Jim Helmreich - Marist College
Propensity Score Analysis: Attempting to Reduce Bias
You say you have an experiment to analyze? No problem - you did set it up to be double blind didn't you? What? Not quite? Not even single blind? Have you heard about this minor problem called the placebo effect? About conscious and unconscious bias? You have problems, but no, the study isn't really worthless the way your Intro Stat teacher said way back when . . . we'll see what we can do . . .
I will present a brief survey of techniques used to control bias in studies. Problems, strengths, and weaknesses of some of these techniques will be discussed. Logistic regression and Tree methods will be presented as methods of producing propensity classifications. Finally, the issue of sensitivity analysis of results will be touched upon.
The talk will be accessible to most undergraduates, and of interest to faculty members in the sciences and social sciences in addition to the usual gang. Knowledge of basic statistics will be very helpful.
[Program Information | Faculty | Mathematics Seminar | Mathematics Links]
[Math Department Home Page | Marist College]