amendmentto include sex was in large part a "mischievous"tactic used by certainTitle VII opponents to try to delay or derail the bill (Graham 1990, p. 136). Although the amendment occasioned some amusement in Congress, serious congressional proponents of women's rights seized upon it as a useful opportunityand Title VII incorporated sex discrimination. 3. Cong. Rec., 88th Cong., 1st Sess., 1964, vol. 110, pt. 6: 7207, 7213. See also statement of Title VII supporter Senator Hubert Humphrey (D-Minnesota), at Cong. Rec., 88th Cong., 1st Sess., 1964,vol. 110, pt. 4: 5423. 4. U.S.Code,vol. 42, secs. 2000(e)-5(g). To Title VII supporters,this explicit reference to intent simply clarifiedearlier sections of Title VII (Cong. Rec., 88th Cong., 1st Sess., 1964, vol. 110, pt. 10: 12723-12724). 5. It stated that employers could legally apply different terms of employment pursuant to "bona fide seniority or merit system[s]"and that they could "give and act upon the results of any professionally developed ability test provided that such test .., .is not designed, intended or used to discriminatebecause of race."(U.S. Code,vol. 42, secs. 2000[e]-2[h]). 6. It stated that "nothingcontained in this Title shall be interpretedto requireany employer ... to grant preferential treatment to any individual or to any group because of the race ... of such individual or group on account of an imbalance which may exist with respect to the total number or percentage of persons of any race ... employed by any employer ... in comparison with the

total number of percentage of persons of such race..,. in any community."(U.S. Code, vol. 42, secs. 2000[e]-2[j]). 7. Unless otherwise noted, the following discussion of disparate treatment and disparate impact prior to the 1989 Supreme Court term are based on Player (1988). As a legal treatise in West's Hornbook Series, Player (1988) gives scholars and attorneys a detailed, objective overview of the law. I needed such an objective overview of legal interpretation prior to the 1989 Wards Cove decision, so a hornbook with a publication date of 1988 provided an excellent startingpoint. I supplemented it by consulting and analyzingdiverse primarysources, includingthe relevant statutes, case law,and administrativeregulations and guidelines. 8. Burstein's (1991) survey of all appellate equal employment cases other than those of reverse discriminationfrom 1964-1985 showed that, on average, the plaintiff's likelihood of winning was 58 percent. The likelihood of victory decreased slowly over time. Sixteen percent of the appellate cases included government as a party,and 45 percent were class actions. Plaintiffswon 74 percent of the time when government was a party,versus 54 percent of the time when government was not. Plaintiffs also won a substantiallygreater percentage of class actions than of nonclass action suits. 9. Employers may not appeal disparate impact cases they see as hopeless. If appellate cases are those in which employers are most optimistic about winning,it could be that the percentage of plaintiff wins at trial is greater than the percentage of wins at the appellate level. But this would run counter to Siegelman and Donahue's (1990) findings about district court cases. These authors do not report figures for disparate impact cases alone, but they do report that the plaintiff victory rate was 35 percent in district cases with published opinions and 25 percent in district cases without published opinions. On the one hand, both these percentages are less than the 58 percent average plaintiff victory rate reported by Burstein (1991). On the other hand, settlement levels are very high in district court cases without published opinions (68 percent) and substantial (35 percent) in cases with published opinions. We do not know what percentage of the total settled Title VII cases predominantlybenefit plaintiff as opposed to defendant, but most of the argumentsfor

