Estimating a treatment effect under biased sampling.

摘要

Methods are presented for the estimation of a treatment effect based on before- and after-treatment values, where for ethical reasons all and only those patients are treated whose before-treatment values exceed a given constant.