A True Test: In Defense of the SAT

Question: On average, blacks and Latinos score well below whites and Asians on the Scholastic Assessment Test (SAT). What should we do about this? (a) Take race into account when considering test scores. (b) Allow blacks and Latinos to go underrepresented, as a proportion of the population, on elite campuses. (c) Scrap the tests. (d) Overhaul our educational system from the bottom up.

Until recently, of course, the answer of most college admissions offices was (a). But in November 1996 California voters decided to declare that answer unconstitutional when they passed Proposition 209, banning affirmative action. Ever since then a vocal group of Californian academics and activists has been arguing for replacing (a) with (c), in order to maintain levels of minority admissions to the state university system.

To be sure, it's not the first time the SAT has come under fire. For the past 35 years there has been no shortage of people who think that standardized tests are the work of the devil: Marxists who regard them as instruments of class segregation, postmodernists who ridicule them for making naive distinctions between "right" and "wrong," and, most important of all, ethnic activists who argue that they are suffused with white middle-class values.

But Proposition 209 has provided the anti-testing movement with a new cause—and a new legitimacy. Last September, for example, the University of California's Latino Eligibility Task Force proposed disregarding SAT scores in university admissions entirely in order to counteract the effects of abolishing affirmative action. So far the university has resisted such moves, but the university's president, Richard Atkinson, has pronounced the idea "intriguing." And the anti-testers, led by Eugene Garcia, the dean of Berkeley's School of Education, continue their agitation.

Garcia and his allies have undoubtedly hit on a subject that worries opponents and supporters of affirmative action alike: the huge gap between the average SAT scores of whites, on the one hand, and Latinos and blacks on the other. In 1995, for example, only one black student scored 750 or above on the mathematics portion of the SAT for every 89 whites. Even Ward Connerly, the architect of Proposition 209, briefly welcomed the idea of scrapping the SAT.

In picking on the SAT, the anti-testers have also found a particularly vulnerable target. Apart, of course, from the companies that make millions of dollars out of devising and administering them, nobody much likes tests that reduce knowledge to a set of isolated facts, ripped out of context and divorced from interpretation; turn reasoning into a mechanical exercise in ticking boxes; and drive a fair number of the million or so people who take them each year to despair.

The people who design these tests argue that all this is a price worth paying because the SAT is much more objective than other forms of assessment. Yet it's no great secret that you can boost your SAT score substantially by cramming—nor is it any great secret that minorities and the poor lack the resources to take advantage of classes that provide the most useful cramming. What's more, a good SAT score is hardly a guarantee of academic success. In studies of several hundred colleges, the correlation between test scores and freshman GPAs was only in the range of 0.42 to 0.48.

Some of the anti-testers are populists who are opposed to introducing any system of rationing into the elysian world of the American campus. Some are know-nothings who believe that, since all forms of testing are fallible, we might as well distribute university places by lottery. But there are more persuasive reformers, too, and they base their arguments not on the evils of selection in general but on the particular ills of the SAT. All they want to do, they argue, is to devise a more sophisticated system of selection: one that gives students more of a chance to express their intellectual talents and creative powers than these multiple-choice monstrosities.

Ah, if only it were that easy. One obvious problem with this critique is that it rests on an exaggerated view of how blindly and exclusively public universities rely on the SAT. SATs are primarily used as screening devices at highly competitive institutions such as Berkeley. Most prospective students can get into lesser institutions almost regardless of their SAT scores. And the most exclusive universities use the SAT alongside a range of other indicators, such as grade-point averages (GPAs), standardized achievement tests (which are designed by the same company that administers the SAT and test knowledge in specific subjects, like history or math), teacher recommendations, and other school records.

The real question, then, is what role SATs should play in the mixture of tests that are used to decide who gets into elite universities. And the problem facing the SAT's critics is that putting less weight on these tests could end up making the already fraught system even worse than it is at the moment.

