7.5 pages body of supporting, cited information, Current issues and future trends related to the topic. Either way, this element should be addressed. Discuss your personal interest and takeaways regarding the topic and research (not the same as the closin

Equal Employment Opportunity Legislation and the Income of Women and Nonwhites Author(s): Paul Burstein Source: American Sociological Review, Vol. 44, No. 3 (Jun., 1979), pp. 367-391 Published by: American Sociological Association Stable URL: https://www.jstor.org/stable/2094880 Accessed: 20-11-2019 17:59 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide

range of content in a trusted digital archive. We use information technology and tools to increase productivity and

facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

https://about.jstor.org/terms

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to American Sociological Review

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

EQUAL EMPLOYMENT OPPORTUNITY LEGISLATION AND THE

INCOME OF WOMEN AND NONWHITES*

PAUL BURSTEIN

Yale University

American Sociological Review 1979, Vol. 44 (June):367-391

This article examines the impact of federal equal employment opportunity legislation on the income of white women, nonwhite women, and nonwhite men relative to the income of white men at the national level since the late 1940s. The analysis differs from those previously published in comparing all groups with white men, including changes in attitudes in the analysis, and dealing in detail with enforcement of the laws. The model developed in the article works well when used to examine increases in nonwhite income, but does not deal satisfactorily with the incomes of white women. The findings indicate that there are important differences between race and sex discrimination which will have to be taken into account in theories of labor market discrimination.

The struggle to end discrimination against minorities and women has played a central part in recent American history. Much of the struggle has been devoted to the achievement of equal opportunity in employment. To help women and minorities achieve equal opportunity, Congress passed several equal employ- ment opportunity (EEO) laws in the 1960s and 1970s. The aim of this article is to gauge the national-level impact of these laws on the incomes of women and non- whites relative to those of white men. The focus will be on the four laws most di- rectly related to employment in the entire civilian labor force-the Equal Pay Act of 1963, Title VII of the Civil Rights Act of 1964, and the 1972 laws amending each of them. '

The analysis differs in three significant respects from comparable ones published previously. First, the incomes of white and nonwhite women, as well as nonwhite men, will be compared with those of white men, who were clearly intended to be the comparison group in the EEO laws. No previous aggregate-level studies gauge the laws' impact on women relative to white men.

Second, changes in attitudes toward discrimination will be included in the models. The economic theory of employ- ment discrimination, which provides the theoretical framework for this study, hy- pothesizes that the white male "taste for discrimination" will affect the incomes of those discriminated against. Past applica- tions of the theory, however, make no attempt to examine the consequences of

* Address all communications to: Paul Burstein; Department of Sociology; Yale University; New Haven, CT 06520.

I would like to thank Richard Berk, Claude Fischer, Owen Fiss, Florence Katz, and Stanley Masters for helpful comments on an earlier draft of this paper. An earlier version was presented at the 1979 annual meeting of the Eastern Sociological Society, New York.

I The Equal Pay Act (Public Law 88-38, 77 U.S. Statutes At-Large 56-57) prohibited differential rates of pay for women and men who do equal work, with some exceptions. Title VII (PL 88-352, 78 Stat. 241-268) prohibited many discriminatory employ- ment practices. Enforcement relied very heavily on the private actions of aggrieved individuals; the Equal Employment Opportunity Commission (EEOC) created by the Act had no enforcement powers of its own; its role was limited primarily to conciliation.

Among the significant exceptions to the prohibitions against discrimination was one that permits discrimi- nation on the basis of sex (but not race) when it is found to be a "bona fide occupational qualification" for the job. The Equal Employment Opportunity Act of 1972 (PL 92-261, 86 Stat. 103-113) extended the coverage of Title VII and gave the EEOC the power to go to court when conciliation efforts failed. Sec- tion 906 of the Education Amendments of 1972 (PL 92-318, 86 Stat. 235-381) abolished many of the exemptions in the Equal Pay Act. For detailed de- scriptions and analyses of these and related laws, see Task Force on Women, 1975: chap. 3; Gitt and Gelb, 1977; Harvard Law Review, 1971; Fiss, 1971. Other federal laws may affect the economic situation of women and minorities; this paper deals with those laws intended to have the most direct and wide- spread impact.

367

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

368 AMERICAN SOCIOLOGICAL REVIEW

changes in tastes. This paper will demon- strate the importance of including approx- imate measures of tastes in the analysis.

Third, the paper will refine considerably the treatment of EEO law enforcement, which was measured in very simple ways in past work.

The improved model developed in the paper works well when used to examine changes in the incomes of nonwhite men and women relative to those of white men, but does not deal satisfactorily with changes in the incomes of white women. The findings indicate that there are impor- tant differences between race and sex dis- crimination which will have to be taken into account in theories of labor market discrimination.

EMPIRICAL CONTROVERSY

Have EEO laws had a significant, favorable impact on the economic situa- tion of women and minority group mem- bers? Experts disagree. Legal scholars in- volved in employment discrimination cases tend to see the laws' effects as slow and uncertain. Professor Derrick Bell, Jr. (1977:681) of Harvard Law School, for example, introduces a symposium on equal employment law by writing that he has

serious doubt as to the long-term effective- ness and worth of equal employment laws. ... If the country were really committed to eradicating the social and economic burdens borne by the victims of employment dis- crimination, it would have fashioned a far more efficacious means of accomplishing this result. At present, the law channels charges of employment discrimination into a burdensome, conciliation-oriented adminis- trative structure that functions, in the mean, on a case-by-case basis, depending on effectively-prosecuted litigation and a sym- pathetic judiciary for even the hard-won progress thus far achieved. Even the most wildly optimistic among us cannot reason- ably hope that reliance on this complex and uncertain process will close the wide gap in income standards and unemployment rates between black and white and male and female employees.

Herbert Hill, the foremost expert on black labor and the American legal sys- tem, arrives at a cautious but somewhat less pessimistic conclusion about blacks in

the labor force. He feels that the apparent victory achieved by the passage of the 1964 Civil Rights Act was largely illusory in the short-run because employers and unions vigorously defended their tradi- tional discriminatory practices for years. Years of litigation began to turn the tide, however, and he concludes that "[b]y the mid-1970s the impact of the law was evi- dent as court orders began to make the elimination of racial discrimination an economic necessity" (1977:4).

Neither Bell nor Hill presents quantita- tive evidence to support his conclusions. The most sanguine conclusions about the impact of EEO legislation are those of Richard Freeman and Stanley Masters, economists who attempted to analyze the laws' impact on black-white economic dif- ferences. Freeman (1973:118-9) con- cludes:

The income and occupational position of black workers improved significantly rela- tive to those of whites in the sixties. ... Much of the improvement in the black eco- nomic position that took place in the late sixties appears to be the result of govern- mental and related antidiscriminatory ac- tivity associated with the 1964 Civil Rights Act.

Masters (1975:143) similarly concludes that his findings support "the liberal view that there should be a continuous im- provement in the relative position of blacks once some precipitous event like the Civil Rights Act and the pressures that led to it have occurred.' '2

2 Other works on the aggregate-level conse- quences of federal legislation attempt to show that the gains of black men have been limited to the South (Vroman, 1974; McCrone and Hardy, 1978). There are also some studies which proceed at lower levels of aggregation, such as industries or SMSAs; find- ings about the efficacy of EEOC enforcement, Office of Federal Contract Compliance activities, and other programs, are mixed; see, e.g., Adams, 1972; Beller, 1977; Bergmann and Lyle, 1971; Heckman and Wol- pin, 1976; Smith and Welch, 1977. A study of Cana- dian equal pay legislation claimed to show it had no effect (Gunderson, 1975).

There are also a number of studies which purport to gauge the impact of civil rights laws by simply presenting data on black-white or female-male dif- ferences along some dimension before and after the passage of the laws, without controlling for any other factors; Niemi (1974) is typical. The conclusions of these latter studies, carried out without controls for other relevant variables, cannot be taken seriously.

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 369

Thus, those who deal with the laws tend to see their effects as slow and uncertain. Those who deal with income data tend to see the laws' effect as immediate and con- sequential so far as the black-white in- come ratio for males is concerned, and have nothing to say about the incomes of women as compared with white men.

THEORETICAL ISSUES AND MODELS OF

LEGISLATIVE IMPACT

Economic Models: Theory vs. Application

Almost all the quantitative work on the effects of EEO legislation has been writ- ten by economists who draw heavily on Becker's (1971) Economics of Discrimina- tion. Three studies form the core of empir- ical work on the aggregate-level effects of EEO legislation: Landes's (1968) article on the impact of state fair employment practices laws on the economic status of nonwhites; Freeman's (1973) paper on the impact of the 1964 Civil Rights Act; and Masters's (1975: chap. 6) chapter on the effects of the 1964 Act on the economic situation of black men. Other works use very similar theoretical approaches, vari- ables, and analytic techniques.

Although the economic theory of labor market discrimination has ramified be- yond concise description (see Marshall, 1974; Blau and Jusenius, 1976), those as- pects of the theory applied in the empirical works can be summarized as follows: Dif- ferences in the labor market treatment of racial minorities or women begin with the taste for discrimination of employers, em- ployees, or customers. Because white men do not like members of other social categories in some way-they may not like to work with them, to see them have equal status positions, or whatever-they act as if they are willing to pay a monetary price to avoid associating with members of such categories. White employers who have a taste for discrimination, for exam- ple, may be willing to hire black workers only if the blacks will work for lower wages than whites of equal skill, with the extra money going to the employers as payment for overcoming their aversion to blacks. Under specified conditions, this behavior lowers the relative incomes of blacks. Laws prohibiting discrimination in

the labor market may work if they raise the price discriminators have to pay for indulging their tastes-by threatening them with fines or with having to make compensatory payments to those dis- crinminated against, for example.3

Thus, Freeman (1973:93-4) argues theoretically that two important forces appear to have reduced discrimination in the labor market between 1948 and 1972: a change in the demand or taste for dis- crimination, which plays the critical role in Becker's theoretical model, and a change in the cost of discrimination due to enforcement of Title VII of the 1964 Civil Rights Act and other national policies. Masters (1975:141) similarly argues that changes in the labor market position of blacks could have been brought about by legislation and attitude change.

