Selection
If subjects in the experimental group are substantially different from control subjects, the researcher can’t assume it’s because of the experimental condition-- Maybe these subjects were different even before
Comparing pre-test observations for the experimental & control conditions can rule this out