Common Causes of Discrimination

One direct body of evidence of the persistence of employment discrimination, despite the presence of antidiscrimination laws, comes from the scope and dispensation of job discrimination lawsuits. Discriminatory practices have occurred at highly visible U.S. corporations often having multinational operations. The suits reveal racial and gender discrimination in employment, training, promotion, tenure, layoff policies, and work environment, as well as occupational segregation. 1

Perhaps the most notorious recent case is the $176 million settlement reached between Texaco and black employees after disclosure of taped comments of white corporate officials making demeaning remarks about blacks, remarks that revealed an outlook that translated into corresponding antiblack employment practices. Clearly, neither federal antidiscrimination laws nor the pressures of competitive markets have prevented the occurrence of discriminatory practices that have resulted in significant awards or settlements for the plaintiffs. 2

Another important source of direct evidence are the audit studies of the type conducted in the early 1990s by the Urban Institute. The Urban Institute audit studies sought to examine employment outcomes for young black, Hispanic, and white males, ages 19-25, looking for entry-level jobs. Pairs of black and white males and pairs of Hispanic and non-Hispanic white males were matched as testers and sent out to apply for jobs at businesses advertising openings. 3 Prior to application for the positions, the testers were trained for interviews to minimize dissimilarity in the quality of their self-presentation, and they were given manufactured resumes designed to put their credentials on a par. The black/white tests were conducted in Chicago and in Washington, D.C., while the Hispanic/nonHispanic tests were conducted in Chicago and in San Diego.

A finding of discrimination was confirmed if one member of the pair was offered the position and the other was not. No discrimination was confirmed if both received an offer (sequentially, since both were instructed to turn the position down) or neither received an offer. This is a fairly stringent test for discrimination, since, in the case where no offer was made to either party, there is no way to determine whether employers were open to the prospect of hiring a black or an Hispanic male, what the overall applicant pool looked like, or who was actually hired. However, the Urban Institute audits found that black males were three times as likely to be turned down for a job as white males, and Hispanic males also were three times as likely as non-Hispanic white males to experience discrimination in employment.

Discussions

Bendick, Jackson and Reinoso (1994) also report on 149 race-based (black, white) and ethnicity-based (Hispanic, non-Hispanic) job audits conducted by the Fair Employment Council of Greater Washington, Inc. in the D.C. metropolitan area in 1990 and 1991. Testers were paired by gender. The audit findings are striking. White testers were close to 10 percent more likely to receive interviews than blacks. Among those interviewed, half of the white testers received job offers versus a mere 11 percent of the black testers. 5 When both testers received the same job offers, white testers were offered 15 cents per hour more than black testers. Black testers also were disproportionately “steered” toward lower level positions after the job offer was made, and white testers were disproportionately considered for unadvertised positions at higher levels than the originally advertised job.

Overall, the Fair Employment Council study found rates of discrimination in excess of 20 percent against blacks (in the black/white tests) and against Hispanics (in the Hispanic/non-Hispanic tests). In the Hispanic/non-Hispanic tests, Hispanic male job seekers were three times as likely to experience discrimination as Hispanic females. But, surprisingly, in the black/white tests, black females were three times as likely to encounter discrimination as black males. 6 The racial results for women in this particular audit stand in sharp contrast with the results in the statistical studies described above.

The most severe criticisms of the audit technique have come from Heckman and Siegelman (1993). At base, their central worry is that testers cannot be paired in such a way that they will not signal a difference that legitimately can be interpreted by the prospective employer as a difference in potential to perform the job, despite interview training and doctored resumes. 7 For example, what about intangibles like a person’s ability to make a first impression, or the fact that certain resumes may be unintentionally superior to others?

In an audit study consciously designed to address many of the Heckman and Siegelman (1993) methodological complaints, Neumark, Bank, and Van Nort (1995) examined sex discrimination in restaurant hiring practices. Four testers (all college students, two men and two women) applied for jobs waiting tables at 65 restaurants in Philadelphia. 8 The restaurants were separated into high, medium, and low price, according to average cost of a meal. Waiters at the high price restaurants tend to receive greater wages and tips than their counterparts in low price restaurants; specifically, the authors find that average hourly earnings for waiters were 47 and 68 percent higher in the high price restaurant than the medium and low price restaurant, respectively. One man and one woman applied for a job at each restaurant, so there were 130 attempts to obtain employment. Thirty-nine job offers were received.

