I study the problem of choice between two treatments for a population of observationally identical individuals based on statistical evidence about average treatment effects that does not reveal the best treatment with certainty. I approach the problem from the perspective of statistical decision theory, derive treatment rules that minimize maximum...