The question of persistent racial differences in tested cognitive ability has long been politically awkward for liberals. In "America's Next Achievement Test," which appeared in our September-October 1998 issue, Christopher Jencks and Meredith Phillips confronted that awkwardness, proposing that closing the black-white test score gap could possibly "do more to promote racial equality than any other strategy that commands broad political support."
Drawing on a range of evidence, Jencks and Phillips demonstrated that because the large gap between blacks and whites on tests of cognitive skills has narrowed in recent years, it must be to a significant extent malleable. They also proposed that changing parenting practices and making a greater social investment in early cognitive development were among the most promising avenues for narrowing the gap still further in the future. Finally, in a significant revision of what Jencks had found some 25 years earlier, the authors concluded that because a black man and a white man with comparable test scores can now expect almost comparable earnings, achieving black-white parity on cognitive tests should be a major focus of public policy.
Of course, not everyone will agree. Nor, even among those who do agree, will there be consensus about how best to achieve parity on these tests. With this in mind, we asked prominent experts on economics, race, and child development to submit their comments on "America's Next Achievement Test." Claude Steele, Stephen Ceci and Wendy Williams, Mindy Kornhaber, Jared Bernstein and Richard Rothstein, and Glenn Loury all responded. Taken together, these comments hint at the broad array of issues that attend the test score gap, and make its elimination such a difficult—and politically freighted—project. As Jencks and Phillips make clear in their response to these comments, the challenge now is to take what we know and do something more concrete about the problem.
Next Issue: our readers respond
Claude M. Steele
Some people who would like very much to right racial inequality will not like the idea, proposed by Christopher Jencks and Meredith Phillips, that reducing the black-white test score gap could be a prime target of public policy. African Americans have been hammered by this test score gap for decades. Focusing on it now could seem like treating a symptom while ignoring the disease. Shouldn't we focus first on righting the structures of racial inequity and let test scores take care of themselves? Why even bring up the issue again?
I have a great deal of sympathy with these concerns. But two considerations keep me open to the Jencks-Phillips discussion. First, I can't believe that the questions raised by this gap will ever go away until the gap itself goes away. Second, I found the Jencks-Phillips idea of focusing on closing the gap appealing, as something with the potential to generate fresh policy.
But having gotten on board, I am entitled to quibble a bit about where this train is going. The premise of "America's Next Achievement Test" is that in today's world, blacks who have the skills measured by these tests will get rewarded with earnings almost equivalent to those of equally skilled whites, and that, therefore, racial equality would be served by equalizing those scores. But what really drives black earnings? Tested cognitive skills, or other things that go with these skills? Blacks with higher test scores are very likely to come from better-off families, to have more interracial contacts, to be more assimilated into mainstream walks of life and job networks, to come from more integrated schools, and so on. If blacks' earnings are really determined by factors that correlate with their tested abilities rather than by the abilities themselves, then the first challenge of policy aimed at equalizing earnings between the two races should be to find the right things to focus on.
Two things lead me to feel that this focus has to be on more than test scores per se:
First, the level of one's skill on these tests typically does not predict things about one's future particularly accurately, even when those factors are similar in kind to the test itself, like school grades. The Scholastic Assessment Test (SAT), for example, which is among the better of these tests, measures only 18 percent of the things that influence freshman college grades. So it isn't very plausible that the actual skills measured on these tests have a very direct influence on something as different in kind as earnings. The article doesn't report how strong this relationship is. I worry that there isn't much of a relationship, and that the real causes of the earnings gap are other problems for which both low test scores and low earnings are symptomatic.