DisparateImpactand the Quota Debates

39

disparateimpact's resource value to minorities and women hinge on the likelihood that employers preemptively change their practices because they fear disparate impact litigation. For the same reasons that this is likely, it is also likely that when early disparate impact cases were settled, outcomes disproportionatelyfavored plaintiffs. 10. Donahue and Heckman (1991) caution that, controlling for other factors, the covariation between litigation rates and relative employment outcomes for minorities will not necessarily capture the market impact of Title VII enforcement. This is precisely because employers probably change their practices to preempt litigation (see also Edelman, Uggen, and Erlanger 1999). Donahue and Heckman's (1991) review of the literature on the labor market effects of Title VII and of the Labor Department's Office of Federal Contract Compliance (OFCC) affirmativeaction programs suggests that the latter had a positive effect on the relative employment of African Americans, especially by improving minority employment among government contractors who experienced an OFCC compliance review. Results of studies trying to link overall Title VII enforcement to labor market improvements for blacks are more equivocal, in large part due to methodological difficulties (see especially Donahue and Heckman 1991; Smith and Welch 1984). Donahue and Heckman's (1991) careful detective work suggests that direct measures of Title VII and OFCC enforcement will underestimate the federal government's role in improvinglabor market opportunities for blacks.They conclude that federal antidiscriminationin employment policy, in conjunction with federal antidiscriminationefforts aimed at education and voting, improved African Americans' status between 1965 and 1975-the very period in which disparateimpact was invented. The early 1970s also are the period in which plaintiffs achieved their greatest appellate victory rates in disparateimpact cases (Burstein and Edwards 1994). 11. Courts recognized restrictions on the appropriateness of applicant flow data stemming from how employer selection rules could discourage applications from qualified blacks (Player 1988,p. 306). Holzer's (1996) study of employment possibilities for less-educated workers suggests that courts ought to be even more concerned about applicant self-selection in response to known or presumed employer selection procedures. 12. There are limits to the applicabilityof the four-fifthsrule. A selection rate greater than 80 percent may constitute adverse impact if it is statistically significant and practically important, Conversely, a selection rate less than 80 percent may not constitute adverse impact if the difference is not statistically significant, the number of applicants or selections is tiny, or affirmative action in recruitmenthas resulted in an atypical applicantpool. 13. After Beazer, there were two lines of case language interpretingbusiness necessity, but the lower courts generally did not adopt the more relaxed Beazer standard.Mindful of both Beazer and the tougher standardin Albemarle,the lower courts did allow some flexibility in proving business necessity.They varied how closely predictive tests or other selection devices had to be of onthe-job performance or whether statistically significant predictive validity was required. According to legal precedent, the degree of difficultyfor employer proof of business necessity varied with the skill level of the job to be filled, the risk of human or economic loss from bad job performance, and the strength of a plaintiff's case. In general, courts required a high and statisticallysignificant correlation between the challenged selection device and job performance when the jobs were unskilled and the risks of human and economic loss from unsuccessful job performance were small.When jobs were professional, required high skill, artistic ability,or subjectivejudgment, and especially when large risks would result from employing unqualified persons (e.g., in the case of airline flight engineers), courts lightened the employer's burden of proving business necessity. Similarly,since Beazer could legitimately be read as expressing a lesser burden of proof for a defendant because a plaintiff's evidence was so weak, the weight of "necessity" that the lower courts required to be proved was related to the level of impact established by plaintiff's data. Where that impact was very large, the employer would have to establish that the selection device was absolutely required for efficient and safe business operations. But where the impact was very

