BackFire : Commentary and Extension

Reviewed by
Edward M. Miller
Professor of Economics and Finance
University of New Orleans
New Orleans, LA 70148
To appear: Journal of Social, Political, and Economic Studies, Vol. 21 (Winter 1996) No 4.

For American readers the term affirmative action will need no definition. Non-US readers should know it is a polite term for discrimination in favor of certain groups, including minorities and women. These readers may find Backfire interesting as a case study of how the US system of preferences works, and what to watch out for in programs proposed for their own countries. In spite of the problems affirmative programs have experienced in the United States, many countries have programs designed to benefit particular minorities, such as Malays in Malaysia and scheduled castes (untouchables) in India.

However, such in the attraction of American ideas that some countries may be considering adopting American style affirmative action. Recently, the Wall Street Journal (August 6, 1996) ran a front page story on the pressures in Brazil saying "The debate on "acao afirmativa," notes Brazil's leading magazine Veja,' was inevitable in a country that always so loved American inventions like blue jeans, Coca-Cola and Disney." The story reports that so institutionalized are affirmative action and diversity programs (another American term for essentially the same thing) that US firms operating in Brazil are frequently adopting programs to favor blacks in employment even though they are not required to do so by law, and there is little local pressure to do so.

Thus it seems useful to review a book by a US critic of affirmative action, and to summarize some of the problems that have been observed in the US program. To a large extent this book can be viewed as a detailed account of unexpected consequences; of how programs intended to accomplish one purpose end up having unanticipated side effects.

Backfire starts by making the obvious point that affirmative action is discrimination in favor of one group, and that discrimination in favor of one group logically discriminates against the non-favored groups. Admittedly, affirmative action programs started as a way to oppose discrimination. The first job quotas were part of the "Philadelphia Plan", an effort to get blacks into the unionized construction union jobs there. There were very few blacks, and there was strong opposition to having more. Quotas in the guise of "goals" were set. However, once the quota precedent had been set, goals and timetables quickly expanded. A program to fight discrimination on the basis of race became one that demanded discrimination on the basis of race.

An early court decision created the adverse impact doctrine. Any test or procedure that resulted in "adverse impact" on a group (i.e. group members were selected less than in proportion to their numbers in the relevant labor market) was presumed illegal, unless its use could be shown to be required by "business necessity," a vague standard that in practice was hard to meet. The premise behind the policy was that there were no real ability differences between the groups, and that hence any differences in hiring should be presumed to be due to discrimination. As will be discussed later, there is strong evidence that this belief is false.

However, once the belief became established wisdom, one of the largest of the hidden costs of affirmative action programs appeared. This was a general attack on the idea of merit (p. 9).

Chapter 4 deals with testing. It is describes how the "adverse impact" standard first appeared in the case of Griggs versus Duke Power. Duke Power had historically refused to hire blacks for any but menial jobs. It had stopped this discrimination when it became illegal, but had adapted at the same time employment tests that blacks passed at much lower rates than whites did. With such a case, it was easy to imagine that the newly established testing was a subterfuge for continued discrimination. This made plausible a decision that any test that had "adverse impact" was illegal unless proved to be required by "business necessity". As earlier noted, the premise was that the two racial populations differed little in actual ability to do the job. Once this principle had been accepted the clear language in the US law that permitted use of professionally developed tests could be worked around. While the "adverse impact" standard was created by the court system, it was written into US law in the 1991 Civil Rights Act.

In practice, it proved very hard to meet the "business necessity" standard and to "validate" a test, providing adequate proof that it really did select better workers. This was because the civil rights bureaucracy and the courts were dedicated to increasing the number of minorities in good jobs, and this goal was hindered by the use of tests. The book provides a remarkable quotation (p. 97) by Norton, the head of President Carter's Equal Employment Opportunity Commission in which she states "There is not any way in which black people tomorrow as a group are going to, no matter what kind of test you give them, score the same way white people score. And I can't live with that." Thus even the regulators came to recognize the problem. As the book points out, someone whose primary concern was productivity would probably be working in a agency other than a civil rights one.

Any test or procedure at which blacks did worse than whites was presumed to be biased and could not be used. The cost was high for only moderate benefits to the targeted groups. In most cases competition for jobs, school admissions etc. was between people of the same race. The effect of not using the most effective selection rules was to place the less competent into jobs. Most of the less competent who benefited from failure to test were not minorities. When a minority got the job, failure to test frequently mean that it was not be the most competent of the minority applicants. Another unanticipated effect of a well intentioned government program!

