University of Michigan
Ben Hansen's research interests center around causal inference in comparative studies, particularly observational studies in the social sciences. He contributes to a line of research that, following R.A. Fisher, sees controlled, randomized experiments as the embodiment of an important methodological ideal, while holding to W.G. Cochran's conviction that empirical knowledge can be had even in the absence of experimental control: the keys being to accumulate evidence from diverse sources, and to make one's observational studies experiment-like in relevant ways.
His current work centers around providing methods for statistical adjustment that enable researchers to mount focused, specific analogies of their observational studies to randomized experiments. Such methods include matching, propensity scores, and analogues and extensions of both of these, as well as randomization-based methods of inference. He has papers in print and in development on these topics, and also software, in the form of add-on packages for the open-source statistical environment R. Relatedly, he has interests in confidence procedures that optimize for expected length (while maintaining a given confidence coefficient), and in disclosure risk.