Within this theoretical context, those studying the consequences of EEO legis- lation pursue a similar course: they basic- ally regress some measure of black-white differences in the economic sphere, such as the black-white income ratio, on vari- ables gauging the general state of the economy and the passage or enforcement of EEO laws. Finding significant coeffi- cients for the EEO variables, they con- clude that the legislation has significantly reduced some economic differences be- tween blacks and whites (though Landes finds this may have been at the cost of increased black unemployment).

Unfortunately, the studies are flawed in ways that make it difficult to accept their findings. Both Freeman and Masters argue that changes in black income may be due to two factors: change in the de- mand for discrimination and EEO legisla- tion. The empirical analyses include only one of the factors, however-EEO enforcement-and attribute changes in black incomes to this factor alone. Al- though both are tentative in their conclu- sions (with Masters admitting that his es-

3 Although the notion that discrimination can be reduced by making it expensive is formally derived from economic theory, economic sanctions are seen as the most effective way to reduce discrimination by many others as well; see Hill, 1977:4; Levitan et al., 1975: chap. 13; for evidence of corporate attitudes, see "Business Resisting U.S. on Halting Work Bias," New York Times, May 26, 1978: Di.

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

370 AMERICAN SOCIOLOGICAL REVIEW

timates are upperbound estimates of the true impact of the laws), neither tries to include changes in the taste or demand for discrimination explicitly in the analyses (see Freeman, 1973:94, 105; Masters, 1975:141, 143).

Past analyses thus appear to have a serious specification problem. EEO en- forcement activities are positively corre- lated with increases in favorable attitudes towards blacks and women in the labor force. Consequently, including a measure of demand for discrimination in the analysis would reduce the apparent im- pact of EEO legislation, possibly to in- significance.4

Economists have developed an elabo- rate rationale for not using attitude data to measure tastes or demand, and the argu- ment makes sense in many cases.5 To a sociologist, however, it seems plausible that the factors that probably helped pro- duce EEO legislation in the first place- changing attitudes, in particular- continue to operate after the law is passed and have an independent impact on the economic standing of groups protected by the legislation. It seems much better to include a rough proxy for taste for dis- crimination in the analysis than to ignore it completely.

Unfortunately, the hypothesis that atti- tudes about discrimination and EEO laws may affect the economic status of women or nonwhites has not been tested by soci- ologists. Changes in the economic status of blacks relative to whites and women relative to men are a major concern in the recent literature on stratification, but most works either attribute such changes partly to changes in laws and attitudes without providing evidence (Farley, 1977:206), mention the possibility that laws may have had an impact without investigating it (e.g., Featherman and Hauser, 1976:464; Snyder and Hudis, 1976:215, 231), or sim- ply ignore the legislation when discussing the economic status of women and blacks.

Realism in the Analysis of Enforcement

There is an additional difficulty in the extant work. Although Landes and Freeman try alternative specifications of their EEO enforcement variables, their measures do not seem like realistic operationalizations of their theory.6 They are interested, essentially, in costs im- posed on employers (or unions or employ- ees) for discriminating; their notion is that employers will change their behavior once they realize that it will be costly not to do so. Ideally, therefore, the best predictor of EEO legislation efficacy should be a mea- sure of the costs incurred by employers who continue to discriminate after dis- crimination is prohibited-a measure of the probability of being caught multiplied by the penalty imposed would be best. Such a measure is impossible to find (though some possible alternatives will be discussed below), but the measures actu- ally employed, EEOC expenditures mea- sured different ways, seem insufficient. During the entire period studied by Freeman, the EEOC had no enforcement powers; it was limited to attempting to conciliate complaints of discrimination. Freeman acknowledges that the EEOC variable is a proxy for all federal enforce-

4 The only empirical attempt to measure taste for discrimination in a study of EEO legislation impact is Bergmann and Lyle's (1971) study of black-white occupational differences across metropolitan areas and industries. Using the 1968 Wallace vote as a proxy for white taste for discrimination, they found that this variable explained black-white differences better than differences in education did, and better than the existence of a state fair employment prac- tices law. But they did not use actual attitude data or follow up on the idea.

5 Strictly speaking, there is a difference between a taste for discrimination and a demand for it. Changes in tastes, which economists tend to consider entirely exogenous to their models, may lead to changes in demand; this may be seen very roughly as akin to the gap between attitudes and behavior in social psycho- logical research. In practice, neither tastes nor de- mand for discrimination are measured directly. Berger's sophisticated discussion of the factors af- fecting employment discrimination states many of the same issues in sociological language, and is con- sistent with the argument made here; unfortunately, his empirical work cannot answer the questions raised by his theoretical formulation; see Berger, 1967: chap. 4; Freeman, 1973:93-4; Becker, 1971: chap. 9.

6 Landes is less relevant here because he was con-

cerned with state legislation. Masters's operationali- zation is the simplest of all-he simply includes two time trend variables and a dummy variable to mark the passage of Title VII.

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 371

ment activities, but this does not seem adequate justification for the magnitude of the gap between the variables required by the theory-actual costs-and those used.

In order to assess satisfactorily the im- pact of EEO legislation, realism about the enforcement process seems essential; the measures of enforcement effort should have some fairly clear relationship to the actual likelihood that those who discrimi- nate will have to pay a penalty for doing so.

This consideration leads to three changes in the way enforcement will be dealt with here, as compared with past work. First, although there is no way to measure the actual cost to employers of disobeying or complying with the laws, data are available on the amount of money awarded under the provision of the Equal Pay Act to those discriminated against. These data will be included in the analysis.

Second, because the binding enforce- ment procedures of Title VII involve court proceedings (rather than administratively issued cease-and-desist orders or some other mechanism), some measure of the decisions being made by the higher U.S. courts on EEO cases should be included in the analysis. When the laws were passed, it was not known, of course, how they would be interpreted by the courts. Employers would be most likely to take the law seriously, it can be argued, only if it were clear that the law would be inter- preted to make employers liable to sub- stantial financial penalties if they contin- ued to discriminate (see, e.g., Levitan et al., 1975:290-2). The first U.S. Court of Appeals decision that could be seen as making this point clearly was not decided, however, until 1971 (Robinson v. Loril- lard Corp., 44 F.2d 791; petition for writ of certiorari dismissed, 404 U.S. 1006 [1971]; also see Albermarle Paper Co. v. Moody, 422 U.S. 405 [1975]). The Su- preme Court did not decide any Title VII cases until 1971 (see Equal Employment Opportunity Commission, 1972:23-31), and doctrine remains unsettled in many areas (see Venick and Lane, 1977; com- pare University of California Regents v. Bakke, 46 U.S. Law Week 4896-4936 with Communication Workers of America v.

EEOC, 46 U.S. Law Week 3801; see also Newsweek, July 10, 1978:32, "The Hard Cases Coming").

Past empirical work has thus ignored the cumbersomeness of the legal process and the slowness with which the meaning of the laws becomes clear (see Bell's [1977] statement quoted above, and, e.g., Levine and Montcalmo, 1971). This paper will attempt to take actual enforcement possibilities into account by examining all EEO decisions by the U.S. Supreme Court and Court of Appeals to determine the number and proportion of cases de- cided in favor of women and racial minorities.7

Third, EEO legislation can be effective only if it is utilized; a law that is too cum- bersome and costly for people to use is not likely to be effective. Measures of EEOC expenditures, Equal Pay Act payments, and court decisions essentially gauge the supply of enforcement activities. It seems important to gauge the effective demand for EEO enforcement as well; this can be done by considering the number of charges brought to the EEOC and the number of cases brought before the courts.

As applied to the study of EEO legisla- tion, Becker's theory has been taken to imply that a decline in the taste for dis- crimination and an increase in its cost will lead to an increase in the relative income of the group that has been discriminated against, everything else being equal. That is also the point of view of this paper. But this article differs from past work in pro- posing that changes in taste should be in- cluded explicitly in the analysis and that it is necessary to be as realistic as possible in the analysis of costs likely to be im- posed upon discriminators.

RELATIVE INCOMES OF WOMEN AND

NONWHITES: DESIGN AND DATA

What Is to Be Explained

The analysis focuses on changes in the incomes of nonwhite men, nonwhite

7 In addition, EEOC expenditures could be hy- pothesized to be more effective after the EEOC was given enforcement powers in 1972 than they were before. Unfortunately, the time since 1972 is too brief to permit tests of differential impact.

