A large part of the recent literature on program evaluation has focused on estimation of the average effect of the treatment under assumptions of unconfoundedness or ignorability following the seminal work by Rubin (1974) and Rosenbaum and Rubin (1983). In many cases however, researchers are interested in the effects of programs beyond estimates of the overall average or the average for the subpopulation of treated individuals. It may be of substantive interest to investigate whether there is any subpopulation for which a program or treatment has a nonzero average effect, or whether there is heterogeneity in the effect of the treatment. The hypothesis that the average effect of the treatment is zero for all subpopulations is also important for researchers interested in assessing assumptions concerning the selection mechanism. In this paper we develop two nonparametric tests. The first test is for the null hypothesis that the treatment has a zero average effect for any subpopulation defined by covariates. The second test is for the null hypothesis that the average effect conditional on the covariates is identical for all subpopulations, in other words, that there is no heterogeneity in average treatment effects by covariates. Sacrificing some generality by focusing on these two specific null hypotheses we derive tests that are straightforward to implement

Randomized studies are designed to estimate the average treatment effect (ATE) of an intervention. Individuals may derive quantitatively, or even qualitatively, different effects from the ATE, which is called the heterogeneity of treatment effect. It is important to detect the existence of heterogeneity in the treatment responses, and identify the different sub-populations. Two corresponding statistical methods will be discussed in this talk: a hypothesis testing procedure and a mixture-model based approach. The hypothesis testing procedure was constructed to test for the existence of a treatment effect in sub-populations. The test is nonparametric, and can be applied to all types of outcome measures. A key innovation of this test is to build stochastic search into the test statistic to detect signals that may not be linearly related to the multiple covariates. Simulations were performed to compare the proposed test with existing methods. Power calculation strategy was also developed for the proposed test at the design stage. The mixture-model based approach was developed to identify and study the sub-populations with different treatment effects from an intervention. A latent binary variable was used to indicate whether or not a subject was in a sub-population with average treatment benefit. The mixture-model combines a logistic formulation of the latent variable with proportional hazards models. The parameters in the mixture-model were estimated by the EM algorithm. The properties of the estimators were then studied by the simulations. Finally, all above methods were applied to a real randomized study in a low ejection fraction population that compared the Implantable Cardioverter Defibrillator (ICD) with conventional medical therapy in reducing total mortality.

Originally published in 1986, this valuable reference provides a detailed treatment of limit theorems and inequalities for empirical processes of real-valued random variables; applications of the theory to censored data, spacings, rank statistics, quantiles, and many functionals of empirical processes, including a treatment of bootstrap methods; and a summary of inequalities that are useful for proving limit theorems. At the end of the Errata section, the authors have supplied references to solutions for 11 of the 19 Open Questions provided in the book's original edition. Audience: researchers in statistical theory, probability theory, biostatistics, econometrics, and computer science.