40

THE SOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

small, a selection device that "significantlyserved" business goals might be enough, even if the device were not required by these goals (Player 1988, p. 369, quoting Beazer). 14. Executive Order 11246, enacted by President Lyndon Johnson in 1965, stated: "[T]he contractors will not discriminateagainst any employee or applicant for employment because of race, creed, color, or national origin. The contractor will take affirmative action to ensure that applicants are employed, and that employees are treated during employment without regard to their race, creed, color, or national origin" (Code of Federal Regulations 1964-1965 compilation, vol. 3, pp. 339-348, emphasis added). Executive Order 11375 amended Executive Order 11246 to add sex. Government contractors and politicians argued that the Executive Order itself prohibited goals and timetables, as did Title VII and equal protection, but these argumentswere defeated in the federal courts and Congress in the late 1960s to early 1970s.This paved the way for the Labor Department to institutionalizeits affirmativeaction requirements,including goals and timetables. 15. Section 713(b) indicates that even if an employer violates Title VII in reliance on an EEOC opinion or written interpretation like the agency's Affirmative Action Guidelines, employer reliance bars the suit, even if the EEOC provision the employer relied on is later judged invalid by the courts. 16. Since the Weberprogram occurred during federal contract compliance, Weberalso established that affirmative action plans required to receive government contracts under Executive Order 11246 would be considered "voluntary"for purposes of Title VII enforcement. Unionemployer agreements on affirmative action made during collective bargainingand even affirmative action plans adopted to settle Title VII suits and approved by courts as part of consent decrees likewise are considered voluntary. 17. The definition of a federal contractoris very broad, so broad that a 1973 article in the Personnel Administratorestimated that 80 percent of employers were required to specify affirmative action goals and timetables and to make a good faith effort to meet them pursuant to Executive Order 11246 (Pearson 1973, p. 23; see also Graham 1990). 18. Smith and Welch (1984) suggest that the combination of increased preferences of federal contractors for blacks counterbalanced by declines in black employment among employers not covered by Executive Order 11246 accounts for the modest aggregate gains in black employment since the advent of the EEOC and OFCC. Some preference for minority and female employees through the use of race- and gender-conscious hiring is, of course, essential for effective affirmative action, as Reskin (1998) underscores in her review of affirmativeaction programs.But this is not the same as rigid quota hiring- preferential treatment based solely on race or gender. 19. Leonard's (1984) finding that affirmativeaction is not associated with declines in minority and female productivityrelative to white male productivity(as the composition of the workforce shifts to include greater numbers of minorities and women) is consistent with this conclusion. A critic of affirmativeaction could argue, however, that average employee productivitywould have been still better without affirmativeaction. Surveying the totality of research on the productivity effects of affirmativeaction, Reskin (1998) finds that there is no evidence that affirmativeaction reduces workplace productivity or that employees hired under affirmative action programs are less qualified than other employees even though a majority of Americans think that affirmative action programsmean the hiring and promotion of less-qualifiedemployees. Majorityperceptions do not square with the demographicbreakdown of self-reportedinstances of discriminationin surveys or with case filings of traditional versus reverse discrimination cases, and Taylor's (1995) results suggest that experiencewith affirmativeaction enhances, rather than detracts,from whites' positive evaluations of race-targetedgovernment programs. 20. Results of Edelman's (1990, 1992) quantitative research show that rates of adoption of affirmativeaction offices or rules were positively related to firm size, firm closeness to the public sector, firm status as a government contractor,and the existence of a firm personnel department. This is consistent with the coverage of Executive Order 11246, the targeting of Title VII enforcement to large firms,and the publicity given by the personnel journals to such enforcement.