The book goes on to summarize what is now known about ability testing, how it does help to select employees, how it is now known to work as well for black candidates as for white ones (earlier some thought tests that had been shown to work for predominantly white applicants might not work for minority employees) The book summarizes Schmidt and Hunter's work (see Schmidt, 1988) and the review conducted by the National Research Council (Wigdor & Garner, 1982). More references to the literature can be found elsewhere (Gottfredson, 1986; Gottfredson & Sharf, 1988; Miller 1994b, 1995b)

The problem is indicated by the hiring experience of the US government (p. 93). When it used a test that was essentially a test of general intelligence, it hired 9.5% black or Hispanic employees. It was challenged in the Courts as having disparate impact, and under consent decrees both the Carter and Reagan administrations stopped using it and shifted to a "more narrowly tailored test (far less predictive)" and the Black-Hispanic representation was raised to 27.4%.

The book describes how many firms wishing to retain the benefits of tests, but yet wishing to avoid discrimination charges, adopted race norming. This procedure adjusts scores to reflect relative rankings within each ethnic or racial group. Thus blacks at the 50th percentile for blacks would be compared with whites at the 50th percentile for whites. This procedure became widely used by state employment offices to refer workers to employers. However, the obvious racial discrimination offended people's ideas of fairness, and race norming was outlawed in the 1991 Civil Rights Acts.

Paradoxically, if one is going to have to meet a quota for minorities, race norming minimizes the damage done to workforce quality. Hence by forbidding the procedure, while still retaining the pressures to have ethnic proportional representation, large economic costs have been imposed. Currently, it is hard to use tests without risking adverse impact litigation.

Politically (in my opinion) the problem in getting ability tests legalized is the need to admit honestly that blacks (and to a less degree certain other groups) really are lower in ability. If one says they are lower in ability, one is accused of being racist. The best political strategy for getting the economic benefits of ability testing is not to talk about testing itself, but about education and providing an incentive for students to study. In the near future, if tests are legalized or standards for their validation are liberalized, it will be done as part of an effort to improve the educational performance of American children. Giving preference to those who do well on certain tests of literacy and mathematical ability could be justified as providing an incentive to students to learn the material. Employers requiring good performance on such tests would be depicted as being more than just their efforts to hire a well qualified workforce.

Tests of literacy and numeracy are correlated with basic intelligence. They work in hiring good employees because not only are these skills job related in most cases, but because they also test for high general intelligence. Thus they may be a politically acceptable alternative to test labeled intelligence tests.

Yet the U. S. civil rights establishment is aware of the adverse impact of testing for scholastic ability and can be expected to oppose such proposals. The testing related chapter opens with the history of the Clinton program to set up a certification programs for job related skills. This was accompanied by much rhetoric about seeking excellence. However, in the end the law as passed included language making clear that use of such government sponsored certification in hiring would not protect against violations of the "adverse impact" tests of the Civil Rights laws. Since such adverse impacts were very likely, a fatal obstacle was created to widespread use of the program.

If employees can not be selected by tests, employers give weight to educational credentials, including college degrees. This brings us to the chapter on race and university admissions. It opens with a fascinating discussion of how the Cornell School of Industrial Relations' affirmative action program, showing how minimum standards vary by race. It describes how the number of blacks admitted was halved by a faculty imposed requirement to accept no one with a Scholastic Aptitude Test (SAT) score below 500 on mathematics. A new required mathematics course (pre-calculus) resulted in half of the affirmative action admittances getting a D, and a third actually failing. This experience shows both that admission tests serve a purpose, and that there are real costs to the students admitted on an affirmative action basis.

There is some fascinating data on just how small the pool of highly qualified black students is. On the verbal portions of the Scholastic Aptitude Test, 0.6% of the black test takers and 4.2% of the white ones earned a score of 650 (which would be required of white students at the elite US schools), This is a seven fold difference. For a 650 math score, 1% of the black test takers and 13% of the white ones qualified, producing a 13 fold difference. Although this extreme discrepancy is very striking, test scores are known to be roughly normally distributed. A property of normal curves is that a moderate difference in the means (there is usually about a one standard deviation between white and blacks on IQ type tests), produces such an extreme difference in the ratios at the extremes.