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

372 AMERICAN SOCIOLOGICAL REVIEW

women, and white women, each measured as a percentage of the income of white men. Three time series are analyzed: me- dian wage and salary income of year- round full-time workers, 1955-1975; me- dian wage and salary income of all those with such income, 1948-1975; and median total money income, 1948-1975 (these are the entire periods for which data are avail- able on an annual basis; for information on the sources and construction of all vari- ables, see the Appendix). Each time series can be usefully distinguished from the others. The measure of the earnings of full-time workers is probably the best measure of what EEO legislation is de- signed to affect-opportunities available to full-time employed members of the labor force. Median earnings of all labor force participants captures changes in opportunities available to those employed part-time, voluntarily or otherwise. And median total money income, by including income from social security payments, welfare, dividends, interest, etc., is prob- ably the best measure of overall economic well-being (see Freeman, 1973:73; Mas- ters, 1975:142).8

As described above, past analyses have compared the incomes of nonwhite or black men with those of white men. Freeman also analyzes the determinants of black female income, measured as a proportion of white female income; no analyses have compared female income with white male income in the context of EEO impact.9

This analysis breaks precedent by com- paring the incomes of nonwhite and white women with those of white men. The jus- tification for doing this is straightforward. The. manifest purpose of the legislation was to help ensure that those who had been discriminated against would in the future be treated as well as white men. Because white men are the implicit com- parison group in the law, they should be the comparison group in any study of rela- tive incomes.10 In addition, comparing nonwhite women with white women, as has been customary, rather than with white men, exaggerates their progress, because they are being compared with a group that suffers itself from discrimina- tion (Bell, 1972:365; Figures 1 and 2 will make this very clear).

Explaining Income Differences

Changes in the relative incomes of blacks and whites and men and women customarily are seen as determined mainly by four types of variables: those measuring the relative productivity of members of the different groups, the overall demand for labor in the economy, changes in attitudes toward different groups, and EEO enforcement activity (see, for example, Arrow, 1972; Bell, 1974; Haworth et al., 1975; and, of course, Freeman, 1973, and Masters, 1975).1 All four will be included in this analysis.

Even if all labor market discrimination were to disappear immediately, blacks and women would not achieve economic equality with white men for a long time, 8 Determinants of mean total money income were

also analyzed, but the result did not differ in any material way from those concerning median total income, and have been excluded to shorten the pre- sentation. Means and medians are the best simple measures of group income, but they do not ade- quately represent all aspects of income distributions, such as increasing variance in income within social groups; see Wilson, 1978; Villemez and Rowe, 1975. These problems will be dealt with elsewhere.

9 Stevenson (1975) analyzes changes in white female income relative to that of white males, but does not consider the impact of EEO legislation. This analysis contrasts whites to nonwhites, not blacks. This is partly because the EEO laws prohibit all racial discrimination, not just discrimination against blacks, and partly because the data for much of the time series are available only for non- whites and whites. Because blacks are about 90%o of nonwhites, the results would be altered very little in any case.

10 One justification in past work for comparing women of different races with each other rather than with men was probably that this implicitly con- trolled for many differences between men and women in labor force behavior. Although some re- main convinced that many of these differences are voluntary (Gwartney and Stroup, 1973), the women's movement and the inclusion of sex dis- crimination in the EEO legislation makes this "vol- untariness" problematic. Because the legislation is intended to reduce many of the presumed differences in labor force behavior, it is inappropriate to build such presumptions into the analysis.

" Masters and others have tried to show that a variety of other possible influences, such as migra- tion and housing segregation, do not have a signifi- cant impact on black-white differences (Masters, 1975: chaps. 3-5).

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 373

because they (particularly blacks) have been denied equal access to education and training. Members of groups that have been discriminated against can expect to achieve equality in incomes only as they approach equality in productivity. In this analysis, the measure of relative prod- uctivity will be the ratio of median years of education completed by members of the relevant groups. This is a crude gauge of relative productivity, but is standard and practical (see Welch, 1973).

The economic progress of groups that have been discriminated against is widely seen as dependent upon tight labor mar- kets. Because white men can resist open- ing new job opportunities to nonwhites and women-through union seniority rules, on-the-job hostility and sabotage, etc.-nonwhite and female progress is much more likely when there are lots of jobs available (see, e.g., Bell, 1974). The demand for labor will be measured by the overall national unemployment rate.

Attitudes toward nonwhites and women in the labor force, taken as rough indices of tastes for discrimination, will be mea- sured in terms of the only three questions asked repeatedly in national surveys dur- ing most of the period since 1948. The first was asked of whites six times between 1944 and 1972: "Do you think Negroes should have as good a chance as white people to get any kind of job, or do you think white people should have the first chance at any kind of job?" The second was asked of adult Americans five times between 1942 and 1962: "Do you approve of paying women the same salaries as men, if they are doing the same work?" The third was asked five times between 1938 and 1976: "Do you approve of a mar- ried woman earning money in business or industry if she has a husband capable of supporting her?" Public opinion toward blacks and women in the labor force be- came steadily more favorable over time, making it possible to estimate attitudes for those years when the questions were not asked (R2's for regressions of attitudes on time were .99, .58, and .94 respectively; see Figure 3 and Appendix for details).12

12 These data are clearly not the best indicators of taste for discrimination that could be devised.

EEO enforcement activity was mea- sured in a variety of ways. The supply of administrative EEO enforcement was measured in terms of expenditures by the EEOC, expenditures per black and female member of the labor force, cumulated ex- penditures, expenditures per charge pro- cessed, and Equal Pay Act underpay- ments disclosed. Judicial enforcement was gauged by the annual and cumulative number and proportion of race and sex discrimination cases decided in favor of women or minorities by U.S. Courts of Appeals and the U.S. Supreme Court. The demand for EEO enforcement was mea- sured in terms of the number of actionable complaints of discrimination-race, sex, and total-processed by the EEOC each year. The aim was to gauge in many ways the intensity of the problem, the resources available to cope with discrimination- both total incidence and complaints-and the timing of effective implementation of the law.

Some Expectations

The basic hypothesis of the study is straightforward: everything else being equal, increases in the productivity of women and nonwhites, decreases in un- employment, decreases in discriminatory attitudes, and increases in the supply of and demand for EEO enforcement activi- ties should have independent positive ef- fects on the incomes of nonwhite men, nonwhite women, and white women rela- tive to those of white men.

One caveat is in order. Although neither Becker nor those who have tested his theory generally distinguish between race discrimination and sex discrimination (see Becker, 1971: 106-7), other people have argued that the two are different with different causes and consequences that bear on the analysis. As it turns out, the

Ideally, we would like annual data on attitudes of theoretically distinguishable groups-employees, employers, customers-toward various aspects of labor market relationships (willingness to work with members of particular groups, to have them as supervisors, etc.). But the data used are the only relevant time series data available. For a more gen- eral review of changes in attitudes toward blacks, see Burstein, forthcoming.

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

374 AMERICAN SOCIOLOGICAL REVIEW

hypothesis is generally supported by the evidence for nonwhite men and women, but not for white women. Some possible reasons for this will be discussed below.

FINDINGS

A Preliminary Picture of the Data

Figures 1 through 5 and Table 1 provide some essential preliminary information about the phenomena under considera- tion.

Figures 1 and 2 portray changes in the ratio of the incomes of nonwhite men, nonwhite women, and white women to those of white men. The figures show that nonwhite men and women have made considerable gains in relative income dur- ing the last 20 to 30 years, but that they still have a long way to go before catching up with white men. The figures show a slight upward trend for the entire period; the trend appears to have accelerated since Title VII went into effect in 1965, and may even have accelerated slightly again since 1972, when the law's enforce- ment provisions were strengthened. Thus, for example, the ratio of nonwhite to white male wage and salary incomes among year-round full-time employees rose only from .63 in 1955 to .66 in 1964, but then rose to .70 in 1972 and .77 in 1975.

The picture is quite different, however, for white women. The ratio of white female to male wage and salary incomes for full-time workers fell fairly steadily from approximately .64 in 1955 to .58 in 1974, and the ratio of median total money income has also fallen (the picture of earn- ings for all in the labor force is like the picture for full-time workers, and is left out to save space). The ratio for total money income stopped its decline and then began to rise after the passage of Title VII and the Equal Pay Act, but the ratio for full-time workers did not. White women have not been gaining from recent social and legal changes in the same way as nonwhite men and women. (The figures also demonstrate that analyses which compare nonwhite women with white women rather than white men will exag- gerate the progress of nonwhite women in the struggle against discrimination.)

Figure 3 presents estimated changes in public opinion on the labor force partici- pation of women and nonwhites since the 1940s. At least with regard to statements of principle, prejudice against women in the labor force seems to be a greater prob- lem than prejudice against blacks. Almost all whites agree that blacks should have an equal chance at jobs, and have for years; change since the end of World War II has been dramatic. Changes in attitudes toward female participation in the labor force have also been great, but consider- able antipathy remains; as recently as 1976, almost a third of the public still felt that women who had husbands who could support them should not work for pay (women were only slightly more favorable in their responses than men were). Never- theless, attitudes toward both blacks and women have become much more favor- able. This trend is consistent with the trend in nonwhite incomes (male and female), but is contrary to the trend in white female incomes.

Figure 4 documents the very significant increases in levels of education attained by nonwhites relative to whites. It also shows that white and nonwhite men have been catching up to the women of their own races educationally. The relative de- cline of educational attainment among white women is consistent with the de- cline in relative income documented above; whether such an interpretation makes sense substantively is a question that will be dealt with below. In any case, the educational levels of the four groups reached virtual parity by the mid-1970s.