One interesting twist to this methodology is that three reasonably comparable resumes were constructed, and over a three-week period each tester used a different resume for a period of one week. This resume-switching mitigates any differences that may have occurred because one resume was better than another. 9 To reduce other sources of unobserved ability–for example, the ability to make a good first impression–the testers were instructed to give their applications to the first employee they encountered when visiting a restaurant. That employee was then asked to forward the resume to the manager. In effect, personality and appearance were eliminated as relevant variables for the interview decision, if not for the job offer decision.

Neumark et al. (1995) find that in the low-priced restaurants, the man received an offer while the woman did not 29 percent of the time. A woman never received an offer when the man did not. In the high-priced restaurants, the man received an offer while the woman did not in 43 percent of the tests, while the woman received an offer while the man did not in just 4 percent of the tests. 10 Also, at high-priced restaurants, women had roughly a 40 percent lower probability of being interviewed and 50 percent lower probability of obtaining a job offer, and this difference is statistically significant. Hence, this audit study shows that within-occupation employment discrimination may be a contributing source to wage discrimination between men and women.

Another way to overcome some of the difficulties of the audit approach is the “correspondence test,” which has been used overseas in Britain and Australia, but not (to our knowledge) in the United States. This test involves investigators sending letters of inquiry from prospective “applicants” to employers, where the letters signal the “applicants’ “ethnicity, typically by using a name that provides a strong clue about ethnic affiliation. 11 Of course, the letters of inquiry are designed to demonstrate comparable written skills across the hypothetical members of each group and, again, manufactured resumes are submitted with the letters to present comparable credentials to employers.

Riach and Rich (1991-2) report that in the British studies, letters that appeared to be from Afro-Caribbean, Indian, or Pakistani applicants often received replies that indicated that the positions had been filled, while simultaneously, letters that appeared to be from Anglo-Saxon applicants received responses inviting them to interviews from the same employers. A similar pattern occurred in the Australian audits; inquiries from applicants with Vietnamese-or Greek-sounding names met with information that the position had been filled while Anglo-Saxon-sounding “applicants” again were asked to come for interviews. 12 This is impressive direct evidence of discrimination from a powerful test procedure. However, the correspondence test is limited to identifying discrimination at the initial stage of the hiring process. It cannot identify discriminatory practices during the interview stage, at the point of job offer, or the terms of the job offer like the job audit using trained testers.

Yet another interesting direct test of discriminatory practices based upon gender can be found in Goldin and Rouse’s (1997) assessment of the effects of an alteration in audition procedures for symphony orchestras. In the past, juries watched candidates audition. However, many orchestras now have candidates audition behind a screen, so that their identity is unknown. Goldin and Rouse find that hiding the identity of the players behind a screen raises the probability that a woman will be hired by 50 percent. The implication is obvious: prior to the adoption of the screen on identity there was sex discrimination in the selection of musicians for symphony orchestras.

The direct evidence from the court cases, audit studies, and even symphony auditions confirms the persistence of discriminatory practices in employment. The evidence is consistent with the characterization of employer beliefs and actions found in the joint Russell Sage-Ford Foundation Multi-City Study of Urban Inequality (MCSUI), newly reported by Holzer (1997). Employers seem to possess strong racial and gender preferences in hiring. 13 These preferences are the consequence of enduring stereotypical beliefs, which leads them to set up a racial/ethnic gender ranking of potential hires: white men generally preferred over white women (unless the job is female-typed), Hispanics of either gender preferred over blacks, black women preferred over black men.17 The MCSUI findings suggest the primacy of race/color as a marker for disadvantageous treatment by employers.

When we consider economic disparities by race, a difference emerges by gender. Using a Blinder-Oaxaca approach in which women are compared by their various racial and ethnic subgroups, Darity, Guilkey and Winfrey (1996) find little systematic evidence of wage discrimination based on U.S. Census data for 1980 and 1990.7 However, when males are examined using the same Census data a standard result emerges. A significant portion of the wage gap between black and white males in the United States cannot be explained by the variables included to control for productivity differences across members of the two racial groups.

