Along with the prayer ruling today from the Supreme Court, the New York times reports on a study from USC on the justices' biases in voting to defend free speech rights.
Surprising no one, the headline finding is that the more liberal or conservative a justice is, the more likely they are to vote in favor of a liberal or conservative speaker, respectively. Here's the key graphic.
(By the way, side note on the roles of journalists and scientists here. I think this graphic was constructed from the scientists' raw data by the Times; it's not found in the original paper, and it really conveys some key points very clearly. If so, good communication on the part of the Grey Lady there.)
So far, so human, although we can and should demand more of our highest justices. People favor their side even when they think they're being fair. But there are two things I noticed that I think are much more important than the evidence of a normal human bias among all Supreme Court justices.
First and by far the most important, while a difference is visible for liberal justices -- the justices fit a model with a nonzero slope, with Justice Stevens at the most liberal end -- the study calculates that the difference is not statistically significant for any liberal justice but him. Take a look at that graph again. Does being a conservative speaker penalize you as you move from the conservative to the liberal justices? A little. But compare that to how much being a liberal speaker penalizes you as you move from the liberal to the conservative justices.
The liberals are, for the most part, not statistically different in their support for conservative and liberal speech. They can fairly claim to be impartial when evaluating speakers. The conservative justices are statistically different in their support for conservative and liberal speakers. This is an objective measure that calls their commitment to impartiality into serious question, and -- in the rare cases that Supreme Court Justices open themselves to questioning by the public -- an accounting should be demanded.
(A note: the Times article opens with Scalia's famous vote to strike down a law banning flag-burning. One wonders if in the years since, Scalia is suffering from a version of the Snackwell effect: having inoculated himself against charges of favoritism in his own mind, he is free to play favorites. Recall any pleasant surprise you may have felt when Chief Justice Roberts voted to uphold the ACA.)
The second thing to notice is that the study looked at votes from 1953 to 2011. Discarding justices with fewer than 100 votes (insufficiently strong basis to form a sound ideological measure), who has the most extreme bias in the last half-century plus? Justices Scalia and Thomas. I've downloaded the study's original data and I'm going to see if I can determine whether the margin of bias has become more extreme over time. Technically my working hypothesis is the null, but the good Reverend Bayes is whispering for me to put it all on "yes."
Sat May 10, 2014 at 11:11 PM PT: I have completed the time series calculations. The scientists' raw data was already helpfully sorted by the time the justices was appointed, so that made a handy guide. I might have sorted by midpoint of time on the court, but I doubt that would have affected anything.
At any rate, the upshot: turns out no, bias isn't getting worse -- at least, I can't say with confidence that it is. There is basically a band of justices for whom the differences in support for liberals and conservatives (whatever the justice's base ideology) is 10 to 20 percent or so. If you ignore the justices with fewer than 50 free-speech votes, the only justices with a difference above 30% are Douglas, Warren, Scalia, and Thomas.
There's kind of a trend-ish in the lower band of justices, of bias increasing over the years, but we're getting perilously close to the Texas sharpshooter fallacy by this point.