Figure 5 gives some indication of the timing and intensity of the EEO enforce- ment process; it pictures the cumulative number of race and sex discrimination cases-brought under the four laws dealt with here-decided by U.S. Courts of Appeals and the Supreme Court. To the extent that a complex and controversial set of laws like the EEO laws can have substantial impact only after they have been interpreted by higher courts in a way offering some relief to the disadvantaged, the impact of the laws would be slow in coming. Although the Equal Pay Act was passed in 1963 and Title VII in 1964, with each taking effect a year later, no cases

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 375

0)~~~~~~a

as)An

0_Cd 4-4

0)

I 44 4

0~~~~~~~~~~~~~~~~~~~~0

Figure/1. Median Incomes (Wages or Salary) of Year-Round, Full-Time Workers, As Percentage of White Male Incomes

I~~~~~~~~~~~~~~~~~~~~~~~Y /~~~~~~~~~~~~~~~~~~~~~~~~~_

I 0'~~~~~~~~~~~~~~L 0'

0) / L

Figure 1. Median incomes (Wages or Salary) of Year-Round, Full-Time Workers, As Percentage of White Male Incomes

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

376 AMERICAN SOCIOLOGICAL REVIEW

Ln

r-4I~~~~~~~~r

~\\

H *r4~~~~~~~~~~r

I~~~~~~~~~~~~~~~- ,1)

*r4~~~~~~~~I-

100,~ ~ ~~~~0

'1~~~~~~~~~~~~~~~~!'

X . .1 _.____ M_ _______, I 4 ~~~~~~~~~~~4 r~~~~~~~~~~~~~~~~~~~~~~~~~~~-4

w ) Ln) 0 Ln)

Figure 2. Median Total Money Income, As Percentage of White Male Income

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 377

o4 c)

C r-4 0101

co e e

o v- 0)

4-4~~~~~~~4

0 r-4~~~~~~~440 -r4~~~~~~~~~~~0r

p-4 U)~~~~~~~~"4"

Figure 3. Public Opinion (Estimated) on Equal EmploymentOpportunity4

14-

0

o 0

'-4

Lj00

r-4

o C~~~~~~~~~nC n4

Figure 3. Public Opinion (Estimated) on Equal Employment Opportunity

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

378 AMERICAN SOCIOLOGICAL REVIEW

-V

\ \' \I / 0\< r-4~~~~~~~~~~r 0 0 01

0 ~~~~~~ v-40 ~~~~~~~~r4 %0~ jJ\ \ QC~

*1-4~~~~~~~~~r4O W ~~~\ 00 0

H o co o~~~~~~~~~~~~~c

Fiue .MdinYer f dcain A ecetgeo hieMaeEucto 4-) 'I~~~~~~~~~~~~~~~~0 4

~~~~~~ 0 0~~~~~~~~~~~~~~~~0

v-I

Figure 4. Median Years of Education, As Percentage of White Male Education

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 379

320

/ 240

/ 160

/~~~~ / ~~~//

80 / / / / '

, .

1968 1972 1976

Figure 5. Court Decisions under EEO Laws-Numbers Concerning and Favoring Minorities and Women

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

380 AMERICAN SOCIOLOGICAL REVIEW

under these statutes were decided by ap- pellate courts until 1967, and the numbers decided grew slowly for several years thereafter. The meaning of EEO legisla- tion and the possible magnitude of sanc- tions that could be applied to dis- criminators were not clear for years after the passage of the laws (and are still fuzzy in important areas).

Table 1 provides additional information about the EEO enforcement effort. Objec- tively, it is not obvious what would consti- tute an "adequate" enforcement effort in the short term. EEOC expenditures have risen every year, both in current and con- stant dollars, just about keeping pace with the increasing number of complaints ("ac- tionable charges") made to the Commis- sion. If we view EEOC expenditures as an investment in equal opportunity that will have effects beyond the year of the ex- penditures (Freeman, 1973:100), the total amount spent by the EEOC per black and female member of the labor force, cumu- lated over eleven years, was just under four dollars (1967 dollars). An investment of four dollars a person on a major social problem does not seem like a lot, espe- cially when contrasted with the amount spent on training programs (Masters, 1975:152), and it is, in fact, fairly widely agreed that the EEOC is very inade- quately funded considering the magnitude of its task. ' 3

The table also shows that the number of charges has continued to rise, taking a particularly big jump in 1972, the year the law was strengthened. To the extent that the number of charges reflects the number of grievances, employment discrimination is hardly declining as a problem. Finally,

< 0 3 0

U0 C~

? 0 . *E .,g?',_ cq cq e q 'Ic ? c

W

0 H :

<Us 4-~~~~~~~~~~~-

0 C)E 0S O ~~ c _b<b )

0~~~~~~~~

c E ;; 4 UC o _N W) 00 o0 00 W

M: 0

:~~~~ . .o . . .

O cq M oq ' 'IC 110 "T 0o} O <N

D . < - 0 O- tJd 0

0~~~~~~

_ .2 0

= C~~~-

LL WC.) -W (O vt

0 oC) *&- W Ft-..O CZ)

C.)~~~~~~~~~~

~~~ ~~~~ .0 ~~~~~~0

u CZ &c C CZ 4 a ) 0 W) C r P Q

<H - C1

CIS cis P ~ ~ ~ ~ ~ ~

Hi6(O z p

'3 See Levine and Montcalmo, 1971; U.S. Com- mission on Civil Rights, 1977: 329-35. EEOC expen- ditures are just a part of total government expendi- ture on equalizing employment opportunities (see U.S. Office of Management and Budget, 1978:275- 87), and are used here as an indicator of activity rather than as an inclusive measure (more inclusive data are not available for the entire time period). Politically, it is probably not an accident that the EEOC has been underfunded. Some members of Congress may have underestimated the magnitude of the EEOC's task and felt they were voting sufficient funds, but it is also true that, once Title VII passed, its opponents worked hard to minimize the resources available to the EEOC (see, e.g., U.S. Senate, 1970: 1043ff).

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 381

the table provides evidence that sex dis- crimination lagged behind race discrimina- tion as an object of concern, both among complainants to the EEOC and in the courts, but that it is becoming an increas- ingly important part of the administrative and judicial caseload. In light of the fig- ures on income presented above, this is not surprising.

In sum, the incomes and educational levels of nonwhite men and women have improved relative to those of white men, and attitudes toward blacks have become more favorable. The income and educa- tional picture for white women is much less positive. The EEO enforcement effort was slow to begin and seems modest in intensity, but the demands placed upon the system keep increasing.

Multivariate Statistical Models and Analyses

In order to gauge the impact of likely determinants of income ratios, the non- white male/white male, nonwhite female/white male, and white female/ white male income ratios were regressed on the measures of relative education, un- employment, attitudes, and EEO en- forcement activity. There proved to be serious problems of multicollinearity among some variables, however.

Because all three measures of change in attitudes were simple time trends, they were to all intents and purposes perfectly correlated with each other, so no more than one could be used in any equation.

Changes in relative education were very highly correlated with the passage of time as well, and so were correlated very highly with changes in attitudes. The cor- relation of the nonwhite male/white male education ratio with attitudes on black ac- cess to jobs was .97, for example, and introducing both variables into a regres- sion simultaneously produced the out- comes that would be expected in the presence of multicollinearity-large and substantively unlikely variances in the pa- rameter estimates, insubstantial changes in R2 when variables were added or re- moved, and decreases in computational accuracy (see the useful discussion in Rao

and Miller, 1971:46-52). This high correla- tion is probably not a coincidence, and a study of the relationship between the liberalization of attitudes and black access to education would be interesting itself. Within the data set and context, however, there was no way to disentangle the ef- fects of the two variables. The same prob- lem arose with regard to nonwhite women. With regard to white women, liberalization of attitudes has gone along with a decline in relative educational at- tainment. Substantively, this probably implies that changes in years of education is not a good measure of changes in prod- uctivity so far as white women are con- cerned (Madden, 1978), but in this case the negative correlation between public attitudes and educational attainment was so high that both variables could not be used in the equations simultaneously.

In the equations presented, therefore, measures of attitudes and of education are never included in the same equation (cf. Freeman, 1973:101). The extremely high correlation of the variables with each other and with the passage of time reduces the degree to which the results can answer theoretical questions about their separate effects. This is itself a significant finding, however, if only because it implies that past studies may have attributed to changes in education some consequences that were in fact the result of changes in public opinion. The multicollinearity prob- lem does not reduce the predictive value of the equations, of course.

Similar problems arise with regard to the measures of EEO enforcement effort. Although measures such as court deci- sions, EEOC expenditures, and charges of discrimination can be distinguished theoretically, and although there is no theoretical reason why they should be very highly correlated, in fact nearly all the measures proved to be simple time trends very highly correlated with each other. This was true for EEOC expendi- tures, expenditures per covered worker, cumulated expenditures, Equal Pay Act payments, annual and total court deci- sions, charges handled by the EEOC, ex- penditures per charge, and other mea- sures. Here, as with education and public attitudes, it proves impossible to assess

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

382 AMERICAN SOCIOLOGICAL REVIEW

separately the effects of different interest- ing variables.

The following variables were included in at least some of the ordinary least- squares regressions actually run. For nonwhite men: the national unemploy- ment rate (denoted Unempl in the tables); the ratio of their education to that of white men (Educ); the measure of attitudes toward blacks' access to jobs (Attitude); the percentage of court cases dealing with racial discrimination decided in favor of minorities each year, lagged one year, in- cluded because employers might get a sense for how court enforcement has been proceeding in the short term and respond accordingly (% Cases); cumulated EEOC expenditures per nonwhite and female member of the labor force, in constant dollars (EEOC $); the cumulated number of court decisions favoring minority com- plainants (N cases); and the number of actionable charges based on race pro- cessed annually by the EEOC (Charges).

For nonwhite women. In theory, non- white women should gain from efforts to improve the economic situation of non- whites and of women; in practice, because of the multicollinearity problems just de- scribed, it was not possible to separate the effects. The same variables were used for nonwhite women as for nonwhite men, with the exception of education, which was the ratio of nonwhite female educa- tion to white male education.