DisparateImpactand the Quota Debates

41

21. Welch's (1981) theoretical approach to deducing upper bounds for the impact of affirmative action programs discusses a variety of incentives, some intended and some perverse, that are built into these programs under diverse conditions, assuming that all the programs operate as strict quotas. Welch's calculations are highly illuminating but irrelevant here, because he assumes precisely what is at issue here--employer use of quotas. 22. For public opinion about equal employment and affirmativeaction policies, see, e.g., Bobo and Kleugel (1993), Lipset (1992), Kleugel and Smith (1986), Kahlenberg (1996), Schuman,Steeh, and Bobo (1985), Steeh and Krysan (1996), and Entman (1997). As for small businesses, available evidence suggests that at both the attitudinal and behavioral levels, smaller firms in the 1970s and 1980s were much less committed to any sort of affirmativeaction than were their larger counterparts (Malbin 1973;Noble 1986;Edelman 1992). 23. Early 1990s attacks on disparate impact equating it with quotas also were disingenuous because it was opponents' own attacks on affirmativeaction begun during the first Reagan administration that helped create the very situation that some Republicans later criticized for putting employers between the alleged rock of Title VII disparate impact liability and the alleged hard place of Title VII liability for discriminatingagainst white males (Days 1984; USDJ 1987;Belton 1990).That is, if- by 1989-aggressive affirmativeaction in response to disparateimpact enforcement began to threaten employers with legal liability for reverse discrimination, it was in large part because: (1) the Reagan administrationattacked the EEOC Affirmative Action Guidelines attempting to insulate affirmativeaction employers from Title VII liability for reverse discrimination, (2) the administrationintervened aggressivelyon behalf of white plaintiffs in reverse discrimination suits, and (3) the administration tried repeatedly to persuade the courts to narrow affirmative action remedies to adjudicated victims of intentional discrimination. Interestingly, while employers litigatingTitle VII claims in the late 1980s attacked disparateimpact theory when such attacks could help build their cases, in the early and mid-1980s,many large employers feared that Republican attacks on both Executive Order 11246 and Title VII enforcement would go too far, underminingemployee selection procedures these employers preferred to maintain and upsetting a fragile economic and political compromise between minority and majority groups ("New Bias,"1981;Noble 1986;Belton 1990). Such large business attempts to maintain "balance"in their affirmative action politics and policies are consistent with my "competing incentives" picture of employers working to create good employment relations with all groups of employees. 24. The Bush administration and its employer allies tried to insert into the 1990-1991 Civil Rights Acts specific language that would have interpreted the employer's burden to prove business necessity so as legislatively to weaken that burden. Civil rights groups and their liberal allies in Congress tried to include specific language that would have strengthened the employer's burden legislatively.Neither group succeeded, so the legislative restoration of Griggs came complete with ambiguities about proving necessity that had emerged from the two lines of lower court cases interpretingthe Supreme Court opinion in Beazer (Ritz 1992).

REFERENCES Abrams,Elliot. 1972."The Quota Commission."Commentary,October, pp. 54-57. Antonovky, Aaron. 1960."The Social Meaning of Discrimination."Phylon 21:81-95. Baron, James N. 1992. "OrganizationalEvidence of Ascription in Labor Markets."Pp. 113-143 in New Approaches to Economic and Social Analyses of Discrimination, edited by Richard Cornwall and P.V.Wunnava.Westport,CT:Praeger. Baron, James N., and Andrew E. Newman. 1989. "Pay the Man:Effects of Demographic Composition on PrescribedWage Rates in the California Civil Service."Pp. 107-130 in Pay Equity: Empirical Inquiries,edited by R. T. Michael, H. L. Hartmann, and B. O'Farrell.Washington DC: National Academy Press.