To succeed in the peculiar conditions of American higher education, selection systems must have two attributes. First, they must be susceptible to mass production. At Oxford, for example, the selection system avoided the most obvious defects of the SAT. Candidates wrote large numbers of one-hour essays; their prospective teachers spent days pouring over their manuscripts; and what mattered was the quality of the arguments rather than the mere ability to regurgitate facts.

But applying that model in the United States would be difficult. Oxford sits on the top of a ruthlessly selective system of education that sorts people into streams early in life and devotes a disproportionate amount of public resources to grooming the elite. In the United States, with its emphasis on providing everybody with second chances, such a system would be impossible. The University of California, which selects more than 25,000 students each year, has no choice but to reduce selection to something of a routine. Essays and projects will inevitably end up being evaluated by admissions officers according to one predetermined formula or another.

The second requirement of selection systems is that they must appear as objective as possible, relying on precise measurement rather than fine judgment. In a country as ethnically divided and litigious as the United States, universities need to be able to point to value-free measures in order to minimize strife. The Oxford dons who pronounce mystically on whether a candidate possesses "a trace of alpha" would soon create a backlash in the United States--particularly since such a large number of those who show up favorably on the alpha detectors turn out to be Old Etonians.

To be sure, an alternative system could involve a more fine-tuned standardized examination instead of a group of professors scrutinizing essays. But there is no obvious reason to believe that more sophisticated tests would be more likely to favor underrepresented minorities than the SAT. Why should somebody who has never been taught about Archduke Ferdinand suddenly blossom when asked to write an essay on the origins of World War I? The only way to stop children of privilege from pulling still further ahead under such a system would be to rig it in favor of minorities--by awarding extra marks to candidates who bring a "different perspective" to their work or maybe by asking questions less dependent on specific areas of knowledge.

But students who lost out as a result of such rigging would be unlikely to stay quiet. A more sophisticated selection system might rapidly degenerate into a game of power politics, in which vocal pressure groups agitate for their definitions of merit to prevail and in which recourse to the courts becomes routine. The left would give high marks for self-expression (particularly if the self being expressed had been touched by suffering or shaped by social exclusion); the right would counter that what matters is knowledge of traditional academic subjects; and the nation's campuses would be torn apart by a culture war that would make today's guerrilla skirmishes look benign.

Because of these manifold problems, the critics of the SAT are being forced to resort, for practical purposes, to more mundane measures of promise, such as class ranking and GPA. Yet there is little reason to think that these devices will do anything to improve the fortunes of racial minorities. In 1995, almost twice as many white students as black ranked in the top tenth of their class, according to the College Entrance Examination Board. Only eight percent of black students had an "A" average compared with 21 percent of whites and 27 percent of Asians.

Indeed, when the University of California looked into the implications of abolishing the SAT, it found that the impact on the racial composition of the university would be mixed: The number of Latinos eligible for places would rise by five percent, but the number of blacks eligible would decline by 18 percent. Fourteen percent more whites would qualify for places, while the number of eligible Asians would fall.

These alternatives are also less reliable than the SAT. A recent study by the Educational Testing Service found that SATs are better than high school GPAs in predicting freshman grades in nearly all subjects. By the same token, the Law School Admissions Test—the LSAT—seems to be a substantially better predictor of first-year performance in law school than undergraduate GPA.

GPAs have always been unreliable indicators because different schools use different grading systems. And, even within schools, different teachers grade according to different standards. A 1994 study by the Department of Education, for example, found that students receiving an "A minus" in schools in impoverished areas might well have received a "C" or "D" in schools in more affluent areas. GPAs are getting steadily less reliable, too, because of the impact of grade inflation. The percentage of college-bound seniors reporting an "A" average increased from 28 percent in 1987 to 37 percent a decade later, and this inflation is only likely to get worse if universities start giving more weight to school grades. Judging students by their GPAs will inevitably involve admission officers in judging the reliability of schools as much as the promise of their graduates, a process that is unlikely to redound to the advantage of the inner cities.

