The Late Pretest Problem in Randomized Control Trials of Education Interventions.

To improve precision of estimated treatment effects, education randomized control trials (RCTs) often use pretest–post-test experimental designs. For logistical reasons, however, pretest data often are collected after random assignment, so including them in the analysis could bias post-test impact estimates. Whether to collect and use late pretest data in RCTs involves a variance-bias trade-off. This article addresses this issue both theoretically and empirically for several commonly used impact estimators, using a loss function approach grounded in the causal inference literature.