42

THE SOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

. 1990. "ForWhat It's Worth:Differences across Organizations,Occupations and the Value of Work Done by Women and Nonwhites."American Sociological Review 55:155-175. Belton, Robert. 1990. "The Dismantling of the Griggs Disparate Impact Theory and the Future of Title VII:The Need for a Third Reconstruction."YaleLaw and Policy Review 8:223-257. Belz, Herman. 1991. Equality Transformed:A Quarter-Centuryof AffirmativeAction. New Brunswick, NJ:TransactionBooks. Bielby,WilliamT., and JamesN. Baron. 1986."Men and Women at Work:Sex Segregation and Statistical Discrimination."American Journalof Sociology 91:759-799. Bills, David. 1992. "A Survey of Employer Surveys:What We Know about Labor Markets from Talkingwith Bosses."Researchin Social Stratificationand Mobility 11:3-32. Biskupic, Joan. 1989. "Rights Bill Is Almost Ready after Long Delay, Debate." Congressional QuarterlyWeeklyReport 47:3055-3102. Blau, Peter, and Otis Dudley Duncan. 1967. The American Occupational Structure.New York: John Wiley and Sons. Blumrosen, Alfred W. 1971. Black Employment and the Law. New Brunswick, NJ: Rutgers University Press. . 1972. "Strangersin Paradise: Griggs v. Duke Power Co. and the Concept of Employment Discrimination."Michigan Law Review 71:59-110. . 1993. Modern Law: The Law TransmissionSystem and Equal Employment Opportunity. Madison:University of Wisconsin Press. Bobo, Lawrence, and James R. Kleugel. 1993. "Opposition to Race-Targeting: Self Interest, StratificationIdeology or Racial Attitudes?"American Sociological Review 58:443-464. Bobo, Lawrence, and Susan A. Suh. 1996. "Showing Racial Discrimination: Analyses from a Multi-Ethnic Labor Market."Workingpaper 75. New York:Russell Sage Foundation. Brimelow, Peter, and Leslie Spencer. 1993. "When Quotas Replace Merit, Everybody Suffers." Forbes,February15, pp. 80-102. Bullock, Charles, and Harrell Rogers. 1976. "Institutional Racism: Prerequisites, Freezing and Mapping."Phylon 37:212-223. Burstein, Paul. 1991. "Legal Mobilization as a Social Movement Tactic:The Struggle for Equal Employment Opportunity."AmericanJournalof Sociology 96:1201-1225. . 1993. "AffirmativeAction, Jobs and American Democracy: What Has Happened to the Quest for Equal Opportunity?"Law and Society Review 26:901-922. Burstein, Paul, and Mark Evan Edwards. 1994. "The Impact of Employment Litigation on Racial Disparityin Earnings:Evidence and Unresolved Issues."Law and SocietyReview28:79-111. 1986. "The Economic Analysis of Labor Market Discrimination:A Survey."Pp. 693Glen. Cain, 785 in Handbook of Labor Economics, vol. 1, edited by O. Ashenfelter and R. Layard. Amsterdam:Elsevier Science Publishers. Carmichael, Stokely, and Charles Hamilton. 1967. Black Power: The Politics of Liberation in America. New York:Vintage. Chambliss,Elizabeth. 1989. "Role of Confrontation and Equal Employment Advocacy."Institute for Legal Studies working paper no. 4-6. Madison:Institute for Legal Studies. University of Wisconsin Press. Chayes,Antonia Handler. 1974. "Make Your Equal Opportunity Program Court Proof."Harvard Business Review 52:81-89. Crenshaw, Kimberle Williams. 1988. "Race, Reform and Retrenchment: Transformation and Legitimizationin Anti-Discrimination Law."HarvardLaw Review 101:1331-1387. Culp, Jerome. 1985. "A New Employment Policy for the 1980's:Learning from the Victories and Defeats of Twenty Years of Title VII." Rutgers Law Review/Civil Rights Developments 37:895-919. Drew S., III. 1984. "TurningBack the Clock:The Reagan Administration and Civil Rights." Days, Harvard Civil Rights/CivilLibertiesReview 19:309-347.