For white women. The following vari- ables were used: the unemployment rate; the ratio of their education to that of white men; the measure of attitudes toward equal pay for equal work for woman; the percentage of court cases dealing with sex discrimination decided in favor if women each year, lagged one year; cumulated EEOC expenditures per nonwhite and female labor force member; cumulated number of court decisions favoring female complainants; and the number of actiona- ble charges based on sex processed an- nually by the EEOC.

The results are presented in Tables 2 to 4. Table 2 presents the results for non- white men ("D-W d' is the Durbin- Watson d statistic). Overall, the results are consistent with the hypothesis that the demand for labor, educational attainment,

public attitudes, and EEO enforcement ef- forts all have a significant impact on the relative incomes of nonwhite men.'4 It is, however, as noted above, impossible to disentangle the effects of education from those of attitudes, or of the various mea- sures of enforcement effort from each other. '5 Perhaps the most important result is that the impact of EEO legislation is not absorbed into the attitude variable-EEO laws do have an independent effect.'6

Interpreting the coefficients concretely, the coefficients for education cluster around approximately .20, meaning that a 5% increase in the ratio of nonwhite male education to white male education re- sulted in a 1% increase in the income ratio, everything else in the equations being equal. With nonwhite male educa- tion having improved from 71% of white male education in 1948 to 96% in 1975, an increase of approximately 5% in the in- come ratio could be attributed to the in- crease in education. Similarly, an increase of about 8% of the white population feel- ing that blacks should have an equal chance at jobs seems to be associated with a 1% increase in the income ratio.

The unemployment rate has no signifi- cant impact on the relative income of full- time workers, as might be expected, since increases in unemployment would simply

14 The t-statistics should be interpreted cautiously.

Berk and Brewer (1977) have argued that statistical inference&in many econometric models rests upon untested and even untestable assumptions. Tests of statistical significance in this article mean just what they do in the other relevant studies.

'5 Both Freeman and Masters include time-trend variables in their analyses. As a practical matter, the estimates of EEO impact derived here for nonwhite men will not differ all that much from Freeman's, because the attitude variables included in this study are so highly correlated with the time trend variable he included in his equations. Theoretically, however, the use of attitude data is clearly preferable-it plays a definite role in the economic theory of discrimina- tion, while time is not a very useful concept theoreti- cally. In addition, in those situations where attitudes and time are not highly correlated, the coefficients in equations not including attitudes would be biased (see Pfister, 1974:189).

6 Although the attitude change was probably partly responsible for the passage of the EEO laws, the relationship between laws and attitudes would probably become more complex once the laws are passed, with changing attitudes aiding enforcement and the enforcement of laws hastening attitude change.

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 383

Table 2. Determinants of Nonwhite Male Income As Percent of White Male Income, Alternative Specifica- tions

Independent Variables-unstandardized regression coefficients (t-statistics in parentheses)

EARNINGS OF YEAR-ROUND, FULL-TIME WORKERS Eq. Constant Unempl Educ Attitude % Cases EEOC $ N Cases Charges R2 D-W d

1 46. .49 .21 .... .010 1.22 .... .... .87 1.96 (6.32) (1.20) (2.50) (.63) (1.61)

2 51. .33 .... .16 .0078 1.85 .... .... .87 2.10 (9.40) (.86) (2.66) (.47) (3.16)

3 43. .62 .25 .... .014 .... .023 .... .86 1.84 (6.05) (1.44) (3.13) (.87) (1.09)

4 49. .39 .... .18 .016 .... .046 .... .86 1.97 (8.75) (.95) (3.05) (.92) (2.60)

5 43. .74 .24 .... .0098 .... .... .000075 .86 1.79 (5.60) (1.96) (2.70) (.55) (.91)

6 49. .62 .... .16 .0066 .... .... .00016 .85 1.86 (8.29) (1.64) (2.47) (.35) (2.36)

EARNINGS OF ALL WORKERS

Eq. Constant Unempl Educ Attitude % Cases EEOC $ N Cases Charges R2 D-W d 1 51. -.99 .16 .... .026 3.17 .... .... .84 2.11

(7.12) (-2.33) (1.73) (1.09) (3.56) 2 58. -1.14 .... .083 .027 3.76 .... .... .84 2.18

(15.69) (-2.72) (1.82) (1.19) (5.14) 3 47. -.94 .21 .... .038 .... .076 .... .82 2.00

(6.61) (-2.06) (2.36) (1.57) (2.98) 4 56. -1.14 .... .11 .046 .... .099 .... .81 2.06

(14.50) (-2.48) (2.20) (2.01) (4.43) 5 51. -.72 .15 .... .018 .... .... .00031 .82 2.02

(6.41) (-1.70) (1.44) (.71) (3.00) 6 57. -.81 .... .067 .019 .... .... .00037 .82 2.06

(14.57) (-1.91) (1.35) (.74) (4.51)

MEDIAN TOTAL INCOME Eq. Constant Unempl Educ Attitude % Cases EEOC $ N Cases Charges R2 D-W d

1 44. -1.51 .22 .... .0033 2.49 .... .... .89 2.56 (8.51) (-4.96) (3.44) (.20) (3.91)

2 55. -1.70 .... .086 .012 3.45 .... .... .86 2.32 (18.89) (-5.14) (2.41) (.67) (5.98)

3 41. -1.49 .26 .... .013 .... .063 .... .88 2.49 (8.18) (-4.67) (4.12) (.75) (3.51)

4 54. -1.72 .... .11 .029 .... .093 .... .84 2.24 (17.55) (-4.75) (2.82) (1.58) (5.28)

5 44. -1.30 .21 .... -.0028 .... .... .00025 .87 2.33 (7.69) (-4.29) (2.89) (-.15) (3.35)

6 55. -1.42 .... .070 .0028 .... .... .00035 .85 2.14 (17.88) (-4.29) (1.80) (.14) (5.49)

cause people to drop out of that category; the rate does have a significant negative impact on the other measures of income, also as predicted. This negative impact is made up for by the higher coefficients associated with EEO enforcement vari- ables in the equations for all workers and all those with income. For all those with earnings, for example, a cumulated dollar spent by the EEOC leads to more than a 3% increase in the income ratio. In paral- lel fashion, every ten or twelve favorable court decisions may be interpreted as

associated with a 1% increase in the in- come ratio, as may every 3,000 actionable charges.

Examining the earnings of all workers in the context of equation 1, we note that changes in educational levels since Title VII went into effect have been associated with a two and a quarter percent increase in the income ratio (coefficient of .16 times increase in the education ratio of 14%), while EEOC expenditures would be associated with a rise of 12%, everything else being equal (of course, everything

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

384 AMERICAN SOCIOLOGICAL REVIEW

else was not equal, since the unemploy- ment rate almost doubled in the interim). EEO enforcement activities appear to have had substantial impact.

The results for nonwhite women, presented in Table 3, are similar to those for nonwhite men. The proportion of vari- ance explained is slightly greater, and nonwhite female incomes, unlike male in- comes, seem to be sensitive in some cases to court decisions calculated on an annual basis. These possibilities must be treated very cautiously, however, because the

Durbin-Watson d statistics are in some cases in the range where serial correlation in the errors cannot be ruled out with a high degree of probability. It is also worth noting that the median total income of nonwhite women does not seem to be re- sponsive to changes in the unemployment rate.

The results for white women presented in Table 4 provide some fairly dramatic contrasts with the findings for nonwhite men and women, and indicate that the util- ity of the hypotheses being tested is lim-

Table 3. Determinants of Nonwhite Female Income As Percent of White Male Income, Alternative Specifi- cations

Independent Variables-unstandardized regression coefficients (t-statistics in parentheses)

EARNINGS OF YEAR-ROUND, FULL-TIME WORKERS Eq. Constant Unempl Educ Attitude % Cases EEOC $ N Cases Charges R2 D-W d

1 8.9 .40 .34 .... .037 2.26 .... .... .91 1.47 (.92) (.82) (3.35) (1.81) (3.00)

2 16. .28 .... .28 .030 2.24 .... .... .93 1.52 (2.72) (.66) (4.15) (1.60) (3.37)

3 3.9 .55 .39 .... .047 .... .052 .... .90 1.48 (.38) (1.02) (3.75) (2.23) (2.28)

4 14. .39 .... .32 .040 .... .053 .... .92 1.51 (2.10) (.81) (4.47) (2.02) (2.59)

5 5.5 .77 .38 .... .035 .... .... .00019 .90 1.55

(.52) (1.59) (3.20) (1.50) (2.29) 6 14. .65 .... .29 .029 .... .... .00018 .91 1.56

(2.06) (1.48) (3.75) (1.34) (2.41)

EARNINGS OF ALL WORKERS Eq. Constant Unempl Educ Attitude % Cases EEOC $ N Cases Charges R2 D-W d

1 -6.4 -.57 .42 .... .046 3.17 .... .96 1.73 (- 1.28) (-2.03) (7.34) (2.88) (6.19)

2 15. -.79 .... .21 .055 3.63 .... .95 1.38 (5.00) (-2.36) (5.76) (3.04) (6.24)

3 -9.7 -.57 .46 .... .060 .... .084 .... .95 1.70 (- 1.84) (-1.81) (7.61) (3.61) (5.33)

4 13. -.78 .... .23 .074 .... .096 .... .93 1.39 (4.16) (-2.08) (5.85) (3.89) (5.22)