Which leads to my second point: research by my colleagues and me over the years points to something more primary than the "low cognitive abilities causing low test scores and low life earnings" scenario. That "something else" is the relationship blacks have to the larger American society and the experiences that go with that. They live, work, and perform in the upward-mobility domains of school and the workplace under an extra pressure. It is the extra pressure from historically rooted stereotypes that tell them and those around them—just as other stereotypes tell a white guy trying out for the 100-meter dash—that they don't belong there. This stereotype about blacks, its pervasiveness in society, the devaluation of blacks that it causes, the climate of intimidation it establishes, the mistrust and alienation these things cause in blacks—these all influence the development of skills that underlie both test scores and earnings. And I am afraid that trying to reduce the test score gap without addressing these more fundamental issues will not get very far. The hopeful news of this and other research is that it is possible to address these problems, especially in local school situations, and that when you do, test scores improve.
From this, a conclusion: focusing public policy on closing the test score gap without giving primary attention to the context in which the gap occurs would not only be ineffective, but would also perpetuate a victim-blaming framework for thinking about racial inequity. But focusing public policy on test scores as a way of coordinating a broad remediation of these larger contextual forces—that is, by directing resources and attention to parents, homes, and communities, as well as schools—could foster just the comprehensive strategy that we need, putting, for once, causes before symptoms. And this, I hope, is the effect that the Jencks-Phillips project will have.
Stephen J. Ceci and Wendy M. Williams
Recent analyses of the National Assessment of Educational Pro gress data show that the gap between black and white children's math, reading, and science achievement scores narrowed by at least 30 percent between 1971 and 1989. Unfortunately, according to our research, further reduction in the black-white score gap ceased after 1989.
Why did black children make so much progress in closing the achievement gap between 1971 and 1989? While it is tempting to attribute the closing of the racial gap to social programs put in place during the Great Society programs of the 1960s, such as school desegregation, school lunch programs, and Head Start, the evidence for the impact for these interventions is mixed at best. But there is one unambiguous factor that all agree is associated with the gap closing.
Approximately one-quarter of this gap closing can be linked to the fact that black parents made greater relative gains in their educational attainment than did white parents during the period in question. While it is true that black parents as a group possess less education than do white parents, they nonetheless outpaced white parents in their educational increases from 1971 to 1989. In every category—from the percent of parents graduating from high school, to the percent receiving some college education, to the percent completing college and matriculating in graduate and professional programs—black parents' gains have been far steeper than white parents'. And as black parents re ceived more education, their children's test scores rose. (However, David Grissmer and his colleagues have noted that white students made few gains despite their parents' educational in creases during this same time period.)
Jencks and Phillips correctly note that traditional strategies for closing the racial divide, such as massive increases in spending and the use of race preferences in college admission, will be politically infeasible in the fu ture. We agree that it will be nec essary to identify financially modest interventions acceptable to the American public. Such proposals will probably not involve racial preferences. Our reading of the research leads us to anticipate that the following three proposals represent the best strategy:
- Reducing class size during the first three years of school. (The elevation in minority children's test scores seen in the Student/Teacher Achievement Ratio project, conducted by the state of Tennessee, appeared to cease after the first year, and smaller classes after this point did not lead to further benefits.)
- Developing preschool programs specifically geared to cognitive development.
- Promoting programs that encourage minority-parent educational attainment, such as classes for pregnant teenagers and continuing education for working parents.
Implementation of these proposals represents a solid chance for successfully building on the foundation set during the 1970s and 1980s. Black students' test scores have come a great distance in recent years—but they are still a long way away from the goal of parity.
Only four years ago, The Bell Curve's authors, Richard Herrnstein and Charles Murray, grimly announced that "For the foreseeable future, the problems of low cognitive ability are not going to be solved by outside interventions to make children smarter." In contrast, Jencks and Phillips now claim that, with great effort, the black-white test score gap should be tractable. In a sharp reversal of Jencks's earlier views, they also assert that closing the gap will foster equity in earnings, education, and other social arenas. (That these different conclusions derive from studies of cohorts who came of age before and after the Civil Rights Act of 1964 is worthy of a longer discussion.)
