Can the evaluator be sure that those who dropped out of the reatment group were equivalent to those who dropped out of the control? Extensive
pretesting can be used to help adjust statistically for differences across groups in those who leave the study to maintain the equivalence of those who remain in the study.