The LATE framework is motivated by an analogy between IV and randomized trials. But some instrumental variables really come from randomized trials. IV methods allow us to capture the causal effect of treatment on the treated in spite of the nonrandom compliance decisions made by participants in experiments like the MDVE. In fact, the use of IV is usually necessary in such experiments. A naive analysis of the MDVE data based on treatment delivered is misleading.
Analysis of the MDVE based on treatment delivered is misleading because the cases in which police officers were supposed to coddle suspected batterers and actually did so are a nonrandom subset of all cases assigned to coddling. Comparisons of those who were and were not coddled are therefore contaminated by selection bias...Read More