This is welcome work for anyone concerned with what I call "the problem of cognitive equity." My own interdisciplinary efforts to understand how hu man potential might be developed both equitably and to a high level lead me to offer these few amplifications:
First, we want more equitable achievement and intelligence test scores because the scores are associated with other important things. Jencks and Phillips note the association between scores and educational and economic outcomes. Yet it is essential to recognize that the tests simply do not measure the great bulk of what goes into making a successful student or worker. For example, neither intelligence tests nor the SAT explain even 20 percent of the variance in college students' first-year averages. They are far weaker in explaining the total college transcript. They bear little relationship to success within one's career. Despite these widely agreed-upon facts, test scores continue to be overvalued and thus misused—with consequences that run counter to closing the gap. The increased weight placed on test scores in the wake of California's Proposition 209, which barred the use of race-based affirmative action in college admissions, is the most notable recent instance of this.
Second, while we do want equity in test scores, what we want even more is equity in individuals' abilities to use their minds well. Because test scores are only an indirect and partial proxy for true mental muscle, policies aimed at boosting scores do not always overlap with policies for building minds. For example, in recent years several states and many school districts have renewed their efforts to retain students whose test scores are inadequate. Yet as research has demonstrated, retention undermines achievement and dramatically increases the chance of dropping out. Given existing differences in blacks' and whites' achievement scores, this politically popular policy will perpetuate the gap rather than close it.
Fortunately, there are intersections between the "keen mind" path and the interventions that Jencks and Phillips advocate to close the test score gap. For instance, adding cognitive content in the preschool years will give children more opportunities to learn academic content. Reducing class size may enable teachers to offer more feedback to students. Higher-scoring teachers may be more competent. However, nothing Jencks and Phillips suggest directly bears on two crucial footholds: enabling students to engage the material at length and to reflect upon their efforts. As John Ogbu's and Claude Steele's separate investigations indicate, this is where black and white youngsters' paths diverge. For African-American students (and others whose abilities are negatively stereotyped) academic engagement comes with greater psychological and social costs.
While it is reasonable to hope for the test score gap to narrow if preschool improves, class sizes decrease, and black youngsters have greater access to higher-scoring teachers, if the goal is high-level cognitive equity, or "keen minds," then equipping educators with classroom techniques, organizational structures, and social policies to keep diverse students engaged over the long haul may matter even more.
Jared Bernstein and Richard Rothstein
We agree with Christopher Jencks and Meredith Phillips that reducing the black-white test score gap is important, but we are less convinced that this is the best—or even, as they suggest, the most politically viable—way to close racial pay gaps. Jencks and Phillips place too much emphasis on the positive effects of test scores on earnings and too little emphasis on two central wage determinants: the long-term erosion of labor market institutions that formerly lifted wages of noncollege-educated workers, and labor market discrimination. Jencks and Phillips make three discrete claims:
- Many factors contribute to low black test scores. Cultural explanations should be given relatively more credence, economic explanations relatively less, and genetic explanations almost none at all.
- The black-white test score gap has narrowed significantly in the last generation, and a variety of policies—though no single panacea—could narrow it further. The most important changes, however, will be the most difficult to implement, so policy focus should be on educational policies to reduce class size and better prepare teachers. Eliminating the gap will take more than one generation.
- Eliminating the gap would be "sufficient" to substantially reduce black-white gaps in earnings as well as in crime, health, and family structure.
Our disagreement focuses on the third of these. To support their argument, Jencks and Phillips employ a figure showing relative earnings by scores on the Armed Forces Qualification Test (AFQT): blacks in their early thirties with scores above the median earned 40 percent less than whites in 1964; in the 1990s similar blacks were earning almost the same as whites.
This AFQT finding has struck many social scientists as implausible. In a recent issue of the Journal of Economic Perspectives, William Darity and Patrick Mason grant that part of the black wage gap reflects blacks' "inferior productivity-linked characteristics," but they assert that this fact should not be taken as evidence that there is no longer racial discrimination in hiring practices. They and other economists argue convincingly that racial discrimination in the labor, product, and credit markets persists. Audit studies, for example, where workers of different races but with other similar characteristics—including educational credentials—apply for a job (or loan, or apartment) show very convincing evidence of continued discrimination. Such evidence challenges Jencks's and Phillips's implicit assumption that the labor market is now truly "meritocratic" in that cognition is rewarded without—or with significantly less—regard to race.