The absolute numbers are interesting too. There were 649 blacks at that level on the verbal portion, and 2,050 blacks on the math portion. As the Journal of Blacks in Higher Education pointed out, "There are approximately 45,000 places for entering freshman in the nations highest ranked universities. At the present time about 3,000 of these places are held by African Americans. If the huge racial gap in SAT scores continues to prevail and these institutions make their admission decisions without regard to race, blacks will be able to win only 2% of these positions. As indicated by their share of the top scoring SAT, the number of incoming black freshman at the nation's top-ranked institutions would drop to 900." Actually, this quote probably exaggerates the effect of a return to non-discriminatory admissions because most schools give some weight to high school grades and class rankings where blacks do not do as poorly (partially just because the standards are usually laxer in the predominantly black high schools, where many blacks study). However, the quote does make it clear that a large fraction of the blacks in the elite universities are there only because of affirmative action.

Some very interesting tables show for various schools just how big the gap is between black and non-black admissions is.

While schools are frequently reluctant to reveal how much they have cut standards (partially to avoid stigmatizing the students admitted under affirmative action programs), some information is available from court cases. The Bakke case in which the US Supreme Court declared that some weight could be given to race in university admissions is discussed. This eliminated the rigid quota system which had reserved for blacks sixteen places out of a hundred places in the University of California Medical School at Davis. It is show how much better the test scores of the rejected white candidate, Bakkee, were than those of the blacks that were accepted.

The Hapgood case at the University of Texas Law School is discussed (p. 145). Here it emerged that the score that led to a presumptive rejection for whites was higher than the score that led to a presumptive acceptance for blacks and Hispanics. Apparently, out of the forty-one blacks and fifty-five Hispanics accepted, at most nine blacks and eighteen Hispanics would have been accepted had common standards been applied. This program was thrown out (after it had been operating for a number of years) by the Fourth Circuit Court. One interesting feature that emerged from this case, and from one eliminating a racially exclusive scholarship at the University of Maryland, is that a high fraction of the benefits had gone to out of state of students. Administrators tried to recruit a large number of minority students without cutting standards too far. Of course, it is hard to justify preferential admission for out of state students on the basis of compensation for the state's history of segregation.

While one might argue that the practice of law is zero sum (lowering the average ability of lawyers may not hurt society much), medicine is a different case. Evidence is quoted (from the Spring 1995 Journal of Blacks in Higher Education) showing that the average grade point average of blacks accepted into medical schools was actually lower than that for whites rejected, and the average Medical College Admittance Test score for blacks accepted was below that for whites rejected.

This difference in standards shows up in differential pass rates on the National Board Exams. In 1988 pass rates were 88 for whites, 84 for Asians, 66 for Hispanics, and 49 for blacks. It follows that a substantial fraction of the preferential admissions do not result doctors allowed to practice, an outcome that adversely affects both society, and the minorities who go through medical school and then fail to become licensed. A logical implication is that in choosing a doctor one should avoid minorities that may have benefited from affirmative action programs.

A frequent proposal is to replace affirmative action programs based on race with affirmative action programs based on class. However, some data (taken from Hacker, 1992) shows that while within both Whites and Blacks SAT scores go up with income, the average score for blacks whose family income exceeds $70,000 (854) is actually below that for whites with a family income $10,000 -29,000 (879). At this lower income level, the black average score is only 704. Thus, even in competition among the low income, whites will do well. Income is really a poor surrogate for race. Also, there are large numbers of low income Asians who do very well on intelligence tests and earn high grades in school. Any bias in favor of the poor will benefit many of these Asians.

Although the proponents of affirmative action like to depict the typical beneficiary as a low income black from the inner city, this stereotype is inaccurate. Latter on (p. 191) we learn that the average family income of black freshman at Berkeley is $39,000, somewhat above the national median. In 1987, two thirds of the blacks there had family incomes at or above the national median. Incidentally, this observation questions the "diversity" rationale frequently given for affirmative action, since such individuals are likely to be solidly middle class, and in most ways not to differ radically from middle class citizens of other races. The poor performance of blacks from high income families is evidence that the poor overall black performance is due to more than poverty.