DisparateImpactand the Quota Debates

43

Dhanens, Thomas P. 1979. "Implicationsof the New EEOC Guidelines." Personnel, SeptemberOctober, pp. 32-38. Donahue, John, and James Heckman. 1991. "Continuous versus Episodic Change:The Impact of Civil Rights Policy on the Economic Status of Blacks." Journal of Economic Literature 29:1603-1643. Donahue, John J., III. 1994. "Employment Discrimination Law in Perspective:Three Concepts of Equality."Michigan Law Review 92:2583-2611. Donahue, John J.,III, and Peter Siegelman. 1991."The ChangingNature of Employment Discrimination Litigation."StanfordLaw Review 43:983-1033. Edelman, Lauren. 1990. "Legal Environments and Organizational Governance: The Evolution of Due Process in the American Workplace."American Journal of Sociology 95:1401-1440. --. 1992. "Legal Ambiguity and Symbolic Structures: Organizational Mediation of Law." American Journal of Sociology 97:1531-1576. Edelman, Lauren B., Howard S. Erlanger, and John Lande. 1993. "Internal Dispute Resolution: The Transformationof Civil Rights in the Workplace."Law and Society Review 27:497-534. Edelman, Lauren B., Stephen Petterson, Elizabeth Chambliss, and Howard S. Erlanger. 1991. "Legal Ambiguity and the Politics of Compliance:Affirmative Action Officers Dilemma." Law and Policy Review 13:73-97. Edelman, Lauren, Christopher Uggen, and Howard Erlanger. 1999. "Grievance Procedures as Rational Myth."AmericanJournalof Sociology 105:406-455. England, Paula. 1992. ComparableWorth:Theoriesand Evidence. New York:Aldine de Gruyter. Entman, Robert M. 1997. "ManufacturingDiscord: Media in the Affirmative Action Debate." Press/Politics2:32-51. Epstein, Richard. 1992. Forbidden Grounds: The Case against Employment Discrimination Laws. Cambridge.MA: HarvardUniversity Press. Farrell, Michael. 1978. "Proposed EEOC Regulations." Personnel Administrator,November, pp. 51-60. Feagan, Joe. 1977. "Indirect Institutional Discrimination: A Typological and Policy Analysis." American Politics Quarterly5:177-199. Featherman, David L., and Robert M. Hauser. 1976. "Sexual Inequalities and Socioeconomic Achievement in the U.S., 1962-1973."American Sociological Review 41:462-483. Fisher,Anne B. 1985."BusinessmenLike to Hire by the Numbers."Fortune,September 17, pp. 2630. Goeke, Joseph R., and Caroline S. Weymar.1969. "Barriersto Hiring the Blacks."Harvard Business Review 47:144-152. Gorman, Robert A. 1977. Basic Texton Labor Law: Unionization and Collective Bargaining. St. Paul, MN:West Publishing. Graham, Hugh Davis. 1990. The Civil Rights Era: Origins and Development of National Policy, 1960-1972. New York:Oxford University Press. Halaby, Charles.1979. "Job-SpecificSex Differences in OrganizationalReward Attainment: Wage Discriminationvs. Rank Segregation."Social Forces 58:108-127. Harris,John F 1996. "TheWarriorat the President's Elbow."WashingtonPost Weekly,June 10-16, pp. 12-13. Hoffman, Carl C. 1986. "Reopening the Affirmative Action Debate." Personnel Administrator, March,pp. 36-40. Holzer, Harry J. 1996.WhatEmployersWant:Job Prospectsfor Less Educated Workers.New York: Russell Sage Foundation. Jencks,Christopher.1992. RethinkingSocial Policy: Race, Poverty and the Underclass.Cambridge, MA: HarvardUniversity Press. Kahlenberg, Richard. 1996. "Bob Dole's Colorblind Injustice."WashingtonPost Weekly,June 1016, pp. 24-26.