Black women are likely to have the same school quality and omitted family background characteristics as black men (the same is true for white women and men). Hence, it strains credibility to argue that the black-white earnings gap for men is due to an omitted labor quality variable unless one also argues that black women are paid more than white women conditional on the unobservables. The findings of Darity, Guilkey and Winfrey (1996), Rodgers and Spriggs (1996) and Gottschalk (1997) indicate that in 1980 and 1990 black men in the United States were suffering a 12 to 15 percent loss in earnings due to labor market discrimination. 14

There is a growing body of evidence that uses color or “skin shade” as a natural experiment to detect discrimination. The approach of these studies has been to look at different skin shades within a particular ethnic group at a particular place and time, which should help to control for factors of culture and ethnicity other than pure skin color. Johnson, Bienenstock, and Stoloff (1995) looked at dark-skinned and light-skinned black males from the same neighborhoods in Los Angeles, and found that the combination of a black racial identity and a dark skin tone reduces an individual’s odds of working by 52 percent, after controlling for education, age, and criminal record! Since both dark-skinned and light-skinned black males in the sample were from the same neighborhoods, the study de facto controlled for school quality. Further evidence that lighter-complexioned blacks tend to have superior incomes and life chances than darker-skinned blacks in the United States comes from studies by Ransford (1970), Keith and Herring (1991) and Johnson and Farrell (1995).

Similar results are found by looking at skin color among Hispanics. Research conducted by Arce, Murguia, and Frisbie (1987) utilizing the University of Michigan’s 1979 National Chicano Survey involved partitioning the sample along two phenotypical dimensions: skin color, ranging from Very Light to Very Dark on a five-point scale; and physical features, ranging from Very European to Very Indian on a five-point scale. Chicanos with lighter skin color and more European features had higher socioeconomic status. Using the same data set, Telles and Murguia (1990) found that 79 percent or $1,262 of the earnings differences between the dark phenotypic group and other Mexican Americans was not explained by the traditional variables affecting income included in their earnings regression.

Further support for this finding comes from Cotton (1993) and Darity, Guilkey, and Winfrey (1996) who find using 1980 and 1990 Census data that black Hispanics suffer close to ten times the proportionate income loss due to differential treatment of given characteristics than white Hispanics. Evidently, skin shade plays a critical role in structuring social class position and life chances in American society, even between comparable individuals within minority groups.

Cross-national evidence from Brazil also is relevant here. Despite conventional beliefs in Brazil that race is irrelevant and class is the primary index for social stratification, Silva (1985) found using the 1976 national household survey that blacks and mulattos (or “browns”) shared closely in a relatively depressed economic condition relative to whites, with mulattos earning slightly more than blacks. Silva estimated that the cost of being nonwhite in Brazil in 1976 was about 566 cruzeiros per month (or $104 U.S.). But Silva found slightly greater unexplained income differences for mulattos, rather than blacks vis-à-vis whites, unexplained differences he viewed as evidence of discrimination. A new study by Telles and Lim (1997), based upon a random national survey of 5000 persons conducted by the Data Folha Institute des Pesquisas, compares economic outcomes based upon whether race is self-identified or interviewer-identified. Telles and Lim view interviewer-identification as more useful for establishing social classification and treatment. They find that self-identification underestimates white income and overestimates brown and black incomes relative to interviewer-classification.

Despite the powerful results on skin shade, some continue to argue that the extent of discrimination is overestimated by regression techniques because of missing variables. After all, it seems likely that the general pattern of unobserved variables–for example, educational quality or labor force attachment-would tend to follow the observed variables in indicating reasons for the lower productivity of black males (Ruhm, 1989, p. 157). As a result, adjusting for these factors would reduce the remaining black-white earnings differential?