There is a whole chapter "A State of Excess" devoted to the controversy in California about affirmative action, which has put into the public record much interesting information. Thus, there is a full account of the elaborate systems of preference that have been adapted at Berkeley in order to achieve "diversity". For instance, in 1988, the mean SAT scores were highest for the groups not benefiting from preferences, 1267 for whites and 1269 for Asians (here excluding Filipinos, who averaged 1126, and did benefit from preferences) versus 979 for Blacks, 1013 for Chicanos, and 1053 for Latinos (Chicanos are those of Mexican origin, and others of Latin American or Spanish origin are apparently Latinos). There was a similar pattern for school grades, although an interesting feature was that the Asians had a grade point average of 3.91, somewhat above the white average of 3.79, reflecting an Asian tendency to do well in school, probably due to harder work. Interesting, it is reported (p. 188) that Asians were in the top 12.5% (about the statutory standard for admittance to the University of California system) of their high school classes an amazing 32.3% of the time versus 3.9% of the time for Hispanics and 5.1% for Blacks (whites were at this level 12.5% of the time). Again this strong performance by Asians, many of which are low income, suggests that black problems are due to more than poverty.

The book goes on to document that blacks survive to graduation at lower rates than whites in most universities, so there is some cost (failure to graduate) to some of those admitted under preferential programs. As is pointed out, many of the blacks preferentially admitted to very competitive schools are still above average, and would have done well in less competitive schools. There are probably a few who flunk out or drop out (students just barely passing frequently drop out) who would have made it at a less competitive school. Perhaps a higher cost is that students having academic difficulty go into non-preferred majors or programs that prepare them for no particular careers (notably Black Studies).

A chapter is devoted to the problems of the South's historically black universities. Here again one sees the unanticipated consequences of well intentioned programs. Although blacks had once fought hard for integration and access to the white universities, in recent years they have been frightened of desegregation measures which might eliminate the black universities they had traditionally controlled. Mississippi had by law established higher required admission standards for the white universities than for the black universities, reflecting the substantial differences in average scores between the two groups (and possibly a desire to minimize the number of blacks in the universities attended by whites). When litigation resulting in eliminating the difference (raising the minimum for black universities to the score required for white ones), the historically black Jackson State University found that its admissions were down by 60%, as much of its traditional clientele could not longer qualify..

Another example of unintended consequences is provided by the discussion of the effort to maximize the number of majority black Congressional districts, and hence the number of black representative in the House of Representatives. Although in the US blacks usually vote Democratic, this effort was supported by the Republican party. It realized that taking blacks out of many racially mixed districts and putting them into districts of their own would frequently permit the Republicans to win control more often of the remaining predominantly white districts. This was done at the cost of very oddly shaped districts as the black areas of various cities were connected by narrow belts of land (sometimes only the width of a highway) running through intervening rural areas. It is argued that this shift ended up giving control of Congress to the Republicans. The result was that the Black Caucus (the group of black Representatives) became larger, but lost most of the power they once had when the Democrats had controlled the US House of Representatives, and they were a powerful block within the party.

There is a chapter devoted to preferential purchasing from minority firms. The author is fairly sympathetic, pointing out that black owned firms are much more likely to employ blacks than white owned firms (and not asking whether any of this reflects illegal discrimination). However, again there are unexpected consequences. Because so much government purchasing is of items (planes, tanks etc.) that are only produced by large publicly owned firms (which seldom qualify as minority or women owned firms) what often appear to be moderate quotas expressed as a percentage of overall purchases, end up being concentrated in such fields as construction or maintenance, where they may amount to rather strong discrimination against white owned firms. In 1995 100% of the construction contracts at Barksdale Air Force Base were restricted to firms qualifying for the preferential program for women and minorities.

It is pointed out how the US custom of defining minorities to include Hispanics, Asians, and other groups produces odd results. It describes how in the Washington DC area $60 million dollars in construction contracts were awarded to minority owned firms, and how $36 million of this went to three firms owned by two Rodriguez brothers, who are Portuguese whites, but classified as minorities. It is non-obvious why Portuguese should get preference over, say, Italians.

The point is made that when affirmative action was first invented blacks were 66% of those traditionally benefiting (blacks, Asians, native Americans, and Hispanics). Today, with heavy Hispanic and Asian immigration, blacks make up only 49% of the potential beneficiaries. There are few countries besides the United States that give preference to immigrants over their own natives. Certainly the traditional goals of compensating for the historical wrongs of slavery cannot apply to most of these groups.

