What the judge ate for elevenses

Andy Hilfinger pointed me to an interesting recent paper that looks at the effect of a judge’s daily routine on the decisions that he or she makes.  The judges in question are 8 Jewish-Israeli judges who preside over the parole board hearings for 4 major prisons in Israel.  Each judge hears, on average, 20 cases per day; most of the cases are requests for parole to be granted, while others request changes in the terms of their sentence.  Roughly 2/3 of the cases were rejected.  The authors wanted to know whether extraneous factors entered into the decision to allow or deny the request, and particularly whether there was any truth in the old saying that “justice is what the judge had for breakfast”.

They don’t actually look at the composition of the judges’ breakfasts, however.  What they look at instead is the rate of positive decision-making relative to the breaks the judges take to eat.  The regular routine in these courts, apparently, is that the judge takes both a mid-morning snack break and a lunch break.  The judge chooses when the break happens and how long it lasts.  The judge cannot decide when to take the break given the order of the upcoming cases: s/he doesn’t know the order of the cases in advance.

Positive judgements vs. time -- my sketch from data in Danziger et al.

The data are rather startling.  I’ve sketched the main result for you: the real figure isn’t quite as smooth, but the trend is just as clear.  At the beginning of each session, the judge makes positive decisions about 65% of the time.  As the session continues, the rate of positive decisions drops precipitously.  Then the judge takes a break, and the positive decision rate goes right back up to 65%.  The authors carefully checked whether this pattern could be explained by any other factors — the severity of the crime the prisoner committed, the length of sentence already served, prisoner demographics, etc. — and found that none of them explained this pattern even partially.  Of course some of these factors did affect the judges’ decisions — if a prisoner was a repeat offender, the judge was less likely to allow parole, and if the prisoner could show a plan for rehabilitation, they were more likely to get a positive decision — but none of this explains the remarkable change in positive decision rate between the beginning and the end of a session.  The authors also checked for signs of a quota system, for example an increased tendency of a judge to start saying no after a given number of positive judgements.  If anything, the trend goes the other way: a judge who’s made a large number of positive rulings may be slightly more likely to rule positively in the next case that comes up.  There is also good reason to believe that neither the judge nor anyone else knows the order in which the cases are going to come up: the relative position of the cases is determined by the time of arrival of the prisoners’ attorneys, who are sequestered in a separate room after their arrival and don’t know whether the judge has taken a break recently, and the judge doesn’t see the case paperwork until the case is actually brought forward.

What’s going on?  Making repeated decisions is exhausting: the authors point to previous research that shows that people who are asked to make decisions repeatedly tend to become more likely to make the easy decision.  For a judge, the easy decision is to leave things alone, i.e. say no.  For a car buyer (as shown in another study), the easy decision may be to take the options offered as standard by the manufacturer, after your mental resources have been exhausted by the need to make decisions about many other details.   A break can help — and food could too.  This study doesn’t distinguish between the effects of a break and the effects of increased blood sugar.  The idea that it’s harder for a judge to make the decision to approve parole is supported by the length of time it takes the judge to deliberate in these cases: positive decisions took about 1.5x longer than negative decisions.

Andy points out that one could imagine an effect like this operating in many other situations.  He’s particularly interested in whether NIH study sections show dramatic swings in amiability after a lunch break (though I am not sure how he plans to manipulate the time at which specific grants are considered).  I wouldn’t be at all surprised if grant reviewers reach a point where they just want to make the easy decision and get it over with, although the fact that they do review the grants ahead of time may complicate things.  But I wonder whether the easy decision is always to reject the grant — perhaps the easy decision is instead to go along with the reviewer who makes the most passionate case.  Anyone who has any insight into these matters, please feel free to comment.

Danziger S, Levav J, & Avnaim-Pesso L (2011). Extraneous factors in judicial decisions. Proceedings of the National Academy of Sciences of the United States of America, 108 (17), 6889-92 PMID: 21482790