No organization wields more influence in judicial confirmation battles, be they at the district, appellate or Supreme Court level, than the American Bar Association.
For over half a century, the ABA has issued ratings of federal judicial nominees as "well-qualified," "qualified" or "not qualified," which have shown considerable influence in the Senate confirmation process. For example, the ABA's judgment that Mildred Lillie and Hershel Friday were "unqualified" prevented their likely nominations to the Supreme Court by Richard Nixon.*
Lillie would have been the first female Supreme Court justice, and new research suggests that her gender was taken into account when the ABA rejected her. Maya Sen, a political scientist at the University of Rochester, has a new working paper** arguing that the ABA has for the past 50 years been systematically less likely to recommend the judicial confirmations of women or racial minorities than that of white men.
Sen focuses on nominations to the District Courts -- the lowest level of federal court -- in part because the sheer number of seats means that there's more data to mine than there are for appeals court or Supreme Court nominations.
First, Sen demonstrates that the ratings absolutely matter. While candidates judged qualified or well-qualified were almost certain (probably 96 to 97 percent) to be confirmed, those judged "not qualified" had only a 65 percent chance:
Then, she uses a logistic (update: not linear, thanks to comments) regression to determine what effect various factors have on a candidate's likeliness to get a given ranking. She includes not just race and ethnicity but experiential factors that should affect a candidate's ability to perform well as a judge. Those include the length of time practicing law, services as a prosecutor or public defender, quality of law school attended, and so forth. As you'd expect, time spent serving as a judge, an assistant U.S. attorney or a law clerk all increase a candidate's odds of getting a good ranking (shown by being above 0 in the chart below). But three factors are associated with increased odds of getting a bad rating (below 0 in the chart): being black, being Hispanic and being female.
Sen subjected her findings to a battery of statistical tests, and it held up throughout. The most striking finding is that she also saw no evidence that a good ABA rating has anything to do with judge's quality, as measured by the likelihood of a higher court reversing her or his decisions. The ratings haven't just been race- and gender-biased, she concludes. They've been pretty much worthless, given that reversal rates are a good indication of judge's quality.
Sen's research is in the preliminary stage, and Erik Voeten and Andrew Gelman highlight one shortcoming of the paper -- its reliance on an average of data from 1960 to 2012. So it could be that the ABA was just really, really racist and sexist in the 1960s and has softened or eliminated its bias since.
But Sen notes some evidence that suggests that such prejudice is still a problem. For example: Out of the 14 judicial candidates put forward by President Obama that the ABA has ruled "not qualified," nine are women and eight are racial minorities. You can't say, using Sen's data, that there's a causal relationship there, as Obama might have simply chosen a bunch of unqualified female or minority candidates. But it's certainly suggestive that the ABA's historical bias is still around.
* John Jenkins' new biography of William Rehnquist, argues that this version of the story is wrong and that Nixon intended to have the ABA reject Lillie and Friday so that he could appoint Lewis Powell and then-Sen. Howard Baker, his actual top choices, to the high court. Baker took too long to decide if he wanted the job, so Nixon nominated Rehnquist.
** Hat-tip to Erik Voeten at The Monkey Cage.