Topic: Nonparametric Analysis of Time to
Event Data Under Sequential Sampling
Date: November 16, 2004
Speaker: Scott S. Emerson, M.D., Ph.D.
Abstract:
In the clinical trial
setting and some industrial settings, it is common to collect data under a sequential
sampling plan in order to minimize study costs-scientific, economic, and
ethical. Such sequential sampling causes special problems in the setting of
censored survival data, because the censoring distribution is generally not
constant across the various interim analyses of the data. I present approaches
for investigating the effect that sequential sampling can have on a study's
operating characteristics, and propose modifications to the standard survival
analyses which mitigate some of the issues which arise in this setting. We have found that such modifications can
have relatively minor effects on the efficiency of the analytic techniques,
while allowing great improvements in the ability of the sequential sampling
plan to reproduce the statistical behavior of a fixed sample test.
Biography:
Scott Emerson, M.D.,
Ph.D., Professor of Biostatistics at the
Time/Location:
Please arrive at 6:00
p.m. for dinner. Dinner will be served at 6:30 p.m. followed by the
presentation at 7:15 pm at the
Dinner:
Members
$15, Students $10.
If you
plan to come, please RSVP to Dan Fitzsimmons at dan.k.fitzsimmons@boeing.com or
Charles Phelps at cphelps@ctiseattle.com,
by Monday, November 15. Please specify if you will be having dinner with us, or will
be attending only the presentation.