A natural experiment in health outcomes research is used to frame a discussion of statistical methods for causal inference in observational studies. A narrative description of the study’s design, analysis and reception is interrupted twice, first to describe algorithmic developments in optimal matching in design, second to describe analyses that inform discussions of unobserved biases due to the absence of randomized treatment assignment. Specific topics include: (i) fine balance constraints imposed on minimum distance matched samples, (ii) sensitivity analyses when treatment effects are heterogeneous, and (iii) design sensitivity as a tool to evaluate study designs and analytical techniques.
Online Handouts:
http://www-stat.wharton.upenn.edu/~rosenbap/ovarianhandoutPrint.pdf
http://www-stat.wharton.upenn.edu/~rosenbap/ovarianhandPhone.pdf
|