Interval Censored Data: Some Recent Developments

Moulinath Banerjee
Department of Statistics
University of Michigan

Wednesday, October 25th 2006
3:30pm
Moos 2-690
Minneapolis Campus

Abstract:

Interval censoring arises extensively in epidemiology, especially in the study of HIV/AIDS and has therefore attracted considerable interest in the statistics and biostatistics communities over the past two decades. Interval censoring poses significantly harder challenges for inference, as compared to right censoring, since the exact failure time for an individual is never observed but is only known to lie in a random interval. Maximum likelihood procedures with interval censored data fail to produce estimates consistent at the usual square--root rate of the sample size. Rather, one encounters cube root asymptotics, with limit distributions of MLE's that are no longer Gaussian. This talk will review some of these curious features that are encountered with interval censored data, estimation strategies, recent inferential results, and explore natural connections to binary regression.

A social tea will be held at 3:00 P.M. in A434 Mayo. All are Welcome.
For more details contact 612-624-4655 or see http://www.biostat.umn.edu/seminar_academic.html