Their argument is consistent with a widespread tendency to place primary responsibility for solving our most important economic and social problems on our public schools and then to claim that public education has failed when the problems persist. Consider, for example, the claim made by many liberals a few years ago that the United States was losing a head-to-head economic race with Japan and Western Europe because of the inferiority of American schools. Yet today, with America ascendant, few credit educational institutions with responsibility for our macroeconomic success.
This blame-the-schools conventional wisdom should make us especially careful not to place sole responsibility upon public education for closing the racial wage gap. Because those who oppose economic and social equality have their own reasons for wanting to attack public education (they seek market substitutes for all public institutions), we think an emphasis on the test score gap as the way to equalize earnings is not only less likely to be successful than Jencks and Phillips suppose but also more likely to be misused.
That said, we don't doubt that reducing the black-white test score gap would reduce black-white earnings inequality. But there are alternative strategies to solving the nation's racial gaps in income, crime rates, health care, and family structure. Furthermore, these alternatives—modifications in labor market and macroeconomic institutions—are as politically accessible as lowered class size and improved teaching, and they are more able to generate rapid results. If these alternatives engender greater racial economic equality, they may also, in turn, help to narrow the black-white test score gap. We believe that Jencks and Phillips underappreciate that the cultural differences which contribute to a test score gap may not be wholly separable from economic explanations. As John Ogbu—whom Jencks and Phillips credit for making cultural explanations respectable—has argued, relative cultural aversion to schooling may itself partly result from fatalism about future success in finding a job.
A central problem facing black workers is one that faces all middle- and low-wage workers: long-term wage decline. Minority workers are particularly vulnerable, both because of discrimination and because they are disproportionately represented in groups that have been hurt the most: the noncollege educated, those in blue-collar occupations, those in manufacturing and low-wage service industries, those earning at or near the minimum wage, and those in jobs similar to those formerly covered by collective bargaining. Over the last two decades, workers in these categories have been steadily hurt by factors such as the loss of protective trade policies, lowering of the real wage floor, workplace environments less conducive to union organizing, and, for minorities, reduced enforcement of antidiscrimination policies. Deterioration of these protections has lowered their wages, benefits, and job security relative to other workers and has given rise to unprecedented disconnection between economic growth and living standards of typical families, be they white or minority.
Consider four possible economic interventions to narrow the black-white earnings gap: minimum-wage increases; the renewal of collective bargaining; progressive taxation; and full-employment policies. None are politically easy, but they are no more difficult than closing the test score gap directly, which Jencks and Phillips estimate will take more than a generation. Because more black workers are employed in sectors where the minimum wage has power, increases in the minimum wage will themselves further narrow the earnings gap. Economists John Bound and Richard Freeman showed that a 1 percent decline in the real value of the minimum wage was responsible for about a half a percent increase in the black-white wage gap among young male workers. Over the period of their study (1973 to 1989), de unionization was responsible for 23 percent of the growth in the racial wage gap for high school dropouts, 10 percent for high school graduates. More progressive taxation—such as further expansion of the earned income tax credit—will not, of course, affect the black-white earnings gap, but it will narrow the black-white aftertax income gap and have similar impact on social and economic integration. The current labor market provides an excellent example of the benefits of tighter labor markets for minority employment. Over the last year and a half, while overall unemployment fell 0.9 percentage points, that of young blacks fell 3.5 points. Low unemployment is likely accompanied by a narrowing of the black-white earnings gap for employed workers as well. For example, over the last year and a half, the black-white earnings gap for low-wage workers—whose unemployment rates have fallen the most over this period—narrowed by 2.5 percentage points. Even if the chance of possible inflation due to low unemployment were again to frighten policymakers, a willingness to tolerate mild inflation would probably do more to narrow the black-white earnings gap than any other single policy.