Another chapter deals with the controversy over racial differences in home lending where well publicized media studies showed differences in lending rates based on race, but studies controlling for the applicant's characteristics showed much smaller differences, or no differences. However, the government was able to use threats of enforcement actions and its power over mergers to increase lending in minority areas.

A study (p. 329) is described that controlled for borrower characteristics such as credit history, loan: equity ratio, income, and many others. Apparently Hispanic, Asian, and Native American borrowers had default rates similar to those of white borrowers, while default rates and losses were significantly higher for black borrowers. This suggests fundamental differences in personality between blacks and non-blacks, such as I have argued elsewhere could have been produced by evolution (Miller 1994a, 1995a). It is easy to imagine that the type of personality that planned ahead and deferred consumption would have had an advantage in surviving the cold winters where the ancestors of white, Asians, and American Natives evolved. Banfield (1974) has argued that many of the problems of the American Inner city are due to a tendency (which he asserted was non-genetic) to prefer immediate gratification over future gratification, a trait that would probably make for poor credit worthiness.

As in other chapters there are examples of unintended consequences of government programs. Government pressures apparently led to increased mortgage lending in minority communities through more aggressive marketing (deduced since lending to minorities rose without rejection rates changing). It is pointed out how this increased marketing effort could squeeze traditional lenders (some of which were minority owned) in these communities. A minority owned bank in Boston (Boston Bank of Commerce) was knocked out of the mortgage market by such competition,

Another possible cost noted is that tampering with underwriting standards (such as lower down payments) could increase defaults, which hurts both the borrowers and lenders. High default rates have been noticed among inner-city (which in US usage often refers to minority neighborhoods) beneficiaries of such programs. A response has been to use credit histories (from things like credit cards) in evaluating mortgage applications, something that traditionally had not been widely used. Zelnick reports that in private conversation officials had expressed concern that such credit scoring techniques might become a hindrance to acceptance of minority mortgage applications. The author argues that if this happened, the efforts to use civil rights as a whip to beat lending institution into changing practices will have misfired.

There is another way not mentioned in which efforts to increase mortgage lending to minorities might benefit minorities less than anticipated. Increased credit availability can be expected to raise real estate prices as more buyers come into the market. Many of the initial beneficiaries of this price rise will be owners of rental properties and non-minorities owners. Once prices have reached their new higher level, new buyers may not benefit (and may suffer if lax underwriting standards lead to them not being able to continue payments, and defaulting). Also many minorities will continue to remain renters in the traditional neighborhoods, and the rents will come to reflect the higher real estate prices.

The chapter on insurance points out how efforts are now being made to extend the idea of "disparate impact" from employment law to insurance so that any procedure or rule that has an adverse impact on minorities would be presumed illegal unless justified by "business necessity". For fire insurance, this means practices that make it harder to insure older houses or houses of low value, since such houses are disproportionately lived in by blacks and Hispanics of the inner cities. Even though there are no charges of individuals having been denied insurance because of race, the insurance companies are under pressure to change underwriting standards so as to benefit blacks and other minorities. Although the author does not put it this way, this could be described as a way to subsidize insurance for low income inner city residents at the expense of higher income individuals in the suburbs. Doing so frequently involves abandoning underwriting standards that have proved to predict risk (age of house, price of house, insured value to replacement cost, etc.) and profitability. One of the risks is that if suburban premiums can to involve a high element of subsidy for inner city risks, such homeowners may decide to self insure. This would hurt the insurance companies (which is probably why they oppose these ideas), and the suburban homeowners who would bear higher levels of risk. Although the book does not mention it, there is at least a possibility that some of the underwriting rules work as well as they do is because they are to some extent surrogates for race, and behaviors that lead to fire.

One of the weak points of the books is that it neglects the discussion of why blacks do worse in school and on standardized tests, although it does mention that they do in passing (while avoiding the use of the word intelligence). Although the book presents the results of various mental tests (such as the SAT, it avoids calling these IQ tests, and avoids mentioning the racial differences on IQ tests, or stating that blacks have a lower IQ. Fortunately, the evidence on racial differences in IQ is available elsewhere (Herrnstein & Murray, 1994; Jensen, 1980, 1981, 1985; Seligman, 1992). At various points the book refers to the problems of the black community, families without fathers, poverty, high crime etc. and states these are the causes of the problem. Yet this is never demonstrated, and at various points in the book evidence is given that suggests the problem may be more than this (such as the good performance of various Asian groups, and the fact that low income whites seems able to outperform high income blacks on the SAT).

