Sample Attrition in Teen Pregnancy Prevention Impact Evaluations Fact Sheet

A randomized controlled trial (RCT) is able to produce an unbiased estimate of an intervention's effect. However, when a small or non-representative subset of the initially assigned sample is used, the resulting estimate of effectiveness may be biased. This brief outlines how non-response (i.e., sample attrition) affects individual- and cluster-level RCTs, how the bias from attrition can be assessed, and strategies to limit sample attrition in teen pregnancy prevention evaluations.