No Two Experiments are Identical

We study choice between bets on the colors of two balls, where one ball is drawn from each of two urns. Though you are told the same about each urn, you are told very little, so that you are not given any reason to be certain that the compositions are identical. We identify choices that reveal an aversion to ambiguity about the relation between urns, thus identifying a source of uncertainty different from the usual Knightian distinction between risk and ambiguity. Behavior is studied in a controlled high-stakes laboratory experiment, and the ability to rationalize the experimental findings is examined.