44

THESOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

Kellough, J. Edward. 1989. Federal Equal Employment Opportunity and Numerical Goals and Timetables:An ImpactAssessment.New York:Praeger. Kennedy, Randall. 1986. "Persuasion and Distrust: A Comment on the Affirmative Action Debate." HarvardLaw Review 99:1327-1346. Kirschenman,Joleen, and Kathryn M. Neckerman. 1991. "We'dLove to Hire Them But ... : The Meaning of Race for Employers."Pp. 203-232 in The Urban Underclass,edited by Christopher Jencks and Paul Peterson. Washington,DC: Brookings Institution. Klein, Joe. 1995."AffirmativeInaction?"Newsweek,June 26, p. 23. Kleugel, James R., and Elliot R. Smith. 1986. Beliefs about Inequality:Americans' Views of WhatIs and WhatOught to Be. New York:Aldine de Gruyter. Knowles, Lois L., and Kenneth Prewitt. 1969. InstitutionalRacism in America. Englewood Cliffs, NJ: Prentice-Hall. Leonard, JonathanS. 1990."The Impact of AffirmativeAction Regulation and Equal Opportunity Law on Black Employment."Journalof Economic Perspectives4:47-63. . 1984."Anti-Discrimination or Reverse Discrimination:The Impactof ChangingDemographJournalof HumanResources19:145-174. ics,TitleVII and AffirmativeAction on Productivity." . 1986. "WhatWas Affirmative Action?" American Economic Review 76 (Papers and Pro•ceedings):359-363. Lipset, Seymour Martin. 1992. "AffirmativeAction and the American Creed." Wilson Quarterly 16:52-62. Loury, Charles. 1991. "Beyond Civil Rights."Pp. 435-451 in Racial Preferenceand Racial Justice: The New AffirmativeAction Controversy,edited by R. Nieli. Washington, DC: Ethics & Public Policy Center. Lundberg, Shelly J. 1989. "Equality and Efficiency:AntidiscriminationPolicies in the Labor Market." ContemporaryPolicy Issues 7:75-94. Malbin, Michael. 1973."Employment Report/Proposed Federal Guidelines on Hiring Could Have Far-ReachingImpact."National JournalReportsvol. 1973:1429-1434. Matthies,Mary.1976."The Developing Law on Equal Employment Opportunity."Journalof ContemporaryBusiness,Winter,pp. 29-46. Mayfield, Harold. 1964. "Equal Employment Opportunity: Should Hiring Standards Be Relaxed?" Personnel,September-October, pp. 8-17. Merton, Robert. 1976. Sociological Ambivalence and OtherEssays. New York:Free Press. "The New Bias on Hiring Rules."1981. Business Week,pp. 123-127. Noble, Kenneth. 1986. "Employers Are Split on Affirmative Action Goals." New York Times, March 3, p. 1. Norton, Eleanor Holmes. 1990. "The End of the Griggs Economy: Doctrinal Adjustment for the New American Workplace."YaleLaw and Policy Review 8:197-204. Oliver, Melvin L., and Thomas M. Shapiro. 1995. Black Wealth/WhiteWealth:A New Perspective on Racial Inequality.New York:Routledge. Ornati, Oscar A., and Edward Giblin. 1975. "The High Cost of Discrimination."Business Horizons, February,pp. 35-40. Pearson, David. 1973. "OFCC and EEOC Demands: Guidelines to Frustration." Personnel Administrator,November-December, pp. 21-25. Pedriana, Nicholas. 1999. "The Historical Foundations of Affirmative Action, 1961-1971. Research in Social Stratificationand Mobility 17:3-32. Pedriana, Nicholas and Robin Stryker. 1997. "Political-CultureWars 1960s Style: Equal Employment Opportunity-AffirmativeAction Law and the Philadelphia Plan."American Journal of Sociology 103:633-691. Player, Mack. 1988. EmploymentDiscriminationLaw. St. Paul, MN:West Publishing. -. 1999. FederalLaw of EmploymentDiscriminationin a Nutshell. 4th ed. St Paul, MN:West Publishing.