5 -5.0 -.35 .39 .... .036 .... .... .00033 .96 1.67 (-.97) (- 1.26) (6.47) (2.12) (6.04)

6 15. -.51 .... .19 .043 .... .... .00038 .95 1.34 (4.98) (-1.60) (5.06) (2.24) (6.21)

EARNINGS OF ALL WORKERS Eq. Constant Unempl Educ Attitude % Cases EEOC $ N Cases Charges R2 D-W d

1 -4.0 -.17 .31 .... .026 2.11 .... .... .93 1.66 (-.91) (-.65) (5.91) (1.82) (4.57)

2 11.0 -.32 .... .15 .033 2.44 .... .... .92 1.52 (4.44) (-1.11) (4.90) (2.12) (4.92)

3 -6.0 -.16 .33 .... .035 .... .0056 .... .93 1.63

(- 1.38) (-.58) (6.32) (2.47) (4.08) 4 10.0 -.31 .... .17 .045 .... .064 ... . .91 1.50

(3.87) (- 1.01) (5.12) (2.90) (4.25) 5 -2.8 -.026 .28 .... .018 .... .... .00023 .94 1.66

(-.63) (-.11) (5.40) (1.25) (4.85) 6 11. -.14 .... .14 .023 .... .... .00026 .92 1.51

(4.66) (-.55) (4.44) (1.47) (5.24)

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 385

Table 4. Determinants of White Female Income As Percent of White Male Income, Alternative Specifica- tions

Independent Variables-unstandardized regression coefficients (t-statistics in parentheses)

EARNINGS OF YEAR-ROUND, FULL-TIME WORKERS Eq. Constant Unempl Educ Attitude % Cases EEOC $ N Cases Charges R2 D-W d

I - 15. .35 .72 .... .0024 -.53 .... .... .93 1.89

(-2.28) (2.60) (10.90) (.31) (-2.42) 2 110. -.15 .... -.56 .025 -.097 .... ... . .91 1.45

(20.02) (-.97) (-9.45) (2.63) (-.37) 3 - 17. .39 .73 .... -.032 .... -.031 .... .94 2.02

(-2.60) (2.98) (11.69) (-.05) (-2.85) 4 109. -.11 .... -.56 .026 .... -.011 .... .91 1.42

(20.79) (-.70) (-9.79) (2.97) (-.83) 5 -15. .34 .71 .... .0021 .... -.00011 .94 2.01

(-2.31) (2.86) (11.57) (.33) (-3.07) 6 109. -.13 .... -.55 .026 .... .... .000037 .91 1.43

(20.16) (-.84) (-9.40) (2.97) (- .86)

EARNINGS OF ALL WORKERS Eq. Constant Unempl Educ Attitude % Cases EEOC $ N Cases Charges R2 D-W d

1 -45. .55 .89 .... -.0029 -.71 .... .... .81 .33

(-3.41) (1.53) (7.50) (-.19) (- 1.05) 2 115. -.13 .... -.74 .031 -.041 .... .... .92 1.04

(23.82) (-.61) (-13.26) (1.96) (-.09) 3 -46. .60 .90 .... -.0069 .... -.040 .... .81 .32

(-3.52) (1.61) (7.63) (-.34) (- 1.14) 4 115. -.098 .... -.74 -.034 .... -.0098 .... .92 1.02

(24.2) (-.44) (-13.4) (2.36) (-.44) 5 -44. .54 .89 .... - .0048 .... -.00013 .81 .32

(-3.36) (1.52) (7.46) (-.22) (- 1.11)

6 115. -.13 .... -.74 .032 .... .... .- 000016 .92 1.03 (23.69) (-.58) (-13.2) (2.16) (-.21)

MEDIAN TOTAL INCOME Eq. Constant Unempl Educ Attitude % Cases EEOC $ N Cases Charges R2 D-W d

1 -66. -.080 .96 .... -.0065 2.01 .... .... .70 .39

(-4.16) (-.18) (6.66) (-.22) (2.45) 2 106 -.82 .... -.80 .030 2.73 .... .84 1.24

(15.46) (-2.64) (-10.04) (1.32) (4.46) 3 -62. .13 .93 .... .011 .... .097 .... .69 .39

(-3.85) (-.28) (6.34) (.43) (2.23) 4 104. -.85 .... -.76 .053 .... .13 .... .81 1.16

(14.25) (-2.49) (-9.04) (2.41) (3.79) 5 -68. - .0068 .98 .... .026 .... .... .00034 .69 .38

(-4.25) (-.016) (6.70) (.096) (2.39) 6 106. -.73 .... -.81 .043 .... .... .00046 .83 1.18

(15.18) (-2.37) (-9.94) (1.97) (4.29)

ited unless they are specified better. As shown in Figures 1, 2, and 3, white female incomes have declined relative to those of white men despite increases in the inci- dence of attitudes favorable to female labor force participation on an equal basis; in addition, the decline continued, at least for a while, in the face of legislative change and increasing EEO enforcement efforts. These facts are reflected in the coefficients, which are negative for public attitudes in every equation, and are nega- tive or insignificant for EEO enforcement efforts in some of the equations. It may be

possible, however, that the reversal of the downward trend in total money income has been due to EEO enforcement ac- tivity. The coefficients for education re- main strong and positive, as in Tables 2 and 3. Because relative white female edu- cation has been going down, declining in- comes could be attributed to changes in relative education. Such an interpretation seems far-fetched, however, if only be- cause initial white female educational levels were higher than those for men, while income levels hardly were.

In addition to the unexpected coeffi-

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

386 AMERICAN SOCIOLOGICAL REVIEW

cients, the measures of R2 and the Durbin-Watson d statistic indicate that there is something wrong with the model as it applied to white women. R2 is notably lower in the equation for white women than it is for other groups. More signifi- cantly, the d statistic indicates the likely presence of substantial serial correlation in the error terms; such a result is often interpreted as an indication that important variables have been left out of the equa- tion. 17 A statistical model that worked reasonably well for nonwhites does not appear nearly as useful for explaining changes in the relative incomes of white women.

DISCUSSION

As expected, the unemployment rate generally has had a significant impact on the incomes of nonwhites. In addition, EEO enforcement activities have had a significant impact on their incomes. Un- fortunitely, the extremely high correla- tions between changes in attitudes and educational levels made it impossible to estimate their impacts separately; at this point, it is impossible to say how much of the change in nonwhite income has been due to each. The situation with regard to various measures of EEO enforcement ac- tivity is similar. Since the mid-1960s, im- provements in nonwhite education, changes in attitudes and the law, and EEO enforcement activity all seem to form a historical "package." This collection of closely related forces has led to the im- provement of nonwhite incomes, but it is difficult at a number of points to be very clear about the causal relationships among the variables or to gauge independent im- pact very precisely. The passage of time, providing more data points, the develop- ment of better measures of educational quality and attitude change, and studies at lower levels of aggregation may lead to analyses with statistically neater out- comes.

17 Generalized least-squares estimation could be used to overcome much of the effect of serial correla- tion, but in this context that would miss the point, which is that a particular statistical model derived from prior theoretical and empirical work does not work well when applied to a comparison of groups not hitherto examined.

The major problem raised by the find- ings has been the difference between the results for white women and those pre- dicted for all groups and actually esti- mated for nonwhites. What kind of story could one tell that would explain the re- sults and improve the specification of the model?

A great deal has been written about the similarities and differences between race and sex discrimination (for a general re- view, see Chafe, 1977: chaps. 3-4). There is no objective way to conclude which kind of discrimination is "worse"- although if one's standard is equality of results, sex discrimination seems worse because white women have lower incomes than nonwhite men, and nonwhite women have the lowest of all-nor is it possible at this stage to specify rigorously the relative difficulties to be overcome or the re- sources available to the different groups. Nevertheless, there are a number of fac- tors, some operating before the passage of the laws, some built into the operation of the laws, and some occurring as a conse- quence, which can be described as making sex discrimination more difficult to over- come than race discrimination.

Among preexisting factors are the fol- lowing: There is some evidence that labor market discrimination against women is worserse' than discrimination against nonwhites, in that income differences by race have been attributed largely to higher black levels of unemployment and re- stricted access to many occupations, while women suffer from these problems and from lower wages within job categories (Strauss and Horvath, 1976). Women may have had more education than men, on the average, but the educa- tion may have been less useful as women shied away from the acquisition of practi- cal skills due to restrictions on their labor force activity (Madden, 1978). Women may be less unified in their interests than blacks, because the gains of some women may be at the expense of white men whose wives will suffer as a consequence (Alexis, 1974). Finally, white women may find it more difficult than nonwhites to make the same gains relative to white men simply because there are so many more white women; the changes in social struc-

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 387

ture implied by equal opportunity for white women are much greater than those implied by equal opportunity for blacks (Stem et al., 1976).

Thus, there are a number of preexisting factors that would make it more difficult for women to achieve equality. There are compensating factors, of course. Greater numbers can lead to greater political strength, if unity can be achieved, for example (Alexis, 1974), and white women can have the relatively easy access of white men to informal socialization, useful contacts, etc., when attitudes change. The empirical problem will be determining the net effect of each.

The law itself as it applies to women also makes their situation more difficult than that of nonwhites. Although histori- cally both blacks and women have been subject to many legal restrictions on their activities, blacks have suffered from fewer legal disabilities than women since the mid-1960s. The differences are both con- stitutional and statutory. The Supreme Court has interpreted the Constitution so as to make race a "suspect classification" -laws that distinguish be- tween races are inherently suspect, and will be struck down unless justified by a compelling state interest. Distinctions be- tween the sexes have not attained this status, so arguments against sex discrimi- nation rest on a shakier constitutional foundation than arguments against race discrimination (see Ginsburg, 1976; Get- man, 1973).

