I don’t know, maybe I had something rattling around in my brain about statistics in light of the then still pending Wal-Mart case when I wrote my last post, ostensibly about the use (and misuse) of statistics in baseball analysis, and less ostensibly about the use (and misuse) of statistics in litigation. I certainly wasn’t conscious of any concern on my part about the Supreme Court and Wal-Mart when I wrote that post, but the post now seems to have an air of precognition to it, at least in hindsight (one wonders whether it is possible to have precognition in hindsight, given the dictionary definitions of those words, but nonetheless). I particularly wasn’t thinking about Wal-Mart when I wrote that post because, like most casual observers of that case, I had long ago come to think it of as a case about class certification and class action requirements, and not about issues such as the proper role of statistics in litigation.
Regardless, one of the most interesting things right off the bat to me about the Court’s ruling in Wal-Mart is the extent to which its rejection of the statistical evidence at issue mimics the views of statistics in litigation that permeate my post on baseball statistics. As both the Court’s decision and that post suggest, there is no legitimate basis to rely on statistics when they do not, in fact, establish factually a specific point that could not otherwise be proven in a different manner of presenting evidence. This is another way of saying that statistical analysis does not belong in a court case, at any stage, if it does not illuminate – and do so accurately – a point that could not otherwise be shown.