A Distributional Analysis of Treatment Effects on Subpopulations of a Socioeconomic Experiment
When analyzing treatment effects, the average treatment value is frequently compared to that of the control group. This approach, naturally, is not particularly informative about specific regions of the treatment and control distributions. For this reason and having in view a specific application, in the present paper we consider tests that provide us with more detailed analysis of treatments and their effectiveness. The tests are based on comparing the treatment and control distributions (e.g., whether they are equal, one dominates another, or intersect) over their entire or partial domains of definition. The test of intersection of distributions is introduced in the paper with the scope of pinpointing the region of the tested distribution that is subject to an adverse treatment effect. We illustrate the tests on a simulation study which is based on a matched data and apply them to analyze the Pennsylvania Bonus Experiment.
|Keywords||Stochastic dominance, effective treatment.|
|JEL||Hypothesis Testing (jel C12), Personal Income, Wealth, and Their Distributions (jel D31), Equity, Justice, Inequality, and Other Normative Criteria and Measurement (jel D63)|
|Publisher||Department of Economics|
|Series||Carleton Economic Papers (CEP)|
Wang, Liqun, Voia, M.-C, & Zitikis, Ricardas. (2009). A Distributional Analysis of Treatment Effects on Subpopulations of a Socioeconomic Experiment (No. CEP 09-02). Carleton Economic Papers (CEP). Department of Economics.