As one might imagine, given the framework in which economists tackle the issue of discrimination, considerable effort has been made to find measures of all imaginable dimensions of human capital that could be used to test the presence of labor market discrimination. This effort has uncovered one variable in one data set which, if inserted in an earnings regression, produces the outcome that nearly all of the black male-white male wage gap is explained by human capital and none by labor market discrimination. (However, thus far no one has suggested a reasonable missing variable for the skin shade effect.) The particular variable that eliminates evidence of discrimination in earnings against black men as a group is the Armed Forces Qualifying Test (AFQT) score in the National Longitudinal Survey of Youth (NLSY).

A number of researchers have confirmed with somewhat different sample sizes and methodologies that including AFQT scores in an earnings equation virtually will eliminate racial differences in wages. In this journal, June O’Neill (1990) examined the 1987 sample of men aged 22-29 who had taken the AFQT when they were interviewed seven years earlier. The average AFQT score for black men was 48 and for white men it was 73.9 The unadjusted hourly wage ratio for these men was 83 percent. The ratio adjusted for region, schooling, and potential experience was 88 percent. The ratio adjusted for region, schooling, potential experience, and AFQT score was 95-96 percent, close to parity. Similarly, Maxwell (1994) looked at a cohort of men six years after leaving school, and found that the inclusion of AFQT scores in a wage regression explained two-thirds of the gap.

Ferguson (1995) used the 1988-1992 samples of males aged 25-35 years, and found that while unadjusted gaps in earnings ranged between 13 to 20 percent, AFQT scores could explain one-half to two-thirds of that difference. Neal and Johnson (1996) found that AFQT scores could explain three-quarters of the black-white gap for men and all of the black-white gap for women. Neal and Johnson also found that AFQT’s inclusion in log wage equations can completely explain wage differentials for Hispanic males and females.

The conclusion of this body of work is that labor market discrimination against blacks is negligible or nonexistent. Using Neal and Johnson’s (1996) language, the key to explaining differences in black and white labor market outcomes must instead rest with “premarket factors.” These studies have led Abigail and Stephan Thernstrom (1997) in a prominent Wall Street Journal editorial to proclaim that “what may look like persistent employment discrimination is better described as employers rewarding workers with relatively strong cognitive skills.”

But matters are not so straightforward. The essential problem is what the AFQT scores are actually measuring, and therefore what precisely is being controlled for. There is no consensus on this point. AFQT scores have been interpreted variously as providing information about school quality or academic achievement, about previously unmeasured skills, and even about intelligence (Herrnstein and Murray, 1994)although the military did not design AFQT as an intelligence test. The results obtained by O’Neill (1990), Maxwell (1994), Ferguson (1995), and Neal and Johnson (1996) after using the AFQT as an explanatory variable are, upon closer examination, not robust to alternative specifications and are quite difficult to interpret.

The lack of robustness can be illustrated by looking at how AFQT scores interact with other variables in the earnings equation. Neal and Johnson (1996), for example, adjust for age and AFQT score in an earnings equation, but not for years of schooling, presumably on the assumption that same-age individuals would have the same years of schooling, regardless of race. However, this assumption does not appear to be true. Rodgers, Spriggs and Waaler (1997) find that white youths had accumulated more schooling at a given age than black or Hispanic youths.

When AFQT scores are both age and education-adjusted, a black-white wage gap reemerges, as the authors report (p. 3): . . . estimates from models that use our proposed age and education adjusted AFQT score [show] that sharp differences in racial and ethnic wage gaps exist. Instead of explaining three-quarters of the male black-white wage gap, the age and education adjusted score explains 40 percent of the gap. Instead of explaining the entire male Hispanic-white gap, the new score explains 50 percent of the gap… [B]lack women no longer earn more than white women do, and . . . Hispanic women’s wage premium relative to white women is reduced by one-half.

Another specification problem arises when wage equations are estimated using both AFQT scores and the part of the NLSY sample that includes measures of psychological well-being (for “self-esteem” and “locus of control”) as explanatory variables. The presence of the psychological variables restores a negative effect on wages of being African-American.

Yet another specification problem becomes relevant if one interprets AFQT scores as providing information about school quality. But since there is a school survey module of the NLSY which can be used to provide direct evidence on school quality, using variables like the books/pupil ratio, the percent of students classified as disadvantaged, and teacher salaries, it would surely be more helpful to use this direct data on school quality rather than the AFQT scores. In another method of controlling for school quality, Harrison (1972) compared employment and earnings outcomes for blacks and whites living in the same black ghetto communities, on grounds that school quality would not be very different between them. Harrison found sharp differences in earnings favoring whites?

