Saturday, November 11, 2006
Are the Low Numbers of Female Supreme Court Law Clerks a Statistical Blip? (Or: 'Let's Ask Justice Thomas to Check')
Posted by Alan Childress
David Kaye (Ariz. State, Law) and Joseph Gastwirth (GW, Arts & Sci. [Stats Dept.]) have just posted on SSRN Law & Soc'y: Legal Prof., their new article, "Where Have All the Women Gone? 'Random Variation' in the Supreme Court Clerkship Lottery." Only 7 of 37 of this year's Supreme Court law clerks are women, a drop nearing 50%, leading to a popular-media outcry. Justices Breyer and Souter have publicly defended the low numbers as the product of "random variation in the applicant pool," but apparently did not convince many onlookers, some of whom assert "insidious discrimination" in the drop.
The authors test the claim using their experience with statistics (David Kaye, for instance, is one of the most respected statisticians in the law school world and coauthor of the book Prove It With Figures; and Gastwirth edited Statistical Science in the Courtroom). They conclude that the numbers are not improbable and within an expected range, given the applicant pool. [Elsewhere they have reported, similarly, that this year's drop is not "statistically significant," adding "that for all the raised eyebrows, the numbers are not so dramatic as to establish that this year's decline is anything other than the 'random variation' asserted by Souter and Breyer."] But this relatively clean bill of health only works, they note in the SSRN article, court-wide. Some individual Justices have numbers so low that it is hard to explain them by chance or fluctuations.
In any event, all the statistics-crunching and the conclusion, they point out, depend to some extent upon the foundational truth of public assertions about who applies and particularly the claim that women are underrepresented in that source pool. The authors do not claim, of course, that the numbers may be chance when compared to the larger population of women law students or even students with particular credentials. That was not their investigation. And they ackowledge that their job was hampered by working from public statements rather than disclosed real data about the applicant pool. They call for public disclosure of information on applications and hiring.
The full abstract is below the fold. Also find my take on the matter, especially the problem of using the 'pool' as the reason for this year's dip, or generally for such inquiries.
I would add (and the authors are no doubt aware) that even if the Court's public statement about the available applicants happens to be true, that pool is not randomly created. It is not "neutral" itself, to explain how a drop in numbers might well be neutral. Particular feeder judges, inflexible credential requirements, repeat-player recommenders, and all sorts of other non-random variables initially create the pool. Earlier I argued (as to clerks' influencing decision-making) that "it must be kept in mind that the hiring of the clerks is not a random and independent act--their ideologies are often taken into account both directly and through their references, feeder judges, and mentors." The same may be said for their diversity in terms of gender and race. Pointing to a blip in the source pool does not really explain the intuitively low hiring rate, particularly when the non-randomness of the pool (beyond even the hiring criteria) is controlled to a significant extent by the Justices' actions and traditions themselves.
If the Justices are not going to look for hiring factors other than those 'tried and true,' or find new feeder judges--or else at least impress upon their current sources the importance of their being more diverse in hiring--then all the statistical 'neutrality' of the Scotus hires for matching the "available pool" (or more accurately, the study suggests, for falling within an expected range in that pool, if only at the shallow end) is wholly meaningless. The Court can do better, and should.
The authors, of course, do not deny this. But the conclusion of possibly neutral numbers (and especially their more supportive conclusion as reported elsewhere) is only as good as the randomness of the source pool. That reserve is unarguably filtered through all sorts of sieves that the Justices might not accept from other institutions.
Certainly unarguable is the authors' call for public disclosure of data on which fuller examination can be made. I doubt the Court will do this anytime soon, since the Justices could say they don't answer to federal law since they are federal law. Or they could say nothing. But they did not say nothing or tell us to mind our business; instead they publicly proclaimed this a product of statistically blippery and asked us to trust them without (as far as I know) having either: (a) consulted a trained statistician, (b) consulted a court employee who has experience heading an office for years whose main job was to compile and crunch hiring numbers and claims of mere blips--hey, that's Clarence Thomas!, or (c) provided the audience with source data to verify it.
The Justices really have a duty to give such information, in my mind, now that some of them have come out in public to claim this is neutral and not an issue--once they have effectively referred to the "available pool" as an explanation. Fine, let's see that pool. And even if the source pool is down this year and so is itself not very diverse, the Court should be scrambling to find a cooler pool. [Alan Childress]
In the world of American law, a Supreme Court clerkship is a position desired by many but attained by few. Considering the career trajectory of many clerks, perhaps it is not surprising that the New York Times would give front page coverage to some disturbing facts about this year's clerks -- only seven out of the 37 -- a mere 19 percent -- are women. This outcome, moreover, represents a shocking 50 percent drop from preceding years. Yet, two Justices portrayed this year's percentage as the result of “random variation,” a claim that strikes many observers as incredible.
This essay applies standard statistical reasoning to answer two questions -- what do the numbers prove, and how strongly do they prove it? We show that this year's decline in women is not at all improbable. Likewise, if the percentage of women applying for these clerkships is in the range of what one Justice suggested, then the small proportion of women is about what one would expect.
The situation seems different, however, when one examines statistics on each Justice. Some Justices hire considerably fewer women than would be expected by chance, while others hire somewhat more. There are many possible explanations for this pattern. We marshal data to assess the plausibility of some of them, but in the end, the available records do not allow a definitive conclusion. To assure public confidence in the Justices' assurances of gender neutrality, we recommend that the Court make statistics on the characteristics of those who apply for and receive clerkships publicly available.