In sum, closing the test score gap and working to revitalize labor market institutions are complementary. It would be a serious error for progressives to place their hopes of solving the nation's most serious social divisions and inequalities on one without the other.
Glenn C. Loury
America is at a critical moment in its age-old struggle with the di lemma of race. The progressive surge forward in race relations that began after World War II with a federal onslaught against official segregation in the southern states and that culminated with the civil rights legislation of the 1960s has been exhausted. Race-conscious policies intended to benefit blacks are under sustained attack in courts and legislatures throughout the country. Yet the subordinate social status of blacks remains a stubborn reality. America's prisons are overflowing with young black men. A third of black children depend for their subsistence on a welfare system that has recently become more difficult to negotiate. And blacks continue to lag far behind whites on every available measure of academic achievement and intellectual functioning.
The conventional partisan responses to these facts are by now all too familiar. Conservatives deny any continuing public responsibility to elevate the position of blacks. Historic patterns of discrimination have collapsed, they say, and the problems of racial inequality that remain can be attributed to the failure of blacks to seize existing opportunity. Liberals ex press alarm at the staggering disparity in life chances still separating poor Americans from the rest of us—a disparity that correlates powerfully with race. They liken the current retreat from racial egalitarianism with the late-nineteenth-century failure of Reconstruction to achieve genuine political equality for the newly emancipated slaves.
In my judgment, the conservatives are getting the better of this dispute politically, but the liberals' moral instincts are to be preferred. These instincts need to be backed by workable, saleable programs. Affirmative action–type interventions targeting blacks for special treatment through government policy seem to have a limited political future. Perhaps this is as it should be. Yet given the extent of persisting black disadvantage and the evident, deep connection between this disparity and our unlovely history of race relations, the policy of "benign neglect" implicitly urged by conservatives seems unconscionable. Desperately needed is a program of action around which public support can be mobilized that promises to bring about genuine civic inclusion for the broad mass of the black American population.
What, then, is to be done? There is no easy answer to this question. Nothing being discussed in the area of education reform today—neither expenditure equalization, nor vouchers—comes close to meeting this desideratum. Although Jencks and Phillips fall short of providing a plan of action, their focus on fostering the cognitive development of black American children through universal policies that are particularly helpful to blacks is suggestive. Reducing class size and raising teacher quality are policies that fit this bill, as is an effort to provide more than equal per pupil budgets for schools with exceptionally large numbers of severely disadvantaged children. But, as Jencks and Phillips recognize, it is implausible to imagine that such undertakings could come close to eliminating the substantial racial disparity in cognitive functioning that now exists. Moreover, it would be a long time before they began to make a difference.
It is impossible to avoid the implication here that public policy is only a part of the story. Changing the organization and support of educational institutions can have only a limited impact on this problem, and there is only so far one can go toward implicitly favoring the development of black youngsters through policy shifts that maintain broad public support. As Jencks and Phillips stress, parenting practices are crucial determinants of the early cognitive development of children, and these practices can only be indirectly and weakly impacted through public policy. So while acknowledging that schools serving poor youngsters need more resources, and while understanding that many poor parents are unusually burdened with various stresses and disabilities, we need also to recognize that it is these parents who are ultimately responsible for the intellectual development of their children.
This, of course, is true not only for the poor. Highly publicized controversies over black-white differences in intelligence, and regarding affirmative action in college admissions, make it urgently important to consider with seriousness and candor how the cognitive development of middle-class blacks can be made more effective. In 1997 black high school seniors from families with annual incomes between $70,000 and $80,000 scored an average of 472 on the verbal component of the SAT, compared to an average of 487 for whites from families whose income was less than $10,000 per year. There is no necessity, and much danger, in this situation. It should in my view occasion some serious soul-searching, and communal mobilization, among black Americans.
