Predicting the Health Care Decision from the Oral Arguments

by John Sides on March 29, 2012 · 13 comments

in Judicial

This is from political scientist Michael Evans, and was originally posted to a law and courts listserv.  I thank him for sending it along:


I was curious about the relative number of words directed at the two sides in yesterday’s oral argument and thought the results would be of interest here.

For those not familiar with the research on this (see below), it has shown that Justices tend to direct more questions and words at the side they eventually vote against. (Questions and words are highly correlated, but I prefer words because questions are harder to define.) The theory is that Justices generally do not play “devil’s advocate”—asking questions to help the side they support—but, rather, attempt to expose what they see as the weaknesses of the other side’s arguments.

This table shows the relative number of words uttered by each Justice to the two sides regarding the constitutionality of the individual mandate under the commerce clause. As is typical, Thomas did not ask any questions, but I think his vote is “somewhat” predictable. I excluded questions about the taxing power because, tellingly, no one asked Clement or Cavin a single question about the taxing power:

For what it’s worth, here is my take. Based on this, I think (even more than I did after reading and listening to the oral arguments) that the individual mandate is in serious trouble. This indicator suggests a 5-4 decision striking down the mandate. Of course, it goes without saying that the number of questions/words is not a perfect predictor (~86% accuracy in the published studies). And the most important caveat here is that Kennedy is the least predictable Justice (according to Epstein et al. 2009) and the discrepancy in his questions was lower than the other justices’. Another important caveat is that the aggregate results actually suggest a slightly higher probability that the mandate will be upheld (54% vs. 46% of words).

However, when we look at individual justices, which seems more appropriate here, the Court appears sharply polarized along familiar ideological lines and Kennedy uttered just over twice as many words to Verrilli as he did to Clement/Carvin, which is a greater discrepancy than is typical for him (according to Epstein et al.). For all other justices, moreover, the discrepancy was much starker than the average discrepancy reported by Epstein et al.; and the latter also found that the number of questions and words uttered by conservatives is a fairly good predictor of Kennedy’s (and Thomas’) vote. And note, finally, the whopping 39 percentage point difference between Kennedy and the closest liberal justice (Ginsburg) vs. the 13 percentage point difference between Kennedy and the closest conservative (Alito).

Again, this is just my take. I’d be interested to hear reasons to doubt this interpretation of the data, especially by those who have studied the predictability of votes based on oral arguments

Here is the list of studies:

  • Epstein, Lee, Landes, William M. and Posner, Richard A. 2009. “Inferring the Winning Party in the Supreme Court from the Pattern of Questioning at Oral Argument.” University of Chicago Law & Economics, Olin Working Paper No. 466. Available here or here.
  • Johnson, Timothy R., Black, Ryan C., Goldman, Jerry and Treul, Sarah. 2009 “Inquiring Minds Want to Know: Do Justices Tip Their Hands with Questions at Oral Argument in the U.S. Supreme Court?”9). Washington University Journal of Law & Policy, Vol. 29, 2009.  Available here.
  • Shullman, Sarah Levien. 2004. “The Illusion Of Devil’s Advocacy: How The Justices Of The Supreme Court Foreshadow Their Decisions During Oral Argument.” Journal of Appellate Practice and Process. Vol. 6: 2.  Available here.
  • See also Lindgren, Jim. 2005. “Does Asking More Questions Tip the Outcome of Supreme Court Cases?” Volokh Conspiracy.


Comments on this entry are closed.

Previous post:

Next post: