Tasks in Statistical Inference for Studying Variation in Medicine

Donald B. Rubin
DOI: https://doi.org/10.1097/00005650-199305001-00017
1993-01-01
Medical Care
Abstract:When studying variation in medicine, traditional hypothesis-testing procedures are too limited to obtain useful inferences except in special situations. More generally, full probability modelling is necessary. Even a relatively simple example can illustrate this point rather dramatically. The development and application of full-probability methods for medical problems comprise exciting areas for statistical and medical researchers, especially if working together.
What problem does this paper attempt to address?