|
Activity Number:
|
188
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2008 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #301990 |
|
Title:
|
Semiparametric-Efficient Estimation for Multistage Case-Control Studies
|
|
Author(s):
|
Alastair J. Scott*+ and Alan J. Lee and Chris J. Wild
|
|
Companies:
|
University of Auckland and University of Auckland and University of Auckland
|
|
Address:
|
38 Princes StreetDepartment of Statistics, Auckland, International, 1010, New Zealand
|
|
Keywords:
|
multi-stage ; case-control study ; semiparametric efficiency
|
|
Abstract:
|
In a simple stratified case-control study, a prospective cohort is stratified according to some variables known for the whole cohort. Separate samples of cases and controls are then drawn from each stratum and values of covariates are obtained. In a two-stage study, some of the more expensive or difficult covariates are not measured on all the sampled units, but only on a subsample drawn from them. This process can be continued indefinitely. We derive a set of estimating equations for the efficient semiparametric maximum likelihood estimator with an arbitrary number of stages. These take the form of penalized "pseudo-likelihood" equations, with a term corresponding to an ordinary prospective likelihood plus an extra penalty term for each additional stage of sampling. We illustrate with a three-stage example.
|
