Saturday, January 20, 2007

Law Prof Diversity, Hiring, and Tenure

Do women or non-Caucasian minorities have trouble winning tenure at U.S. law schools? Prof. Harrison worries that class biases might make the tenure review process especially difficult for such professors. As he observes, we for present lack very solid data on that front. But we have got pretty good data about how well women and non-Caucasian minorities do at winning academic jobs at law schools—a necessary prerequisite to winning tenure. That data suggests that, at least in terms of hiring, women and minorities enjoy significant advantages.

For the last fourteen hiring seasons, the American Association of Law Schools (AALS) has collected data about how well the candidates listed in its Faculty Appointments Register did at finding academic jobs with law schools. (Presumably, those candidates got hired, if they did, after being interviewed at the AALS's annual Faculty Recruitment Conference, colloquially known as the "Meat Market.") The AALS has published that data in table form in its Statistical Report on Law School Faculty and Candidates for Law Faculty Positions (2005-06). I here recreate select portions of that data graphically, so as to better illustrate the relative success of women and non-Caucasian minorities.

This chart, using data from Table 13B of the Statistical Report, shows how well women have fared relative to men at landing jobs via the AALS faculty recruitment process:

http://www.tomwbell.com/images/AALS_Hiring_Wom.gif

This chart, using data from Table 13C, shows how well minorities have fared relative to non-minorities:

http://www.tomwbell.com/images/AALS_Hiring_Min.gif


For a summary of how various types of candidates have done, on average, over the last 14 hiring seasons, consider this data, from Table 13E of the Statistical Report:




Candidate Type Success Rate (%)
Minority Women 18.5
Minority Men 17.5
Non-Minority Women 15.0
Non-Minority Men 11.3


I could say a lot more about this data, adding caveats and analyses. I've written about the topic several times before, though, and don't want to tax anyone's patience by repeating myself. For some more recent blogging about the American Bar Association's causal role in these observed hiring trends, see Gail Harriot's recent series of posts.

I'll just say this, for now: Having gone through the meat market process three times, and having served for many years on my school's Appointments Committee, I find the relative success of males and non-minorities in the mid-to-late-90s the only surprising thing in the above data. Perhaps we can explain that divergence from the normal hiring pattern as an effect of the relatively tight job market in that era. Note, after all, that the percentage of all candidates hired, regardless of their sex, race, or ethnicity, hit all time lows around that time.

[Crossposted to MoneyLaw.]

22 comments:

Robert Siegmund said...

I'm no expert with statistics, but it seems to me these graphs don't say much without knowing how many applicants applied and their breakdown.

For example, if there were 1000 men and 100 women applying for 50 available seats, you could very well see a much higher success rate for women applicants. But what would that tell us?

Tom W. Bell said...

Robert: I take it that you mean to say that a small sample set might generate statistically meaningless variations. True enough. But as you can see by recurring to the AALS tables that I cite, the numbers of candidates under consideration are in most instances pretty big. Also, note the table at the end of my post. It covers enough cases to largely iron out the bumps caused by small sample sets.

Jeff Milyo said...

There is a deeper point than sample size here, and this may be part of Robert's concern:

The data above suggest favorable treatment of female and minority candidates under the assumption that
all such candidates share similar distributions of job-relevant attributes within group (ie, you're comparing apples to apples).

However, since favorable treatment of one group over another is an open question, there is no reason to think favorable treatment occurs only at hiring. For example, if women and minorities face significant barriers along the path to law school, then only the most resourceful and diligent will make it into the sample of applicants. This means that women and minority candidates are on avergae "better" applicants than men, so should be more successful in getting jobs (ie, apples and oranges).

Of course, it is also possible that women and minorities recieve favorable treatment all along their educational path; in which case, they should have a lower success rate in a fair hiring process.

These data are an interesting first step to answering issues of bias, but by themselves utterly uninformative without some further information\assumptions about the selection process involved here.

Ryan Waxx said...

I actually think he was saying that unless you produce X substatistic, he'll dismiss the whole thing. Note that he doesn't provide any coherent reasoning for the idea that the applicants-to-hirees is vital, just that if you don't provide, then 'the graphs don't say much'.

In actuality, the way affirmative action works would tend to inflate the minority applicant pool, since AA mandates that the organization conduct extensive searches for applicats of the 'right' gender or color.