Why are middle-class blacks lagging so far behind even poor whites in intellectual performance? The truth is that nobody knows. Although evidence for the genetic inferiority argument is not persuasive, the psychological and cultural dynamics involved here, though much debated, are little understood. Indeed, one of the most important prescriptions made by Jencks and Phillips is their call for a massive scholarly effort at gathering the data and doing the research needed to improve our understanding of such matters. They note—rightly, and ironically—that the scale of this endeavor could be comparable to that undertaken by psychologists beginning nearly a century ago to develop the mental ability tests so widely in use today. But it remains to be seen whether this social problem, so adverse to the egalitarian aspirations of black Americans, can come to be seen as a matter of vital concern to the nation as a whole. The polarized nature of current political discourse on racial matters does not encourage optimism in this regard.
Christopher Jencks and Meredith Phillips respond:
We found all these comments on "America's Next Achievement Test" thoughtful and constructive. The fact that we agree with almost all the commentators' points may mean that The American Prospect should have sought out a broader range of critics. Or it may mean there is more consensus on this topic than we had assumed. In what follows, we will concentrate on trying to clarify some factual questions raised in these comments.
Claude Steele, who has an excellent chapter in our new book (The Black-White Test Score Gap) on how stereotype threat can affect able black students' test performance, raises a crucial question when he asks whether the correlation between blacks' earnings and their test scores reflects the fact that scores on cognitive tests are important in their own right or are a proxy for something else. We do not think the scores themselves are crucial. If the Educational Testing Service secretly added 200 points to every African American's total SAT score but the underlying racial disparity in vocabulary, reading comprehension, and math skills remained unchanged, we do not think blacks' earnings would rise much. Our argument is not that scores matter but that skills matter.
Steele suggests, however, that even cognitive skills may look more important than they really are for blacks, because they may be proxies for growing up in an affluent family, attending racially integrated schools, having white friends, and assimilating into white culture. We agree that blacks who have the same cognitive skills as average whites probably resemble whites in other respects as well. We also agree that this resemblance may help explain why employers pay higher-scoring blacks higher wages (though interviews with employers also suggest that many of them attach considerable weight to cognitive skills per se). We are not certain, though, about the causal relationship between assimilation and cognitive skills. Does cultural assimilation en courage blacks to develop their vocabulary, reading, and math skills, or do vocabulary, reading, and math skills encourage assimilation into white culture? Or do both assimilation and cognitive skills simply reflect some other set of common causes?
Although interracial contact of various kinds surely has some effect on both test scores and earnings, the effect on test scores seems to be modest. Attending a predominantly white elementary school seems to raise black children's reading scores but not their math scores. One reason the benefits of school desegregation are smaller than integrationists once hoped may be that blacks who attend school with whites are even more exposed to negative stereotypes about their mental powers than blacks who attend overwhelmingly black institutions. Having to compete with better-prepared whites may actually discourage some blacks from taking school seriously. But these explanations are really just speculations. Nearly 30 years have passed since the South desegregated its schools, yet we know little more about the possible psychological consequences than we knew before the process began.
We do not believe that vocabulary, reading, and math skills are mainly a proxy for growing up in an affluent family. Higher test scores are associated with higher earnings even when we compare siblings raised in the same family. Furthermore, as Glenn Loury suggests in his commentary, income differences between blacks and whites explain only a small part of the test score gap. A study of this issue by Phillips and colleagues in The Black-White Test Score Gap reaches the same conclusion. Comparing five- and six-year-olds whose parents had had the same average annual income since the child's birth reduced the black-white gap by only a fifth, and most of this modest reduction is traceable to the fact that black and white parents with the same income also tend to have similar test scores.
Both Steele and Mindy Kornhaber emphasize the fact that tests like the SAT measure a fairly narrow range of cognitive skills and explain no more than a fifth of the variation in college grades. It is also true that such tests account for less than a fifth of the variation in annual earnings. But cognitive skills explain the entire black-white gap in college graduation rates, and they also seem to account for most of the earnings gap between blacks and whites, especially among women.