Another alternative is to provide automatic admission for the top performers in each school. This has the advantage of measuring children against competitors from similar backgrounds. It also eliminates one of the most obvious biases from the old system of affirmative action: the fact that the children of black lawyers get an automatic advantage over the children of white valet-parkers, since nowadays school population is more segregated by class than by race.

This solution has been adopted by the University of Texas, which has scrapped affirmative action but substituted a system that automatically admits the top ten percent of students in each high school, regardless of SAT scores, into the state's public university system. Advocates of this system such as Lani Guinier, who sang Texas's praises in a recent New York Times op-ed, claim it will actually boost minority enrollment beyond levels affirmative action produced.

But, even if true, there are two big problems with this scheme. First, it further reduces pressure on schools to improve their performance: If lousy schools send graduates to the best universities, why should they strive to teach better? Second, this scheme would force even more ill-prepared students into elite universities than affirmative action did. Under such a system, the alumni of South Central's sink schools will eventually have to compete with the alumni of Lowell High. They will just do it when they get to Berkeley, where the pressure is intense and the possibilities of remedial teaching are minimal, rather than when they are trying to get into a university.

What makes this whole exercise so ironic, of course, is the fact that, until very recently, standardized testing was an instrument of upward mobility—and not a barrier to opportunity. Napoleon used tests to introduce the aristocrat-destroying principle of a "career open to the talents." British Whigs such as Macaulay and Trevelyan used tests to open Oxbridge and the civil service, hitherto playgrounds for the idle rich, to lower-class meritocrats.

In the United States, testing played a vital role in forcing the WASP-dominated Ivy Leagues to admit Jews and white ethnics—and then blacks, Asians, and Hispanics. When, in 1933, Harvard's President James Bryant Conant launched a revolutionary scholarship program that awarded money solely on the basis of academic promise, the person he employed to detect that promise was Henry Chauncey, who later went on to found the Educational Testing Service. During a 1955 Supreme Court hearing to debate ways of implementing Brown v. Board of Education, Thurgood Marshall, then a lawyer for the NAACP, relied on testing to explode the segregationist argument that the average black child was so far behind the average white that it made no sense to educate them in the same schools. Why not test all children, black as well as white, he argued, and educate them according to their test results rather than by the color of their skin?

Real conservatives have always regarded tests as tools of subversion. T.S. Eliot, for example, argued that "a vast calculating machine that would automatically sort out each generation afresh according to a culture index of each child" would "disorganize society and debase education." Perhaps the oddest thing about the current debate on testing is the way that the contemporary left, corrupted by identity politics, Foucault-chic, and entitlement mania, has turned against what ought to be one of its most powerful tools.

True, yesterday's conservatives opposed tests because they admitted too many members of minority groups, while today's liberals oppose them because they admit too few. But these liberals should remember that pendulums swing. If they can rig the selection process one way, their successors will be able to rig it another. They should also remember that choosing students for elite education is not like choosing guests for a dinner party. One of the glories of the testing movement is that it has taken selection out of the hands of academic patrons, with their weakness for passing fads, and forced it to become a little more rigorous.

Tests do more than just provide opportunities for the deserving. They provide vital information about how both schools and pupils are doing. This allows students to take corrective action and make informed choices. It also allows society to judge the performance of its educational institutions—and to put pressure on them if they seem to be failing in their basic tasks. For all their imperfections, standardized tests are probably the most powerful instruments of accountability in education that America has.

The word "information" here is crucial. The great merit of tests is that they provide us with relatively objective information about both individuals and institutions and the great disgrace at the heart of the anti-testing movement is that it is trying to conceal information that it regards as unpalatable. What makes this even sadder is that tests are getting ever better at spotting and diagnosing learning difficulties as they unfold.

If anything, the real problem with the SAT is that we use them too little and too late. We have relegated them to the depressing role of eliminating students who are not up to scratch, rather than using them to diagnose and eliminate learning difficulties as they emerge. What we need is more testing rather than less—and more willingness to act on the results of those tests so that the poor of whatever race are no longer given such a rotten deal by America's schools.