James Dreier, Brown University
January 15th, 2004, 8:00 to 9:30 pm
In legal reasoning, and especially when judges and juries reason their way to verdicts, we long for absolutely certain, deductive proofs. But the world does not, by and large, cooperate with our ideals, and legal reasoning like almost all practical reasoning involves and often turns on claims and inferences about which we can have only partial, uncertain confidence: probable reasoning, that is to say. Sometimes reasons for thinking a defendant guilty are merely probable in a special way: they are statistical reasons, and the reasoning that judges and juries engage is statistical reasoning. Intuitively there seems to be something wrong with “merely statistical reasoning” when it results in a verdict, especially when it results in a verdict against a defendant. I survey some examples and some accounts of what is wrong with “mere statistical reasoning” in the law. Analysis of the examples and accounts should, I argue, leave us puzzled. I conclude with a radical suggestion.