Conversely, if you spent months or years seeking out minority enclaves and peppering them with ads to increase your minority applicant pool, wouldn't you be predisposed to hire the fruits of those efforts, rather than throw all that effort away and simultaneously risk deaccredation by hiring some white male who applied via the normal process?

Bob Hawkins said...

Rather than try to read the minds of previous commenters, let me make a general point.

There is an inherent problem with interpreting statistics that are the result of aggregating disparate groups. The within-group correlations are usually not the same as the between-group correlations, and this can lead to paradoxical results.

To give a contrived example: suppose that students whose ancestors came over on the Mayflower apply only to elite Ivy League law schools. Their families expect it of them. And suppose that elite Ivy League law schools prefer Mayflower descendants, and accept them at a slightly higher rate than run-of-the-mill applicants.

When all law schools and all applicants are aggregated, you will find that Mayflower descendants are accepted at a lower rate than average. (The rejection rate at elite Ivy League schools is much higher than at the average school, and even their preference for Mayflower babies can't overcome that.) You will conclude that Mayflower applicants are discriminated against, when the only discrimination they encounter is in fact favorable.

This is not a rare situation at all. Groups are often defined in part by having different within-group correlations. Having a much higher rejection rate is a big part of being an elite school.

The rejection rate of Mayflower applicants is the result of the interaction of the internal characteristics of the group "Mayflower descendants" and the group "Ivy League law school." By aggregating, you destroy your ability to properly interpret the results.

(BTW, the erroneous belief that the characteristics of the whole are the characteristics of the part, is called "the ecological fallacy." Nothing to do with whales and wetlands, it's just a label.)

Anonymous said...

I agree that these statistics aren't particularly helpful, because we don't know about, as Jeff Milyo said, "job relevant attributes" within the group. For example, it might be that white males with weak credentials are willing to go on the job market at a greater rate than members of other groups who have weak credentials--perhaps because white males are less risk-averse, perhaps because they are more confident. (I just made this up; I'm not saying this is the case.) If that were true, we would expect to see white males succeed at a lower rate than other groups precisely because of their relative qualifications.

I guess the bottom line is that we can make up a lot of stories to explain this data, and we can't know which story, or mix of stories, is right until we have more information and a rigorous analysis of that information.

Anonymous said...

What would be interesting, then, is somehow to compare the qualifications of minority and female candidates with white male candidates. The perception I have, based on my conversations with friends in the academy, is that the former are generally less qualified than the latter (of course there are exceptions), that is if one considers the normal factors identified in law faculty hiring: law school, clerkship, government service, private practice, advanced degrees, and publication.

Ann Bartow said...

The fact that women are disproportionately hired for nontenure track jobs substantially undermines your claim that women "succeed" at higher rates than men, if success means landing tenure track job. See e.g. this: http://www.aals.org/statistics/0506/0506_T2A_tit4_8yr.

Anonymous said...

It seems rather attenuated to link one's performance on the job market to performance in a tenure review. Women and minorities may get their feet in the door, but there may be all sorts of things that impede one's progress towards tenure once you're actually at a school -- lack of mentors, teaching responsibilities, overwhelming committee responsibilities, etc.

It probably doesn't help one's standing with colleagues and students to have blogs insinuating that minority and women hires likely are the products of affirmative action measures, rather than merit.

Anonymous said...

Maybe it doesn't "help one's standing", but if the insinuation is accurate, what is wrong with getting it out in the open. Is it really disputed that, one the whole, the minority candidates are not as qualified as the non-minority candidates? It seems to me that there is an increasing trend towards hiring minority candidates so long as they are at least marginally qualified. That is to say that the schools value something beyond top-notch pedigrees and are seekign diversity as well. Whatever one might think about that goal, it seems to me that it is hard to dispute. I have even been told that I would do much better on the market if I were a woman or better yet a minority woman. It's definately a sad development, but one which is probably true.

Tom W. Bell said...

Thanks, all, for your comments. I'll try to briefly address some of the remaining issues.

I agree that the data alone offer a merely interesting start to further inquiries. I didn't offer it as the final say, and in fact made clear that it comes with many qualifications. I'm presently seeking data on how certain widely-regarded credentials correlate with the various categories of candidates that the AALS tracks.

Ann: Thanks for your cite to that interesting paper. I agree that the question of women's success at winning tenure merits further inquiry. I deliberately did *not* say that the data at-hand shows that women have particular success in landing tenure-track jobs. It does not. It speaks only of candidates' success at landing academic legal jobs--a necessary but not sufficient condition to winning tenure.