Unfortunately, the evidence that a large part of the racial difference is genetic is never presented, even though a survey of experts conducted in the late eighties showed that three times as many experts believed it to be both environmental and genetic and origin as believed it to be only environmental in origin (Synderman & Rothman, 1988).

Recently, evidence from adoption of black children into white families has shown that racial difference are about as large when both are raised in white families as it is when each are raised in families of the same race. This is strong evidence for a genetic cause for the racial differences (Levin, 1994; Lynn, 1994).

Conclusion

In conclusion, Backfire provides a powerful condemnation of the current US policies of race and sex preferences. The programs appear to have done relatively little for the minorities they were intended to aid. The members of the minority groups who benefited are typically the ones who would have been middle class or better in any case, rather than the truly poor. Efforts to prevent adverse impact on minorities have imposed large costs through sacrifices the hiring of the most productive worker that use of tests would have made possible, even in cases where the competition for the job would been between two white people. A program originally meant to be temporary and limited has gradually become pervasive and apparently permanent. Rather than improving race relations, the affirmative action programs seem to have made them worse by encouraging people to think in terms of race and ethnicity.

Indeed, foreigners looking at the US program and contemplating something similar for their own countries have much to learn for the US experience.

References

Balch, (1996). What Professors REALLY Think of Affirmative Action, Wall Street Journal, November 26 (editorial page).

Banfield, E. C. (1974). The Unheavenly City Revisited Little Brown.

Gottfredson, L. S. (1986). The g Factor in employment: A Special Issue of the Journal of Vocational Behavior, 29 (3).

Gottfredson, L. S. & Sharf, J. C. (1988). Fairness in Employment Testing: A Special Issue of the Journal of Vocational Behavior, 33 (3).

Hacker, A. (1992). Two Nations

Herrnstein, R. J. & Murray, C. (1994). The Bell Curve: Intelligence and Class Structure in American Life. New York: The Free Press.

Jensen, A. R. (1980). Bias in Mental Testing, New York: The Free Press.

Jensen, A. R. (1981). Straight Talk About Mental Tests, New York: The Free Press.

Jensen, A. R. (1985). The nature of the black-white differences on various psychometric tests: Spearman's hypothesis. Behavioral and Brain Sciences, 8, 193-219.

Levin, M. (1994). Comment on the Minnesota transracial adoption study. Intelligence, 19, 13-20.

Lynn, R. (1994). Some reinterpretation of the Minnesota transracial adoption study. Intelligence, 19, 21-28.

Miller, E. M. (1994a). Paternal Provisioning versus Mate Seeking in Human Populations", Personality and Individual Differences, 17:2, 227-255.

Miller, E. M. (1994c). "The Relevance of Group Membership for Personnel Selection: A Demonstration Using Bayes Theorem", Journal of Social, Political, and Economic Studies,, 19:3, 323-359.

Miller, E. M. (1995a). Environmental Variability Selects for Large Families only in Special Circumstances: Another Objection to Differential K Theory", Personality and Individual Differences, 19:6, 903-918.

Miller, E. M. (1995b). "Race, Socioeconomic Variables, and Intelligence: A Review and Extension of The Bell Curve", Mankind Quarterly, XXXV:3, 267-291.

Mischel, W. & Metzer, R. (1962). Preference for Delayed Reward as a Function of Age, Intelligence, and Length of Delay Interval", Journal of Abnormal and Social Psychology, 64, 425-431.

Moffett, Matt (1996). A Racial 'Democracy" begins painful debate on affirmative action. Wall Street Journal, Aug. 6, p. 1.

Schmidt, F. L. (1988). The problem of group differences in ability test scores in employment selection. Journal of Vocational Behavior, 33, 272-292.

Seligman, D. (1992). A Question of Intelligence. New York: Birch Lane Press.

Synderman, M. & Rothman, S. (1988). The IQ Controversy, the Media and Public Policy. New York: Transaction Publishers.

Wigdor, A. K. & Garner, W. R. (Eds.). (1982). Ability Testing: Uses, Consequences, and Controversies, Washington: National Academy Press.

Zelnick, Bob (1996). Backfire: A Reporter's Look at Affirmative Action. Washington: Regnery Publishing.