Inference of Sample Complier Average Causal Effects in Completely Randomized Experiments

Abstract

In randomized experiments with non-compliance scholars have argued that the complier average causal effect (CACE) ought to be the main causal estimand. The literature on inference of the complier average treatment effect (CACE) has focused on inference about the population CACE. However, in general individuals in the experiments are volunteers. This means that there is a risk that individuals partaking in a given experiment differ in important ways from a population of interest. It is thus of interest to focus on the sample at hand and have easy to use and correct procedures for inference about the sample CACE. We consider a more general setting than in the previous literature and construct a confidence interval based on the Wald estimator in the form of a finite closed interval that is familiar to practitioners. Furthermore, with the access of pre-treatment covariates, we propose a new regression adjustment estimator and associated methods for constructing confidence intervals. Finite sample performance of the methods is examined through a Monte Carlo simulation and the methods are used in an application to a job training experiment.