There is no contradiction here. Cognitive skills are only one of many factors that influence earnings. But blacks and whites are far more alike on the noncognitive determinants of earnings than on the cognitive skills that standardized tests measure. As a result, cognitive skills explain a large percentage of the earnings difference between blacks and whites even though they explain only a small percentage of the variation in earnings among individuals of the same race.
The fact that a relatively narrow range of cognitive skills can explain much of the black-white earnings gap could be very good news. It might mean that we could reduce economic inequality between blacks and whites without having to eliminate all the other historical, cultural, and social problems associated with racism. This is only a possibility, not a certainty. Neither we nor anyone else knows for sure that it is possible to raise African-American children's vocabulary, reading comprehension, and math skills to the level attained by European and Asian Americans without changing white culture. But if it is possible, that is surely good news for blacks, because whites show little interest in changing their culture.
If Steele is right—if raising blacks' vocabulary, reading comprehension, and math skills requires us to root out the negative stereotypes about blacks' intellectual abilities that have been part of American life for centuries—the task is truly daunting. It is hard to eliminate racial stereotypes even when they have no foundation in fact. Eliminating stereotypes that are consistent with observed statistical averages seems almost unimaginable. Like Steele, we think that eliminating such stereotypes depends on changing the facts. Thus even if negative stereotypes about blacks' cognitive skills play a major role in discouraging black children from developing such skills, the best way to solve the problem may not be to attack the stereotypes directly but rather to make black and white adolescents' cognitive skills somewhat more equal by changing the schools and homes in which black children are raised. This should weaken negative stereotypes, as well as improve the climate in which the next generation of black children begin their schooling.
We agree with almost everything that Stephen Ceci and Wendy Williams say about the test score gap. Our one possible disagreement is over why the reading and math gap be tween blacks and whites narrowed during the 1980s. This is a crucial issue, because if we understood why things got better in the 1980s, we might be able to revive the process in the years ahead. Ceci and Williams em phasize the role of parental education. They are certainly correct that black parents' educational attainment has risen faster than that of white parents over the past generation. But if the well-known correlation between parental education and children's test scores were truly causal, we should have seen some increase in both black and white children's scores, because white parents' educational attainment also increased during the 1980s. Yet as David Grissmer and his colleagues show in The Black-White Test Score Gap, white children gained far less than one would have expected based on their parents' rising educational level. If an increase in parental education did not boost white children's scores, why should the larger increase among black parents have raised black children's scores? This is an unresolved puzzle.
We also agree with almost everything that Mindy Kornhaber says. Our main reservation concerns her argument that the psychological and social costs of academic engagement are higher for blacks than for whites. While we accept Steele's argument that the psychological costs of trying to do well in school are usually higher for blacks than for whites, we are less convinced that the social costs of academic success are higher for blacks than for whites. It seems to be true that many black adolescents accuse classmates who get A's of "acting white." But many white adolescents also put down their classmates for working hard or caring about grades. Philip Cook and Jens Ludwig show in The Black-White Test Score Gap that if one measures the social costs and benefits of academic success using national data on tenth graders, the benefits slightly exceed the costs for both blacks and whites. None theless, the social benefits of academic success are not large enough to persuade many blacks or whites to work hard in high school. In 1990 the typical white tenth grader, like the typical black tenth grader, reported doing only two to four hours of homework each week. Since blacks start out at a disadvantage relative to whites, the only way they are likely to catch up is by doing more schoolwork than their white competitors. At the moment, that is not happening.
We also favor all of Jared Bernstein and Richard Rothstein's proposals for improving the position of low-wage workers, namely raising the minimum wage, reducing legal obstacles to unionization, making taxes more progressive, and promoting full employment. And we agree that all these policies would help blacks more than whites. But none of their proposals would change the fact that blacks are concentrated at the low end of the earnings distribution. Indeed, there are no color-blind policies that would change that fact. The obvious short-term solution to the problem is to reduce racial discrimination in the labor market.
