Mister P: Challenges in Generalizing from Sample to Population

Andrew Gelman, Department of Statistics and Department of Political Science, Columbia University

With internet surveys, nonrepresentativeness and nonresponse are bigger concerns than ever. The natural approach is to adjust for more information, demographic and otherwise, to align the sample with the population. We demonstrate the effectiveness of multilevel regression and poststratification (MRP, or Mister P) in internet surveys, and we also discuss some of the challenges and risks of this and other approaches.

Details here. It looks like I’ll be speaking at 1:30 pm, and the conference will be at Johns Hopkins School of Public Health, 725 N Wolfe Street, Wood Basic Sciences Auditorium, Baltimore. Enjoy!