One severe difficulty in interpreting what differences in the AFQT actually mean is demonstrated by Rodgers and Spriggs (1996) who show that AFQT scores appear to be biased in a specific sense. They show that if AFQT scores are treated as an endogenous variable and if equations for AFQT are estimated separately for blacks and whites, controlling for family background, school quality and psychological motivation, the coefficients for generating AFQT scores differ substantially between blacks and whites. White coefficients generate significantly higher scores for given characteristics than black coefficients.

Following the Blinder-Oaxaca approach, Rodgers and Spriggs then create a hypothetical set of “unbiased” black scores by running the mean black characteristics through the equation with the white coefficients. When those scores replace the actual AFQT scores in a wage equation, then the adjusted AFQT scores no longer explain black-white wage differences. A similar result can be obtained if actual white scores are replaced by hypothetical scores produced by running white characteristics through the equation with black coefficients.15 Apparently, the AFQT scores themselves are a consequence of bias in the underlying processes that generate AFQT scores for blacks and whites. Perhaps AFQT scores are a proxy for skills that do not capture all skills, and thus leave behind a bias of uncertain direction. Or there may be other predictors of the test that are correlated with race but which are left out of the AFQT explanatory equation.

To muddy the waters further, focusing on the math and verbal subcomponents of AFQT leads to inconsistent implications for discriminatory differentials. For example, while a higher performance on the verbal portion of the AFQT contributes to higher wages for black women versus black men, it apparently has little or no effect on the wages of white women versus white men. However, white women gain in wages from higher scores on the math portion of the AFQT, but black women do not. Perhaps this says that white women are screened (directly or indirectly) for employment and pay on the basis of their math performance, while black women are screened based upon their verbal skills. Perhaps this is because white employers have a greater “comfort zone” with black women who have a greater verbal similarity to whites. Or perhaps something not fully understood and potentially quirky is going on with the link between these test results and wages.

Finally, since skill differentials have received such widespread discussion in recent years as an underlying cause of growing wage inequality in the U.S. economy-see, for example, the discussion in the Spring 1997 issue of this journal–it should be pointed out that growth in the rewards to skill does not mean that the effects of race have diminished. If the importance of race and skill increase simultaneously, then a rising skill premium will explain more of the changes in intraracial wage inequality, which may well leave a larger unexplained portion of interracial wage inequality.

For example, when Murnane et al. (1995) ask whether test scores in math, reading, and vocabulary skills for respondents in the National Longitudinal Study of the High School Class of 1972 and High School and Beyond datasets have more explanatory power in wage equations for 1980 graduates than 1972 graduates, their answer is “yes”–the rate of return to cognitive skill (test scores) increased between 1978 and 1986. However, in these same regressions, the absolute value of the negative race coefficient is larger for the 1980 graduates than it is for the 1972 graduates! These results confirm that there are increasing returns to skills measured by standardized tests, but do not indicate that the rise in returns to skills can explain changes in the black-white earnings gap very well.

The upshot is the following. There is no doubt that blacks suffer reduced earnings in part due to inferior productivity-linked characteristics, like skill gaps or school quality gaps, relative to non-black groups. However, evidence based on the AFQT should be treated with extreme caution. Given that this one variable in one particular data set is the only one that suggests racial discrimination is no longer operative in U.S. employment practices, it should be taken as far from convincing evidence. Blacks, especially black men, continue to suffer significantly reduced earnings due to discrimination and the extent of discrimination.

Conclusion

Critical mass studies of ethnic minority and female faculty in the departments of universities, however, suggested that if female and ethnic minority faculty are mentored and provided with social support during their first 3 years in an academic setting, the number of voluntary academic dropouts can be decreased. Thus, first 3 years in any university, therefore, are a critical period for junior faculty to seek adviser and supportive colleagues, and for institutions to ensure the junior faculty is offered help by tenured faculty and colleagues. Even though great steps have been made in increasing representation for women and ethnic minorities in academe, but up till now not much equity has not been achieved.

