|
Title
|
Room
|
|
* Correlated Errors, Biased Instruments and Measurement Error Correction in Nutritional Epidemiology
|
H-Clayton
|
|
Date / Time
|
Sponsor
|
Type
|
|
08/06/2001
10:30 AM
-
12:20 PM
|
Section on Statistics in Epidemiology*, ENAR
|
Invited
|
|
Organizer:
|
Donna Spiegelman, Harvard School of Public Health
|
|
Chair:
|
Xiao-Hua (Andrew) Zhou, Indiana University
|
|
Discussant:
|
11:50 AM - Rudolf Kaaks, International Agency for Research and Cancer
|
|
Floor Discussion
|
12:15 PM
|
Description
Covariate measurement error can have a profound impact on estimation and inference in epidemiology, where such error is unavoidably large in many cases. Important scientific problems in nutritional epidemiology has motivated statistical research on valid estimation and inference in such settings, using validation studies in which information about the measurement error model can be obtained. Earlier methods proposed rely on assumptions about the measurement error model that may be unrealistic. In this session, three statisticians actively working in this area present new research when the original error-prone measurement is validated by one or more additional instruments, some of which may be biased, have correlated systematic within-person errors, and correlated random errors. A discussant with expertise in the subject-matter area will comment on the utility of this work for practical applications.
|
|
