The truncation-by-death problem: what to do in an experimental evaluation when the outcome is not always defined.

Although experiments are viewed as the gold standard for evaluation, some of their benefits may be lost when, as is common, outcomes are not defined for some sample members. In evaluations of marriage interventions, for example, a key outcome--relationship quality--is undefined when a couple splits up. This article shows how treatment-control differences in mean outcomes can be misleading when outcomes are not defined for everyone and discusses ways to identify the seriousness of the problem. Potential solutions to the problem are described, including approaches that rely on simple treatment-control differences-in-means as well as more complex modeling approaches.