We certainly agree with Bernstein and Rothstein that racial discrimination remains common both in the labor market and elsewhere in American life. The fact that blacks and whites with above-average scores on the Armed Forces Qualification Test now seem to earn relatively equal wages does not mean that they live in a color-blind world. Anyone who imagines that they do should spend a day talking to well-paid blacks. But there is no one-to-one relationship between discrimination and wages. If black engineers thought that Texaco was run by racists, most would have looked for jobs elsewhere. Furthermore, many of those who took jobs at Texaco would have left for a friendlier environment. As a result, Texaco's labor practices would not have had much of an effect on black engineers' lifetime earnings—though they would have made lots of blacks angry and helped perpetuate racial conflict.
Discrimination has a direct negative effect on black wages only when it is so pervasive that large numbers of blacks cannot find work with any firm that provides truly equal opportunity. That was clearly the case a generation ago. It is not so clearly the case today. It is probably more true for unskilled black men than for any other group. Most employers clearly view unskilled black men as "workers of last resort," especially if they are young. When there are not enough jobs to go around, such men will be the first fired and last hired.
That said, we need to recognize that one reason blacks encounter so much discrimination is that most employers think blacks have weaker cognitive skills than whites with the same amount of schooling. This means that even blacks with skills comparable to whites have trouble getting good jobs, because skilled blacks have trouble convincing whites that they are really different from blacks with weaker skills. Racial stereotypes even skew what supervisors "see" when they deal with black and white workers. When a black worker does not know something, supervisors' prejudices are confirmed. When a white worker does not know the same thing, supervisors are more likely to dismiss the lapse as exceptional.
We also need to remember that the federal government has been trying to reduce racial discrimination in employment since the late 1960s. This effort has sometimes been quite vigorous and sometimes quite lethargic. Enforcement efforts have opened up many traditionally white jobs to blacks, but they have also generated a lot of ill will among whites. Without a smoking gun of the kind that became available to Texaco's black workers, the only way to enforce laws banning racial discrimination at work is to emphasize numerical results.
Lawyers and activists who count the number of blacks in a given job category seldom treat racial disparities in vocabulary, reading comprehension, or math skills as a legitimate explanation for racial disparities in employment. Yet when employers use cognitive tests to select workers, whites almost always do considerably better than blacks. Both black workers and government officials charged with rooting out racial discrimination often challenge the legitimacy of these tests, arguing that they do not predict job performance very well. This is usually true. (The only strong predictor of future job performance is past performance.) But if employers stop testing, or if they ignore the results of tests they have given, whites see this as an instance of hiring unqualified blacks to "make the numbers look good."
Such policies have helped undermine popular support not just for "affirmative action" but for all forms of government intervention in the labor market. Indeed, blue-collar whites' sense that the Democratic Party is committed to favoritism for blacks has played a not unimportant role in their defection to the Republican Party and is one reason (among many) why the prospects for labor law reform, progressive taxation, and a generous minimum wage are now so bleak.
We agree with Glenn Loury that black parents will have to take a much more active role in their children's education if we are ever to eliminate the skills gap. Our only disagreement is about federal, state, and local governments' ability to influence parenting practices. Governments cannot change parents' behavior simply by issuing pronouncements. But governments can persuade people. They can support more experiments aimed at finding out what works and what doesn't, and they can do more to synthesize and publicize the results of such research. Does reading to children every day really improve their school performance? We have lots of statistical correlations suggesting that this is the case, but we do not have any experiments that really prove the point. Until we do, mobilizing parents to spend more time reading to their children will remain difficult—not because parents demand experimental evidence, but because those who must pay for efforts to mobilize parents have become skeptical about whether anything can work.
This does not mean, of course, that either blacks or whites should sit around waiting for the government to "do something." The government will not do much in this area unless it is pushed fairly hard. But when black adults try to change the way black children are raised, they should be able to count on help from the government.
All in all, we find these commentaries heartening. Now the challenge is to do something more concrete about the problem.