Standardized Tests under the Magnifying Glass: A Defense of the LSAT against Recent Charges of Bias

Article excerpt

I. INTRODUCTION

In a recent article entitled "Does the LSAT Mirror or Magnify Racial and Ethnic Differences in Educational Attainment?", William Kidder presents a statistical argument designed to show that the familiar Law School Admissions Test is biased against minorities.1 Kidder himself calls his methodological approach "unorthodox" and candidly admits that he chose it precisely because, in his view, more conventional statistical analyses "lend[] credence to regressive claims about the fairness of the LSAT."2 The question is whether his unorthodox approach will stand up to scrutiny.

A lot may turn on that question now that the United States Supreme Court has granted certiorari in Grutter v. Bollinger3 and Gratz v. Bollinger,4 companion cases involving the University of Michigan's affirmative action policies. Kidder is probably correct when he identifies "the role [of] standardized tests, like the SAT and the LSAT" as a "key issue" in the controversy over racial preferences in university admissions.5 Furthermore, if his charge that the LSAT is racially biased is also correct, it will bolster the University of Michigan's argument that racial preferences are a necessary part of a modern state university's admissions policy.6 This Essay, therefore, will examine that charge.

The importance of the Kidder article is not in the claim of test bias-such claims are commonly made, although usually unsupported7-but in the purported statistical demonstration of that bias. For those who are aware of the testing literature, such a statistical claim is likely to be surprising. To demonstrate that a test that attempts to predict law school performance is biased, one must do more than simply prove that, on average, the members of a particular group do less well than others do. One must demonstrate that the test is not doing its predictive job-that members of the group, on average, perform better in law school than the test actually predicts. Otherwise, the test is simply reporting a real gap in skills, and efforts to blame the test are ill advised at best. Nevertheless, the LSAT and other similar tests such as the SAT have been statistically examined for potential racial bias many times,8 and these examinations consistently have concluded that although African-Americans and Hispanics perform less well than Asian-Americans and Whites, the LSAT actually overpredicts the performance of African-Americans and Hispanics in law school.9 Thus, to the extent the LSAT is biased, it is biased in their favor; and in examining the Kidder article, the central question is whether Kidder gives us reason to doubt this conclusion.

It is usually assumed that the supposed benefits of affirmative action, such as enhancing the racial diversity of a class, have to be traded off against losses in law school performance that arise when race (or any other factor unrelated to law school performance) is factored into admissions decisions to override statistical indicators of performance, such as the LSAT and undergraduate grade point average (UGPA).10 If Kidder is right, there may be no need for any such tradeoff. If the LSAT is biased against minorities, one might be able to adopt a race-conscious admissions policy that actually produces a class of students who perform better as measured by law school grade point average (LGPA) than a class formed by a race-neutral reliance on test scores.

But as we show here, Kidder's argument is defective. He offers no reason whatever to believe that law schools can improve upon academic performance by abandoning the LSAT or by using race as a conscious "plus" factor in admissions decisions that could trump the results from a focus on indicators such as the LSAT and UGPA. Our conclusion is that the conventional wisdom that the putative benefits of affirmative action have to be traded off against a loss in academic performance is wisdom after all, and that we are better off not trying to wish away the existence of such tradeoffs among values. …

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.