This paper introduces an empirical strategy to estimate dynamic treatment effects in randomized trials that provide treatment in multiple stages and in which various noncompliance problems arise such as attrition and selective transitions between treatment and control groups. Our approach is...