BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:A superpopulation treatment to case-control data analysis - Yanyua
 n Ma (Pennsylvania State University)
DTSTART:20180619T100000Z
DTEND:20180619T110000Z
UID:TALK107302@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:We study the regression relationship among covariates in case-
 control data\, an area known as the secondary analysis of case-control stu
 dies. The context is such that only the form of the regression mean is spe
 cified\, so that we allow an arbitrary regression error distribution\, whi
 ch can depend on the covariates and thus can be heteroscedastic. Under mil
 d regularity conditions we establish the theoretical identifiability of su
 ch models. Previous work in this context has either (a) specified a fully 
 parametric distribution for the regression errors\, (b) specified a homosc
 edastic distribution for the regression errors\, (c) has specified the rat
 e of disease in the population (we refer this as true population)\, or (d)
  has made a rare disease approximation. We construct a class of semiparame
 tric estimation procedures that rely on none of these. The estimators diff
 er from the usual semiparametric ones in that they draw conclusions about 
 the true population\, while technically operating in a hypothetic superpop
 ulation. We also construct estimators with a unique feature\, in that they
  are robust against the misspecification of the regression error distribut
 ion in terms of variance structure\, while all other nonparametric effects
  are estimated despite of the biased samples. We establish the asymptotic 
 properties of the estimators and&nbsp\;illustrate their finite sample perf
 ormance through simulation studies.  <br><br><br><br><br>
LOCATION:Seminar Room 2\, Newton Institute
END:VEVENT
END:VCALENDAR