The studies of academic salaries and position achievement by gender demonstrated that even after accounting for academic qualifications and length of experience, gender differences continues, and suggested different treatment of academically comparable men and women (Bergmann, 1985). Furthermore, data on withholding revealed that women and ethnic minorities do not receive promotion and tenure at rates comparable to those of white men. This is not because women and ethnic minorities are less deserving in any way to men or work less hard to achieve tenure, or make lower worth or significance contributions than their white male colleagues (Wicherski, Kohout, & Fritz, 1990).

Standard neoclassical competitive models are forced by their own assumptions to the conclusion that discrimination only can be temporary. Perhaps the best-known statement of this position emerges from Becker’s (1957) famous “taste for discrimination” model. If two groups share similar productivity profiles under competitive conditions where at least some employers prefer profits to prejudice, eventually all workers must be paid the same wage. The eventual result may involve segregated workforces-say, with some businesses hiring only white men and other: hiring only black women–but as long as both groups have the same average productivity, they will receive the same pay. Thus, in this view, discrimination only can produce temporary racial or gender earnings gaps. Moreover, alternative forms of discrimination are separable processes; wage discrimination and employment segregation are unrelated in Becker’s model.

Despite the theoretical implications of standard neoclassical competitive models, we have considerable evidence that it took the Civil Rights Act of 1964 to alter the discriminatory climate in America. It did not, by any means, eliminate either form of discrimination. Indeed, the impact of the law itself may have been temporary, since there is some evidence that the trend toward racial inequality came to a halt in the mid-1970s (even though interracial differences in human capital were continuing to close) and the momentum toward gender equality may have begun to lose steam in the early 1990s. Moreover, we believe that the forms of discrimination have altered in response to the act. Therefore, it is not useful to argue that either racial or gender discrimination is inconsistent with the operation of competitive markets, especially when it has taken antidiscrimination laws to reduce the impact of discrimination in the market.

Instead, it is beneficial to uncover the market mechanisms which permit or encourage discriminatory practices. Since Becker’s work, orthodox microeconomics has been massaged in various ways to produce stories of how discrimination might sustain itself against pressures of the competitive market. The tacit assumption of these approaches has been to find a way in which discrimination can increase business profits, or to identify conditions where choosing not to discriminate might reduce profits.

In the customer discrimination story, for example, businesses discriminate not because they themselves are bigoted but because their clients are bigoted. This story works especially well where the product in question must be delivered via face-to-face contact, but it obviously does not work well when the hands that made the product are not visible to the customer possessing the “taste for discrimination.” Moreover, as Madden has pointed out, sex-typing of jobs can work in both directions: “While service occupations are more contact-oriented, sexual preference can work both ways: for example, women are preferred as Playboy bunnies, airline stewardesses, and lingerie salespeople, while men seem to be preferred as tire salespeople, stockbrokers, and truck drivers.” Obviously, group-typing of employment will lead to a different occupational distributions between group A and B, but will it lead to different earnings as well? Madden suggests not necessarily: . . . consumer discrimination causes occupational segregation rather than wage differentials.

If the female wage decreases as the amount of consumer contact required by a job increases, women seek employment in jobs where consumer contact is minimal and wages are higher. Only if there are not enough non-consumer contact jobs for working women, forcing them to seek employment in consumer-contact jobs, would consumer discrimination be responsible for wage differentials. Since most jobs do not require consumer contact, consumer discrimination would segregate women into these jobs, but not cause wage differentials. Perhaps the best attempt to explain how discrimination might persist in a neoclassical framework is the statistical discrimination story, which, at base, is a story about imperfect information. The notion is that potential employers cannot observe everything they wish to know about job candidates, and in this environment, they have an incentive to seize group membership as a signal that allows them to improve their predictions of a prospective candidate’s ability to perform.

However, this model of prejudicial beliefs does not ultimately wash well as a theory of why discrimination should be long-lasting. If average group differences are perceived but not real, then employers should /earn that their beliefs are mistaken. If average group differences are real, then in a world with antidiscrimination laws, employers are likely to find methods of predicting the future performance of potential employees with sufficient accuracy that there is no need to use the additional “signal” of race or gender. It seems implausible that with all the resources that corporations put into hiring decisions, the remaining differentials are due to an inability to come up with a suitable set of questions or qualifications for potential employees. Moreover, models of imperfect competition as explanations of discrimination do not solve the problem completely either.