Two major statutory issues have prob- ably slowed the positive impact of EEO legislation on women. First, the relation- ship between Title VII and state protec- tive legislation which limited certain labor force activities of women was initially un- clear. It is now clear that Title VII over- rides such legislation, but female progress was probably slowed somewhat by the uncertainty (see Gates, 1976). Second, as noted above, women can still be treated unequally in those cases where sex is a "bona fide occupational qualification" for a job; race is never considered a bona fide occupational qualification. Although the EEOC and the courts have tended to interpret this provision narrowly, it still provides a loophole in the law.

Finally, response to the laws may differ between groups, and it is in considering differential response that probably the most immediately useful way of dealing with white women is to be found.

The important variable is probably the rate of labor force participation. During the 1948-1975 period, rates of labor force participation among white women have in- creased considerably, rising from 33.3% in 1954 to 37.5% in 1964 and 45.4% in 1975. Given that a high proportion of the new entrants would have had few skills and little experience, it would not be surpris- ing if the average earnings of all labor force participants went down, and that is what happened. If the elasticity of the supply of white female labor were high, and women were optimistic about oppor- tunities increasing as a result of changes in legislation, the size of the influx of white women into the labor force could have outpaced the increase in opportunities, and continued to bring median earnings down even after the passage of EEO legis- lation. This picture would be consistent with the differences in trends in earnings as opposed to total money income de- scribed above (see Figures 1 and 2). The entry of inexperienced white women into the labor force could bring down the me- dian earnings of those who are working, but increase the median total incomes of all women, since the earnings of the new entrants would go from zero to some higher figure. And, in fact, total money incomes have risen noticeably since 1963-64, even while median earnings con- tinued to decline.

The situation has been quite different for nonwhite men and women. The labor force participation rates of nonwhite men have actually declined in recent years (like white male rates), going from 85.2% in 1954 to 70.3% in 1975, with most of the decline concentrated among older men. The participation rate of nonwhite women has been very stable; it was 46.1% in 1954, 48.5% in 1964, and 48.7% in 1975. For nonwhite men and women, therefore, changes in income due to EEO legislation and changing attitudes would be easier to detect because they would not be con- founded with changes caused by trends in labor force participation.

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

388 AMERICAN SOCIOLOGICAL REVIEW

This discussion raises an additional question, however. In response to what could be seen as increasing economic opportunities brought about by changing attitudes and legislation, white women in- creased their labor force participation, nonwhite women did not change theirs, and nonwhite men actually withdrew from the labor force. That is, each of the three groups in question reacted differently to changing circumstances.

A plausible partial explanation for the behavior of white women might be that they were the group with the most to gain from increased labor force participation in the short term. Their initial rate was low- est, so could increase most easily, and the initial level of education was high enough to promise better jobs than would be available to many nonwhites, whose edu- cational level was lower. With regard to nonwhite women, it might be that in- creased opportunities for nonwhite men led some nonwhite women to drop out of the labor force, partly compensating for the arrival of new entrants. And, with lower educational levels to start with, new nonwhite entrants might have had less to gain than whites.

The model tested above fails when con- fronted with a group whose labor supply seems to be the most elastic of all groups considered, whose level of formal educa- tion was high, but whose labor force expe- rience and practical training was low. It seems reasonable to hypothesize that the initial model and hypotheses about the impact of attitude change and EEO legis- lation would be supported by the evidence for all groups if the model were revised to include measures of labor force experi- ence and actual skill levels. Alternatively, the model tested above should be sup- ported once all other groups reach equilib- rium levels of relative labor force participa- tion and skills.

CONCLUSIONS

This paper has argued for the inclusion of changes in attitudes in the study of changes in the economic situation of women and nonwhites, for realism in the examination of laws' impact, and for comparing the economic status of all

groups protected by EEO legislation with white men, who are the criterion group in the legislation. The findings for nonwhites generally provide evidence consistent with the revised model, although it was impossible to test some hypotheses about the independent impact of variables be- cause of multicollinearity. The model breaks down noticeably, however, when white women are considered, leading to the conclusion that it was improperly specified and should be revised to take labor force participation rates, work expe- rience, and skill levels into account.

Although nonwhite men and women have made major gains relative to white men in recent years, and white women show some signs of reversing a downward trend, the income gaps between groups are still very large. The gains thus far may have been the easy ones, involving the elimination of especially crude and blatant forms of discrimination. As discrimination becomes more subtle, as structural dis- crimination becomes a more important part of total discrimination, and as the legal system is confronted more and more often with conflicts between fundamental values-such as nondiscrimination and af- firmative action-further gains may be- come more difficult to achieve.'8

APPENDIX

DATA SOURCES AND VARIABLE DEFINITION

Income. Median wage or salary income of year- round, full-time workers 14 years old or over, by race and sex: U.S. Bureau of the Census, 1976: 459; U.S. Bureau of the Census, Current Population Re- ports, 1977: Ser. P-60, No. 105. Median wage or salary income of person 14 years old and over: U.S. Bureau of the Census, Current Population Reports, 1970-77: Ser. P-60, Nos. 69, 75, 80, 85, 90, 97, 101, 105. Median total money income: U.S. Bureau of the Census, Current Population Reports, 1977: No. 105.

Unemployment. U.S. Bureau of the Census, 1977: 379. Size of labor force: U.S. Department of Labor, 1975:31-4; 1977:25-8.

Implicit price deflator . U.S. Bureau of the Census, 1976:379.

Education. Median years of school completed, by race and sex: for 1947, 1950, 1960, U.S. Bureau of

18 The most sophisticated and comprehensive analysis linking legal and sociological aspects of em- ployment discrimination is that of Fiss (1971); an analysis of the problems in the context of recent court cases is found in Venick and Lane (1977).

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 389

the Census, 1975:380-1; 1952, U. S. Department of Labor, 1974: 299-301; 1957, U.S. Bureau of the Cen- sus, Current Population Reports, 1958: Ser. P-20, No. 77; 1958, 1961, U.S. Bureau of the Census, Current Population Reports, 1960-61: Ser. P-60, Nos. 33, 39; 1959, 1962, 1964-1975, U.S. Department of Labor, 1978:247-9. Cf. Freeman, 1973. Other years interpo- lated.

Attitudes. Initial data points: "Do you think Neg- roes should have as good a chance as white people to get any kind of job, or do you think white people should have the first chance at any kind of job?" Sources: May 1944, NORC Survey 225; May 1946, NORC Survey 241; April 1947, NORC Survey 150; December 1963, NORC Survey 330; June 1966, SRS Survey 889A; March 1972, NORC Survey 9001 (Hastings and Southwick, 1974); Schwartz, 1967:133.

"Do you approve of paying women the same salaries as men, if they are doing the same work?" Sources: January 1942, AIPO Survey 259; Septem- ber 1945, AIPO Survey 356; April 1946, RFOR Sur- vey 54; April 1954, AIPO Survey 530; June 1962, AIPO Survey 660 (Hastings and Southwick, 1974).

"Do you approve of a married woman earning money in business or industry if she has a husband capable of supporting her?" Sources: October 1938, AIPO Survey 136; October 1945, AIPO Survey 359, June 1970, AIPO Survey 808; March 1972, NORC Survey 9001; March 1976, Gallup Opinion Index 128 (Hastings and Southwick, 1974; Gallup Opinion In- dex, 1976).

The responses were regressed on time and ordi- nary least squares estimates for each year used. Fig- ure 3 presents the estimates. Maximum value permit- ted was 95%. Cf. Burstein and Freudenburg, 1978.

Equal Employment Opportunity Commission. An- nual appropriations: Equal Employment Opportu- nity Commission, 1976:31-2. Charges: EEOC an- nual reports, 1967-1976; reports 8 and 9 (fiscal years 1973 and 1974) use a different reporting system than the other reports, so charges were estimated by in- terpolation.

Equal Pay Act underpaynments disclosed. U.S. Department of Labor, 1975:424;1977:318.

Court cases. Decisions reported in Bureau of Na- tional Affairs (1969-77) for U.S. Courts of Appeals and U.S. Supreme Court coded on annual basis; cases interpreting the four laws and dealing with race or sex discrimination only; cases in which all nontri- vial points of law won by women or minorities coded 1, cases in which all nontrivial points lost coded 0, mixed results coded 1/2, in counting victories. The small number of reverse discrimination cases won by white men excluded from calculations. Detailed cod- ing rules available from author.

REFERENCES

Adams, Arvil 1972 Toward Fair Employment and the EEOC.

Washington, D.C.: Equal Employment Opportunity Commission.

Alexis, Marcus 1974 'The political economy of labor-market

discrimination." Pp. 63-83 in G. von

Furstenberg, A. Horowitz, and B. Harrison (eds.), Patterns of Racial Discrimination. Lexington: Heath.

Arrow, Kenneth 1972 "'Models of job discrimination." Pp. 83-102

in A. Pascal (ed.), Racial Discrimination in Economic Life. Lexington: Lexington Books.

Becker, Gary 1971 The Economics of Discrimination. 2nd ed.

Chicago: University of Chicago Press. Bell, Derrick, Jr.

1977 "Forward: equal employment law and the continuing need for self-help." Loyola University of Chicago Law Journal 8:681- 6.

Bell, Duran 1972 "Occupational discrimination as a source of

income differences: lessons of the 1960s." American Economic Review 62:363-72.