Anon: I suppose that some of my colleagues, indeed some people generally, frown on publication of this data. I find, though, that most people value simply learning interesting and relevant facts.

I don't think that I presented it with any particular insinuation. Elsewhere, however, I made quite clear what I think about the legality and morality of discriminating against would-be law profs on the basis of their sexual, racial, or ethnic characteristics. Other disagree with my views on that count, doubtless. But people of good will can still agree to dispassionately mull over the relevant data.

Anonymous said...

Tom,

I agree that some may find it useful to ventilate the data, but others might say that you are unfairly painting all minority and women hires with the same brush. And there are quite a few minorities and women who are newly hired and have excellent credentials -- indeed, better credentials than yours. I suppose the next step would be to say that they are well-credentialed because they have benefitted from diversity preferences at all stages of their careers...

In any event, I think I would agree with Ann that your data is more nuanced and complicated than you have suggested in your original post.

Anonymous said...

These data are not very useful except to say that for the time period in question, on average, women and candidates of color were hired at hire percentages than white men. There is so much that we don't know that it is irresponsible of Tom to present this the way he does. The irresponsibility is the very strong implication that women and folks of color are getting something that they don't deserve; the normative implication here is so strong as to be irresponsible.
Suppose for example that prior to the 90s white males constituted a large majority of law professors because the legal academy discriminated against all others. Suppose also that they also constitute a large percentage of the applicant pool. These two factors alone would give us reason to believe that the hiring rate for white men who differ from all others. That is there would be a market correction for the previous discrimination in favor of white men and given the fact that they constitute a larger percentage of the applicant pool, we would expect to see some robust differentials: non-white males would be hired at higher rates to make-up for past discrimination and white males would be rejected at higher rates because there are way too many of them in the hiring pool.
We know nothing of these factors (though we can suspect what the facts might be.)
In addition, we don't know anything about the distribution of law schools that hire women and candidates of color. Are they hired primarily by top schools? Evenly by all schools? Primarily by less prestigious schools? You need only look at the raw data of the plethora of schools that have two or three professors of color on their faculties to know that whatever Tom's selling, we should not be buying.

Tom W. Bell said...

I'm not quite sure that I fully understand those who would fault me for republishing data that the AALS itself generated. I noted that it comes with caveats. Indeed, I did so more clearly than does the AALS, which simply lays out the data in a series of tables. True, I might have said still more about how to interpret the data. Had I known the short post was going to be so badly (deliberately?) mischaracterized, I think I would have. I guess I underestimated how worked up people would get.

I get the distinct impression that many of my readers would have preferred that I had never aired this data. It does suggest (not prove!), after all, that women and minority candidates received prefential treatment in seeking legal academic jobs. And I guess some people find that an uncomfortable thing to discuss. I do, too. But I'd rather we confront the facts, such as they are. We're likely to discover we need more and better facts. I'm working on that, now. But, in the meantime, let's not think we can deal with the issues by hiding from them.

Anonymous said...

Tom:
The problem is not the fact that you posted the data, the problem is the inferences that you are drawing from the data that you've posted. The data do not suggest that women and candidates of color are getting preferences. To say that is being irresponsible. It is basic statistics. You can't talk about preferential treatment until you've said a lot more about pool size of each group, past discrimination, and qualifications of each group,etc. If after you've controlled for some even basic variables, you can tell us what the data suggest. All the data says is that for the years in question, on average, women and candidates of color had a better success rate in the market. We have no idea why that is from the data that you've posted. You need to provide a lot more information than that to say anything else. So, the objection is not ideological (that is, it is not the case that I don't like the inference that you are drawing.) The objection is that you are clearly permitting your normative commitments to color your interpretation of the data. I'm sure you're looking into this further and you may ultimately be right, that women and candidates of color received preferential treatment. But on the basis of what you've posted thus far, to suggest such a conclusion is to irresponsible at worst and at best premature.

Tom W. Bell said...

Anon, my post said only that the data suggests that women and minorities enjoy significant advantages. I did not say that those accrue because of preferential treatment.

I do think, as I said in the comments, that the data also suggests preferential treatment at work. And perhaps that over-reaches. But I base that supposition on the AALS data plus other information, and I hardly think it radically unlikely.