The reason for the immutability of the imperfection is rarely satisfactorily explained–and often not addressed at all–in models of this type. Struggle as it may, orthodox microeconomics keeps returning to the position that sustained observed differences in economic outcomes between groups must be due to an induced or inherent deficiency in the group that experiences the inferior outcomes. In the jargon, this is referred to as a deficiency in human capital. Sometimes this deficiency is associated with poor schooling opportunities, other times with culture. But the thrust of the argument is to absolve market processes, at least in a putative long run, of a role in producing the differential outcome; the induced or inherent deficiency occurs in pre-market or extra-market processes.

The success in the academic arena is governed to a large extent by devotion to a set of open and understood rules and priorities that women and ethnic minorities have had little role in shaping and sometimes find estrange and tyrannical.

Individuals in these groups are more likely than white men to experience ambivalence about playing the game according to the demands of the white male-dominated system. Many face problems as they try to successfully negotiate an academic career in a setting in which the established goals may conflict with values they hold (Sandler & Hall, 1986).

Notes

Arce, Carlos H., Eduard Murguia, and W. Parker Frisbie, “Phenotype and Life Chances Among Chicanos,” Hispanic Journal of Behavioral Studies, 1987, 9.1, 19-33.

Bendick, Marc Jr., Charles W. Jackson, and Victor A. Reinoso, “Measuring Employment Discrimination through Controlled Experiements.” In James B. Stewart, ed. African-Americans and Post-Industrial Labor Markets. New Brunswick: Transaction Publishers, 1997, 77-100; originally published in The Review of Black Political Economy, Summer 1994.

Cotton, Jeremiah, “Color or Culture?: Wage Differences Among Non-Hispanic Black Males, Hispanic Black Males and Hispanic White Males,” The Review of Black Political Economy, Spring 1993, 21:4, 53-68.

Donohue, John, and James Heckman, “Continuous vs. Episodic Change: The Impact of Civil Rights Policy on the Economic Status of Blacks,” Journal of Economic Literature, December 1991, 29.4, 1603-43.

Elmslie, Bruce, and Stanley Sedo, “Discrimination, Social Psychology and Hysteresis in Labor Markets,” Journal of Economic Psychology, 1996, 17, 465-78.

Fix, Michael, George C. Galster, and Raymond J. Struyk, “An Overview of
Auditing for Discrimination.” In Michael Fix and Raymond Struyk, eds. Clear and Convincing Evidence: Measurement of Discrimination in America. Washington: The Urban Institute Press, 1993, 1-68.

Freeman, Richard B., “Changes in the Labor Market for Black Americans

Goldin, Claudia, and Cecilia Rouse, “Orchestrating Impartiality: The Impact of ‘Blind’ Auditions on Female Musicians,” unpublished manuscript, Harvard University, June 1997.

James, F., and S. W. DelCastillo, “Measuring Job Discrimination by Private Employers Against Young Black and Hispanic Males Seeking Entry Level Work in the Denver Metropolitan Area,” unpublished manuscript, University of Colorado at Denver, March 1991.

Johnson, James H., Jr., and Walter C. Farrell, Jr., “Race Still Matters,” The Chronicle of Higher Education, July 7, 1995, A48.

Johnson Jr., James H., Elisa Jayne Bienenstock, and Jennifer A. Stoloff, “An Empirical Test of the Cultural Capital Hypothesis,” The Review of Black Political Economy, Spring 1995, 23:4, 7-27.

Mason, Patrick L., “Race, Competition and Differential Wages,” Cambridge Journal of Economics, 1995, 19, 545-67.

Rodgers III, William, W’dliam E. Spriggs, and Elizabeth Warder, “The Role of Premarket Factors in Black-white Wage Differences: Comment,” Unpublished manuscript, College of William and Mary, May 25, 1997.

Telles, Edward E., and Nelson Lira, “Does Who Classify Race Matter? Self Vs. Social Classification of Race and Racial Income Inequality in Brazil,” Unpublished manuscript, UCLA, April 28, 1997.