DisparateImpactand the Quota Debates

45

Prager,Jeffrey.1982. "Equal Opportunity and Affirmative Action: The Rise of New Social Understandings."Researchin Law, Deviance and Social Control 4:191-218. Reskin, Barbara. 1998. The Realities of AffirmativeAction. Washington,DC: American Sociological Association. -- . 2000. "The Proximate Causes of Employment Discrimination." ContemporarySociology 29:319-328. Ridgeway, Cecilia. 1991. "The Social Construction of Status Value: Gender and Other Nominal Characteristics."Social Forces 70:367-386. Ritz, Susan. 1992. The Civil Rights Act of 1991: Its Impact on Employment Discrimination Litigation. Litigation and Administrative Practice Series. New York:PracticingLaw Institute. Rosen, Gerard R. 1974."Industry'sNew Watchdogin Washington."Dun's Review 13:83W-85. Rosen, Jeffrey.1996."The Day the Quotas Died." New Republic,April 22, pp. 21-27. Rosenbaum, James. 1985. "Jobs,Job Status and Women's Gains from Affirmative Actior: iTnplications for Comparable Worth." Pp. 116-136 in Comparable Worth:New Directi6ns for Research, edited by H. J. Hartmann, P. A. Roos, and D. J. Treiman. Washington, DC: National Academy Press. Ruch, Floyd L. 1971. "The Impact on Employment Procedures of the Supreme OCurtDecision in the Duke Power Case."PersonnelJournal,October, pp. 777-783. Kevin. 1996. "Atlanta Leaders See Racial Goals as Olympic Ideals," NeVvYork Times,June Sack, 10, pp. A1, A12. Sanders,Alan. 1990. "A Quotas vs. Voters Dilemma:The President Endangers His Support among Blacks by Digging in His Heels on Civil Rights."Time,Octobet 29, p. 42. Schuman, Howard, Charlotte Steeh, and Lawrence Bobo. 1985. Racial Attitudes in America: Trendsand Interpretations.Cambridge,MA: HarvardUniversity Press. Seligman, Daniel. 1973. "How 'Equal Opportunity' Turned into Employment Quotas." Fortune, March, pp. 160-168. . 1982. "AffirmativeAction Is Here to Stay."Fortune,April 19, pp. 143-162. Sewell, William H., and Robert M. Hauser. 1975. Education, Occupation and Earnings:Achievement in the Early Career.New York:Academic Press. Siegelman, Peter, and John J. Donahue, III. 1990. "Studyingthe Iceberg from Its Tip:A Comparison of Published and Unpublished Employment Discrimination Cases." Law and Society Review 24:1133-1170. Simpson, George Eaton, and J. Milton Yinger. 1985. Racial and CulturalMinorities:An Analysis of Prejudice and Discrimination.5th Ed. New York:Plenum Press. Smith, James P., and Finis R. Welch. 1984. "AffirmativeAction and Labor Markets."Joural of Labor Economics 2:269-301. . 1989. "Black Economic Progress after Myrdal."Journal of Economic Literature 27:519562. Smith, Robert C. 1995. Racism in the Post-Civil Rights Era: Now You See it, Now You Don't. Albany: State University of New York Press. Sovern, Michael I. 1966. Legal Restraintson Discrimination in Employment. New York:Twentieth Century Fund. Sowell, Thomas. 1976."'AffirmativeAction' Reconsidered."Public Interest42:47-56. Steeh, Charlotte, and Maria Krysan. 1996. "The Polls-Trends: Affirmative Action and the Public, 1970-1985."Public Opinion Quarterly60:128-158. Steinberg, Ronnie. 1990. "The Social Construction of Skill: Gender, Power and Comparable Worth."Workand Occupations 17:449-482. Stephanapolous, George, and ChristopherEdley, Jr. 1995. "AffirmativeAction Review: Report to the President."Washington,DC.

46

THESOCIOLOGICAL QUARTERLYVol. 42/No. 1/2001

Stryker, Robin, Martha Scarpellino, and Mellisa Holtzman. 1999. "Political Culture Wars 1990s Style: The Drum Beat of Quotas in Media Framing of the Civil Rights Act of 1991." Researchin Social Stratificationand Mobility 17:33-106. Taylor, Marylee. 1995. "White Backlash to Affirmative Action: Peril or Myth." Social Forces 73:1385-1414. Turner,MargeryAustin, Michael Fix, and Raymond J. Struyk. 1991. OpportunitiesDenied, Opportunities Diminished: Racial Discrimination in Hiring. Washington, DC: Urban Institute Press. United States Department of Justice. Office of Legal Policy (USDJ). 1987. RedefiningDiscrimination: "DisparateImpact" and the Institutionalizationof AffirmativeAction: Report to the Attorney General.Washington,DC: GPO. Vernon-Gerstenfeld,Susan, and Edmund Burke. 1985. "AffirmativeAction in Nine Large Companies."Personnel,April, pp. 54-60. Welch, Finis. 1976. "Employment Quotas for Minorities."Journal of Political Economy 84:S105S139 Willie, Charles. 1974. InstitutionalRacism:In Search of a Perspective.New Brunswick, NJ: Transaction Books.

View publication stats