Perhaps the walk-away claim is this: You should not make any claims about preferential treatment unless you can document them very, very well. That would, I think, pose an evidenciary requirement not required in most other areas of legal inquiry. But people get more worked up about this topic, as a practical matter. I'm inclined to air almost any theory that the extant data renders plausible, myself. Why treat this topic as especially sacrosanct? (I mean, apart from the problem of being villified.)

Anonymous said...

Anyone in legal academia who claims that minority faculty candidates don't receive preferences is either a fool or a liar. But Tom's data really underestimates the situation, because (1) minority candidates on the AALS generally have fewer of the traditional credentials for teaching; and (2) not all "minorities" are treated the same. The ABA, for example, doesn't really care if a law school has any Asian American faculty, yet they are a high percentage of "minority" hires. The data would be far more interesting if it were broken down by minority category.

Nothing in the above paragraph, by the way, is at all inconsistent with (a) the view that many minority candidates have sterling credentials and (b) that some minority candidates wind up more or less where they would be regardless of preferences.

Fabio said...

I think that Mr. Bell is doing the right thing by presenting public data. However, I think that the claim that "... that the data also suggests preferential treatment at work" needs way more support. As the comments noticed, not only do the time series fluctuate (in the mid 1990s, males had higher rates ), but some of the other differences might be statistically insignificant, even with a large sample size.

As a statistically oriented sociologist, I would like to see:

- analysis of hiring rates controlling for individual credentials

- analysis of hiring rates controlling for type of institution.

- analysis of job searching behavior

- analysis of specialties desired by law schools

This analysis would test various hypotheses. For example, one might think that there is a selection process at work: weak minority candidates are less likely to apply for jobs. Also, it might be the case that minorities are sorting into weaker schools, where they might be the strongest applicants. It is also possible that law schools are hiring in areas that are more popular among minority and female candidates (e.g., civil rights law). Finally, one might suspect that certain groups are more persistent, which might explain higher rates. If the effects persist in most of the analyses, then I would believe the claim.

My suspicion is that further analysis would show a small advantage for women/minority candidates in some circumstances. The data Mr. Bell presents already shows modest effects. Let's look at some odds ratios (ratio of success vs. non-success):

Minority Women .22
Minority Men .21
Non-Minority Women .17
Non-Minority Men .12

In other words: Minorities have about 1 in 5 odds of getting a job; majority women have about 1 in 6 odds; majority men have about 1 in 8 odds.

The difference is mainly between majority males and everyone else. But it is also the case that the difference is not huge. When dealing with rare events - remember, 80% don't get jobs - small differences can translate into big jumps in odds ratios. So the gap between majority men and everyone else shouldn't be overplayed.

The issue of faculty hiring practice is important. But I hope in the future, we'll have better data. Maybe Mr. Bell can collect the right data? Why not survey a few hundred randomly selected job seekers this year and gather some basic data about their credentials and job search routines? It would be a huge improvement over this aggregate data, which is really inconclusive.

Anonymous said...

I have the impression that the advantage for women will dissapear, or at least will be much smaller if you focus on top 20, and even more so top 10 law schools.

Northerner said...

Anyone who is familiar with law school faculties could easily name many, many schools where the young white male hires are Supreme Court clerks and/or someone with 4 or 5 published articles, while the minority hires may have a student note and a district court clerkship (if any at all). To be sure, there's a very good argument that clerkships/grades shouldn't be the be-all-and-end-all of law school hiring, but to the extent that law schools DO still use those criteria in determining who to hire, any perusal of the top 50 law schools' websites will show that law schools almost always hold whites (and especially white males) to a higher standard in that regard.

Please don't deny this, folks; you know damn well it's true; and I don't want to have to name names.

Anonymous said...

No one should conclude, on the basis of the figures you've presented here, that Asian male candidates do better than white male candidates. It seems all too likely that, when it comes to a hiring preference for "minority" candidates, we are actually talking about a preference for black, Hispanic, and Native American candidates.
You should split out Asian male candidates if you can and see what effect that has on the numbers. It would obviously be quite unfair, and even harmful, to leave people with the impression that Asian males enjoy an advantage that they do not, in fact, enjoy.

Juan said...

It's great that women and minorities are experiencing success at these recruitment conferences. It wasn't all that long ago that there were very few women and minorities as law school professors.

Juan Rodriguez
diversity careers