Waldfogel, Jane, “Understanding the ‘Family Gap’ in Pay for Women with Children,” Journal of Economic Perspectives, Winter 1998, 13:1, 13756.

White, Katherine, “Simultaneity Issues in the Relationship of Income and Intelligence,” Undergraduate Senior Honors Thesis, Department of Economics, University of North Carolina at Chapel Hill, 1997.

�

Bibliography

Arce, Carlos H., Eduard Murguia, and W. Parker Frisbie, “Phenotype and Life Chances Among Chicanos,” Hispanic Journal of Behavioral Studies, 1987, 9.1, 19-33.

Bendick, Marc Jr., Charles W. Jackson, and Victor A. Reinoso, “Measuring Employment Discrimination through Controlled Experiements.” In James B. Stewart, ed. African-Americans and Post-Industrial Labor Markets. New Brunswick: Transaction Publishers, 1997, 77-100; originally published in The Review of Black Political Economy, Summer 1994.

Bergmann, 1985. Discrimination, affirmative action and equal opportunity. The Fraser Institute

Cotton, Jeremiah, “Color or Culture?: Wage Differences Among Non-Hispanic Black Males, Hispanic Black Males and Hispanic White Males,” The Review of Black Political Economy, Spring 1993, 21:4, 53-68.

Donohue, John, and James Heckman, “Continuous vs. Episodic Change: The Impact of Civil Rights Policy on the Economic Status of Blacks,” Journal of Economic Literature, December 1991, 29.4, 1603-43.

Elmslie, Bruce, and Stanley Sedo, “Discrimination, Social Psychology and Hysteresis in Labor Markets,” Journal of Economic Psychology, 1996, 17, 465-78.

Fix, Michael, George C. Galster, and Raymond J. Struyk, “An Overview of
Auditing for Discrimination.” In Michael Fix and Raymond Struyk, eds. Clear and Convincing Evidence: Measurement of Discrimination in America. Washington: The Urban Institute Press, 1993, 1-68.

Freeman, Richard B., “Changes in the Labor Market for Black Americans

Goldin, Claudia, and Cecilia Rouse, “Orchestrating Impartiality: The Impact of ‘Blind’ Auditions on Female Musicians,” unpublished manuscript, Harvard University, June 1997.

James, F., and S. W. DelCastillo, “Measuring Job Discrimination by Private Employers Against Young Black and Hispanic Males Seeking Entry Level Work in the Denver Metropolitan Area,” unpublished manuscript, University of Colorado at Denver, March 1991.

Johnson Jr., James H., Elisa Jayne Bienenstock, and Jennifer A. Stoloff, “An Empirical Test of the Cultural Capital Hypothesis,” The Review of Black Political Economy, Spring 1995, 23:4, 7-27.

Johnson, James H., Jr., and Walter C. Farrell, Jr., “Race Still Matters,” The Chronicle of Higher Education, July 7, 1995, A48.

Mason, Patrick L., “Race, Competition and Differential Wages,” Cambridge Journal of Economics, 1995, 19, 545-67.

Rodgers III, William, W’dliam E. Spriggs, and Elizabeth Warder, “The Role of Premarket Factors in Black-white Wage Differences: Comment,” Unpublished manuscript, College of William and Mary, May 25, 1997.

Sandler & Hall, 1986. “The employee rights handbook.” Answers legal questions from interview to pink slip. Journal of planning and employment.

Telles, Edward E., and Nelson Lira, “Does Who Classify Race Matter? Self Vs. Social Classification of Race and Racial Income Inequality in Brazil,” Unpublished manuscript, UCLA, April 28, 1997.

Waldfogel, Jane, “Understanding the ‘Family Gap’ in Pay for Women with Children,” Journal of Economic Perspectives, Winter 1998, 13:1, 13756.

White, Katherine, “Simultaneity Issues in the Relationship of Income and Intelligence,” Undergraduate Senior Honors Thesis, Department of Economics, University of North Carolina at Chapel Hill, 1997.

Wicherski, Kohout, & Fritz, 1990. the workplace Discrimination. New York, NYC.