Assessing the Claim that Law School Affirmative Action May Actually Lead to Fewer Practicing Minority Lawyers: State Bar Data Relating to Claims of a "Mismatch" Effect
By VIKRAM DAVID AMAR
|Friday, Aug. 31, 2007|
Three years ago, UCLA Professor Richard Sander published a very controversial law review article in which he questioned, among other things, whether race-based affirmative action and other preference programs in law schools actually increase the number of minority lawyers practicing today.
Sander focused on the fates of law students who attend law schools at which the median LSAT and/or college GPA numbers are substantially higher than their own. He asserted that such students, many of whom are students of color, would have ended up with far better law school grades - and a significantly higher chance of passing the bar exam - had they instead attended law schools that were more commensurate with their LSAT scores and college grades. He contended that the legal education system has "mismatched" such students, placing them in schools where they are unlikely to succeed and learn. He argued that if more minority students were "matched" into environments where their law school grades would be higher - in less competitive law schools, that is - then the system could improve the overall bar pass rate of racial minorities, and actually increase the racial diversity of the bar.
Earlier, I wrote a series of essays commenting on some of the methodological moves Sander's piece made, and questioning what the policy implications ought to be, even if some of his empirical analyses were correct. Other scholars took direct issue with Professor Sander's empirical findings themselves, arguing (among other things) that the data on which he relied was too imprecise and too dated to generate the kind of conclusions he drew from it.
Over the past year or so, Professor Sander and some other scholars - whose views on race-based affirmative action are quite varied - have starting seeking more thorough and more recent data from various state bars, especially California's, on this issue. If procured, such data may well go a fair way towards either debunking or confirming Professor Sander's hypotheses about so-called "mismatch effects."
Although I don't necessarily share Professor Sander's intuitions about precisely what the information will show, and indeed I may end up interpreting the data differently than he does, I joined the research group's request to have the data analyzed because I think questions like these should be tested against available real-world information.
However, many state bar officials seem reluctant to facilitate the inquiry. California, for example, has, at least for the moment, turned down the request before it, although the reasons it asserts - protection of individual privacy and internal bar precedent - don't hold up very well, given the specifics of how the project is designed and the fact that the bar has provided similar information to other research efforts in the recent past.
I believe that this project, and the larger issue of which it is a part, are quite important. Moreover, neither is going away anytime soon; one of the Commissioners of the US Civil Rights Commission wrote an Op Ed in last week's Wall Street Journal discussing the topic and highlighting its centrality.
In the balance of this column, I'll set forth in a bit more detail why I think state bars ought to facilitate empirical inquiry of this sort.
What the Bar Data Could Show
First, the data is very relevant to the current debate. Although each state's bar might have different kinds of information in its files, the data fields seem rich. The bar data from California, for example, is likely to include, for some years at least, specific information about each test taker's race, her actual bar exam performance (not just whether the taker passed or failed, but her raw bar exam score), her law school and undergraduate grades, her LSAT score, and the median LSAT score and college GPA of the law school she attended, among other things.
Thus, it may be possible to compare the actual California bar exam performance of, say, the following two groups (whose numbers I have made up for these purposes): The 900 persons over the past 5 years with a 3.4 college GPA and an LSAT score of 160 who attended a law school where that GPA and LSAT score were very close to the medians, with the 250 persons over those same 5 years with the same college GPA and LSAT score but who attended a law school where the medians were substantially higher (i.e., so-called mismatched students).The Case for the Bar's Releasing Data for Professor Sander's and Similar Studies
In a letter I wrote to the California bar this spring, I provided an overview of why analysis of this data is warranted. Here are some (slightly edited) excerpts:
From my vantage point as a scholar and (sometimes) law school admissions committee member within the University of California, I have generally tended to favor the use of some form of race-based admissions programs. I believe such programs are permissible under the U.S. Constitution but, of course, since 1996, they have been forbidden to public entities in California. I was thus troubled by the thesis asserted in Dr. Sander's noteworthy Stanford Law Review article a few years ago, which was that such problems were actually harming minority law students.
I spoke and wrote about Professor Sander's study after it was released, trying to suggest both some limitations on his findings based on some unanswered empirical questions, and some alternatives to improve, rather than cut back on, race-based affirmative action policies, even if Sander's basic assertions turn out to be descriptively accurate.
After reading, analyzing and discussing Dr. Sander's work and that of his critics, I became convinced of two things: First, there are too many assumptions and suppositions that underlie various rather entrenched arguments and conclusions that people embrace on these matters. Yet these assumptions and suppositions could, in theory at least, be verified or disproved by additional empirical data that should, with effort, be discoverable. Second, many people - liberals and conservatives alike-- are uncomfortable discussing the factual realities regarding this issue. I believe Professor Sander's project has the potential to substantially alleviate both of these constraints on the national conversation about race
The relevant bar data can help resolve (or move towards the resolution of) many empirical questions to which one must know the answers if one is to form an intelligent view on the mismatch debate. Importantly, I haven't seen or heard anyone - even critics of the project proposal - who denies the significant relevance of much of the data bars have within their control to this debate.
To be sure, there are important questions of how far particular pieces of data can take us. For instance, individualized bar exam performance is not, of course, an ideal measure of what a student has learned or mastered during three years of law school. But everyone who takes the bar test does try to pass it with some margin of comfort; and if "mismatched" students are not achieving the same margin of comfort as their non-mismatched counterparts at other (less elite) schools, that is undeniably a relevant piece of information in this larger debate.
So too, while we might each quibble over whether a particular law school is or ought to be considered more "elite" than another, having information about the graduates and test takers of each and every law school makes identification of "mismatched" students much, much easier. And the more easily you can identify such students and their (otherwise similarly situated) non-"mismatched" counterparts at other law schools, the more easily you can then do the work of measuring any mismatch effect that may or may not exist.
I myself don't know how much of a mismatch effect our analysis will find. I would be quite pleased personally if regressions and other analyses debunk or tend to cut against the mismatch hypothesis advanced in Dr. Sander's Stanford Law Review piece. And I want very much the research group to run as many different kinds of tests and analyses as we can think of - using a variety of alternative assumptions about the way schools pick students, and the way students go about the task of learning the law - to make sure the research can be used by scholars across the ideological spectrum to make their best arguments.
But more than anything else, I want to move the debate forward and encourage - perhaps even force - people to confront whatever reality seems most likely to exist, even if our sense of that reality is not free from uncertainty, so that we can adapt policy (or perhaps just leave it alone) in a way that is most likely to accomplish our shared objectives of diversity and fairness.
Select a Job Title