|
Abstract:
|
Nonignorable nonresponse is a common problem in randomized clinical trials. Standard methods for handling nonignorable nonresponse typically require specification of the response mechanism. This paper introduces a reverse regression approach that does not require a model for the response mechanism but rather assumes that nonresponse is conditionally independent of treatment assignment given the outcome. The latter assumption is motivated by the observation that, when patients are effectively masked to the assigned treatment, their decision to drop out or not cannot depend on the assigned treatment directly. Under this assumption, one can estimate parameters in the reverse regression model, test for a favorable treatment effect, and in some cases estimate the outcome distributions. The methodology can be extended to longitudinal outcomes under natural conditions.
|
