Center for Demography of Health and Aging

This series of workshops were devoted to causal inference from data with non-random assignment. In particular, there was discussion of the so-called propensity scoring matching and its advantages and shortcomings. The first workshop consisted of discussion of several empirical problems by faculty members in the departments of Sociology and Political Sciences. The second workshop focused on specialized literature on propensity scoring as well as a review of available software. Michael Sobel from the Columbia University gave a presentation at the third workshop while Petra Todd from the University of Pennsylvania was the speakers at the fourth workshop. The main objective of the last two workshops was to examine in detail the assumptions on which the procedure is based and the class of causal inferences that are possible and those that are not possible.

Statistical Methodology Workshop Series II
Fall of 2006

The main theme of the second workshop Series was sampling of hidden or hard-to-find populations. Giovanna Merli from the University of Wisconsin-Madison gave a presentation in the first workshop about a problem related to sampling from sexual networks in china. The second workshop consisted of discussion of literature on sampling from hidden populations and, in particular, examination of the method proposed by Douglas Heckathorn from Cornell University. Jesus Ramirez-Valles from the University of Illinois at Chicago, who has applied the Heckathorn's procedure to study IDU and HIV in the Chicago area discussed problems of implementation and shortcomings. At the third workshop, Steve Thompson from Simon Fraser University gave a critique of the Heckathorn procedure and presented an alternative Bayesian strategy proposed by Anderson and his colleagues Keith Sabin at the Centers for Disease Control and Prevention. The final workshop consisted of potential solutions for Merli’s original problem.

Statistical Methodology Workshop Series I
Spring of 2006

The first workshop was devoted to the issue of multiple imputation. Dick Campbell from the University of Illinois at Chicago presented a problem that could, in theory at least, be resolved via imputation. In the second seminar there was discussion of available multiple imputation techniques and their rationale and justification. The third and fourth workshops were designed around lectures offered by a recognized expert, Trivellore Raghunathan from the University of Michigan in multiple imputation. In the fifth workshop Dick Campbell reformulated and resolved his original problem.