1974 "The economic basis of employee discrimi- nation." Pp. 121-35 in G. von Furstenberg, A. Horowitz, and B. Harrison (eds.), Pat- terns of Racial Discrimination. Lexington: Heath.

BelIer, Andrea 1977 "The effect of women's earnings on en-

forcement in Title VII cases." Monthly Labor Review 100:56-7.

Berger, Morroe 1967 Equality by Statute. Rev. ed. New York:

Doubleday. Bergmann, Barbara and Jerolyn Lyle

1971 "The occupational standing of Negroes by areas and industries." Journal of Human Resources 6:411-33.

Berk, Richard and Marilynn Brewer 1977 "Feet of clay in hobnail boots: an assess-

ment of statistical inference in applied re- search." Paper presented to the annual meeting of the American Sociological Association, Chicago.

Blau, Francine, and Carol Jusenius 1976 "Economists' approaches to sex segrega-

tion in the labor market." Signs 1:181-99. Bureau of National Affairs 1969- Fair Employment Practice Cases. Vols. 1978 1-15. Washington, D.C.: BNA.

Burstein, Paul Forth- "Public opinion, demonstrations and the comingpassage of antidiscrimination legislation."

Public Opinion Quarterly. Burstein, Paul and William Freudenburg

1978 "Changing public policy: the impact of pub- lic opinion, anti-war demonstrations. and war costs on Senate voting on Vietnam war motions." American Journal of Sociology 84:99-122.

Chafe, William 1977 Women and Equality. New York: Oxford

University Press. Equal Employment Opportunity Commission

1966- Annual Report. Washington, D.C.: U.S. 1975 Government Printing Office.

Farley, Reynolds 1977 "Trends in racial inequalities: have the

gains of the 1960s disappeared in the

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

390 AMERICAN SOCIOLOGICAL REVIEW

1 970s?" American Sociological Review 42:189-208.

Featherman, David and Robert Hauser 1976 'Sexual inequalities and socioeconomic

achievement in the U.S., 1962-1973." American Sociological Review 41:462-83.

Federal Reporter 1971 2nd Ser. St. Paul: West.

Fiss, Owen 1971 "A theory of fair employment laws." Uni-

versity of Chicago Law Review 38:235-314. Freeman, Richard

1973 "Changes in the labor market for black Americans, 1948-72." Brookings Papers on Economic Activity 1:67-120.

Gallup Opinion Index 1976 Princeton.

Gates, Margaret 1976 "Occupational segregation and the law."

Signs 1:61-74. Getman, Julius

1973 "The emerging constitutional principle of sexual equality." Pp. 157-80 in P. Kurland (ed.), The Supreme Court Review 1972. Chicago: University of Chicago Press.

Ginsburg, Ruth 1976 "Gender in the supreme court: the 1973 and

1974 terms." Pp. 1-24 in P. Kurland (ed.), The Supreme Court Review 1975. Chicago: University of Chicago Press.

Gitt, Cynthia and Marjorie Gelb 1977 "Beyond the equal pay act: expanding

wage differential protections under Title VII." Loyola University of Chicago Law Journal 8:723-66.

Gunderson, Morley 1975 "Male-female wage differentials and the

impact of equal pay legislation." Review of Economics and Statistics 57:462-9.

Gwartney, James and Richard Stroup 1973 "Measurement of employment discrimina-

tion according to sex." Southern Economic Journal 39:575-87.

Harvard Law Review 1971 "Developments in the law: employment

discrimination and Title VII of the Civil Rights Act of 1964." Harvard Law Review 84:1109-316.

Hastings, Philip and Jessie Southwick 1974 Survey Data for Trend Analysis.

Williamstown: Roper Opinion Research Center.

Haworth, Joan, James Gwartney and Charles Haworth

1975 "Earnings, productivity, and changes in employment discrimination during the 1960s." American Economic Review 65:158-68.

Heckman, James and Kenneth Wolpin 1976 "Does the contract compliance program

work?" Industrial and Labor Relations Re- view 29:544-64.

Hill, Herbert 1977 Black labor and the American legal system.

Race, Work, and the Law. Vol. 1. Washing- ton, D.C.: BNA.

Landes, William 1968 "The economics of fair employment laws."

Journal of Political Economy 76:507-52. Levine, Marvin and Anthony Montcalmo

1971 "The equal employment opportunity com- mission: progress, problems, prospects."

Labor Law Journal 22:771-9. Levitan, Sar, William Johnston and Robert Taggart

1975 Still a Dream: The Changing Status of Blacks Since 1960. Cambridge, Ma.: Har- vard University Press.

McCrone, Donald and Richard Hardy 1978 "Civil rights policies and the achievement

of racial economic equality, 1948-1975." American Journal of Political Science 22:1-17.

Madden, Janice 1978 "Economic rationale for sex differences in

education." Southern Economic Journal 44:778-97.

Marshall, Ray 1974 "The economics of racial discrimination: a

survey." Journal of Economic Literature 12:849-71.

Masters, Stanley 1975 Black-White Income Differentials. New

York: Academic Press. Niemi, Albert, Jr.

1974 "The impact of recent civil rights laws." American Journal of Economics and Sociology 33:137-44.

Pfister, Richard 1974 "Comments on Vroman's 'changes in

black workers' relative earnings.' " Pp. 189-90 in George von Furstenberg, Ann Horowitz, and Bennett Harrison (eds.), Pat- terns of Racial Discrimination, Vol. 2. Lexington: Lexington Books.

Rao, Potluri and Roger Miller 1971 Applied Econometrics. Belmont:

Wadsworth. Schwartz, Mildred

1967 Trends in White Attitudes Toward Neg- roes. Chicago: National Opinion Research Center.

Smith, James and Finis Welch 1977 "Black-white male wage ratios: 1960-70."

American Economic Review 67:323-38. Snyder, David and Paula Hudis

1976 "Occupational income and the effects of minority competition and segregation." American Sociological Review 41:209-34.

Stern, Robert, Walter Gove and Omer Galle 1976 "Equality for blacks and women: an essay

on relative progress." Social Science Quar- terly 56:664-72.

Stevenson, Mary 1975 "Relative wages and sex segregation by oc-

cupation." Pp. 175-200 in Cynthia Lloyd (ed.), Sex, Discrimination, and the Division of Labor. New York: Columbia University Press.

Strauss, Robert and Francis Horvath 1976 "Wage rate differences by race and sex in

the U.S. labour market: 1960-1970." Eco- nomica 43:287-98.

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms

LEGISLATION AND THE INCOME OF WOMEN AND NONWHITES 391

Task Force on Women and Employment of the Twentieth Century Fund

1975 Exploitation from Nine to Five. Lexington: Lexington Books.

U.S. Bureau of the Census 1958 Current Population Reports. Ser. P-20.

Washington, D.C: U.S. Government Print- ing Office.

1960- Current population reports. Ser. P-60. 1977 Washington, D.C.: U.S. Government Print-

ing Office. 1977 Social Indicators 1976. Washington, D.C.:

U.S. Government Printing Office. U.S. Commission on Civil Rights

1977 The Federal Civil Rights Enforcement Ef- fort, 1977. Washifigton, D.C.: U.S. Gov- ernment Printing Office.

U.S. Department of Labor 1974 Manpower Report of the President. Wash-

ington, D.C.: U.S. Government Printing Office.

1977 Handbook of Labor Statistics. Washington, D.C.: U.S. Government Printing Office.

1978 Employment and Training Report of the President. Washington, D.C.: U.S. Gov- ernment Printing Office.

U.S. Office of Management and Budget 1978 Special Analyses, Budget of the United

States Government, Fiscal Year 1979. Washington, D.C.: U.S. Government Print- ing Office.

U.S. Reports 1971 Washington, D.C.: U.S. Government Print-

ing Office.

U.S. Senate. Committee on Appropriations 1970 Hearings on State, Justice . . . Appropria-

tions. 91st Congress, 2nd session. Washing- ton, D.C.: U.S. Government Printing Office.

U.S. Statutes At-Large 1963- Washington, D.C.: U.S. Government Printing 1972 Office.

Venick, Shelley and Ronald Lane 1977 'Doubling the price of past discrimina-

tion." Loyola University of Chicago Law Journal 8:789-812.

Villemez, Wayne and Alan Rowe 1975 "Black economic gains in the sixties: a

methodological critique and assessment." Social Forces 54: 181-93.

Vroman, Wayne 1974 "Changes in black workers' relative earn-

ings: evidence from the 1960s." Pp. 167-96 in George von Furstenberg, Ann Horowitz, and Bennett Harrison (eds.), Patterns of Racial Discrimination, Vol. 2. Lexington: Heath.

Welch, Finis 1973 "Education and racial discrimination." Pp.

43-81 in Orley Ashenfelter and Albert Rees (eds.), Discrimination in Labor Markets. Princeton: Princeton University Press.

Wilson, William 1978 The Declining Significance of Race.

Chicago: University of Chicago Press.

ERRATA

The Items column in the February 1979 issue of the REVIEW (Vol. 44: p. 183) incorrectly identified the professional af- filiations of J. Allen Whitt (Towards a Class-Dialectical Model of Power). Whitt holds a joint appointment in the Urban Studies Program and in the Department of Sociology at Brown University.

The final sentence of Robinson and Kel- ley's article, "Class As Conceived by Marx and Dahrendorf' (ASR February, 1979), contained an error. The correct version is: "For this reason we have taken this first step toward a merger of the class and status traditions.*'

This content downloaded from 129.7.158.43 on Wed, 20 Nov 2019 17:59:34 UTC All use subject to https://about.jstor.org/terms