Some Challenges to the "Israeli Judge Lunch" Study
A famous study in 2011 suggested that Israeli judges become less lenient in granting parole as lunch approaches and then more lenient immediately after.
Based on a quick look at this more recent study, it appears that, soon after the original study was published, some concerns were raised about whether the original authors properly controlled for whether potential parolees were represented by attorneys. Apparently, the original authors believe the general result remained even when the data are better controlled, though they may have overstated the effect size. The study I link to here raises additional concerns that might reduce effect size (suggesting, for example, that more difficult cases might be saved until after lunch in ways that might largely explain the purported results).
I haven't done my own careful examination of these issues, but it seems likely that the original study will propagate in the legal literature for a long time before any of these qualifications bubble up.