I very much enjoyed reading Craig Lerner's new paper, Reasonable Suspicion and Mere Hunches, linked to earlier today by co-blogger Todd, but I think it overlooks the critical reason why courts have required objective, articulable facts to conduct a Terry stop. As I read Craig's paper, he argues that the requirement of ex ante articulable facts should be dropped, and replaced with an ex post examination of whether the officer's conduct was reasonable based on the entirety of the circumstances. The entirety of the circumstances should factor in whether the officer's behavior was effective in identifying criminal conduct: Specifically, if the police are acting on hunches but catching lots of bad guys, the courts should let them act on hunches.

  The problem with this argument, I think, is that courts regulate Terry stops almost exclusively through the exclusionary rule. This results in a selection bias problem: Of all the Terry stops that the police conduct, the ones that appear in appellate opinions are the the subset in which the stop led to an arrest and conviction. These are the cases in which the suspect was guilty, and the officer was right. Civil cases alleging improper Terry stops are theoretically possible, but almost never brought: qualified immunity makes such cases very hard to win, and damages usually are very low. (Lerner alludes to the possibility of a selection bias on page 72, but only in the course of cautioning the reader as to why we can't conclusively say that officer hunches are almost always right based on a reading of published cases.)

  My sense is that the current law of Terry stops can be explained in large part as a response to this selection bias. Judges require objective evidence rather than "mere hunches" because they realize that they only see the cases in which "mere hunches" proved correct. It's hard to craft a rule on that basis: If you as a judge or Justice only see the subset of successful stops, then you'll have no idea how to apply the law to permit stops that have a high likelihood of success but prohibit stops that have a low likelihood of success. The judges don't get to see any empirical evidence of the unsuccessful stops, as those cases don't normally lead to litigation. This context makes it quite hard for judges to perform the kind of ex post reasonableness analysis that Lerner seeks, and I think may explain why current law sticks to the requirement of objective and articulable facts ex ante. It's not an ideal solution, but it's probably better than the realistic alternatives.

  That's my tentative reaction, at least. I just skimmed the piece, so it's quite possible that I'm missing something important. And as Todd says, it's a very thought-provoking paper.

  UDPATE: I wonder what Justice Lammers thinks of Lerner's proposal.