|
Activity Number:
|
392
|
|
Type:
|
Invited
|
|
Date/Time:
|
Thursday, August 15, 2002 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
ENAR
|
| Abstract - #300123 |
|
Title:
|
Informative Missingness with Application to Statistical Genetics
|
|
Author(s):
|
Glen Satten*+ and Andrew Allen and Paul Rathouz
|
|
Affiliation(s):
|
Centers for Disease Control and Prevention and Duke University and University of Chicago
|
|
Address:
|
Mailstio f-24, Buford Highway, Atlanta, Georgia, 30341, USA
|
|
Keywords:
|
TDT ; parental genotypes
|
|
Abstract:
|
Data are informatively missing when the probability of data missing is a function of the (unmeasured) value. Informatively missing data are difficult because the fraction of the sample with complete data are not representative of the observations with missing data. We consider here a general approach to informatively missing data motivated by the problem of missing parental data in a TDT (transmission disequilibrium test). Broadly, a TDT compares the alleles transmitted by parents to an affected child with the parental alleles that were not transmitted. If parental genotypes are missing, some TDTs attempt to reconstruct this data using data from intact families assuming noninformative missingness. However, these approaches will fail if the parental genotype itself determines the frequency that the parental data is available. We show it is possible to account for the effect of informatively missing parental genotype in a TDT under the assumption that given the parental genotype, offspring and spouse genotypes do not further affect missingness. We use simulations to compare our approach to other TDTs that assume noninformative missingness.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |
