Quantitative Methods Colloquium Series: Will Shadish
4/8/2013, 2:00 pm - 3:30 pm
Speaker: Will Shadish, School of Social Sciences, Humanities and Arts, University of California, Merced
Title: "When Can Nonrandomized Experiments Approximate Results from Randomized Experiments: Some Findings and Hypotheses."
Abstract: Abstract: The randomized experiment is rightly regarded as a gold standard for making inferences about the effects of interventions on human participants in many settings. But it is often not feasible and sometimes not ethical. So much interdisciplinary work in the last 20 years has focused on improved causal inference from nonrandomized experiments. This includes empirical tests of how well these improvements actually do in practice when compared to randomized experiments. Pertinent topics include propensity score analysis, regression discontinuity designs, and short time series analysis. This research lays the groundwork for an empirical theory of the design of nonrandomized experiments.