|
Activity Number:
|
227
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #308085 |
|
Title:
|
Statistics and Lies: Correcting Questionnaire Ordering Effect via Multiple Imputation
|
|
Author(s):
|
Xiao-li Meng*+ and Jingchen Liu and Chih-nan Chen and Margarita Alegria
|
|
Companies:
|
Harvard University and Harvard University and Boston University and Cambridge Health Alliance
|
|
Address:
|
1 Oxford Street 7th Floor, Cambridge, MA, 02138,
|
|
Keywords:
|
questionnaire ordering effect ; survey design ; Bayesian modelling ; high-order interactions ; psychiatric epidemiology ; MCMC
|
|
Abstract:
|
National Latino and Asian American Study (NLAAS) is a complex survey of psychiatric epidemiology, with multiple embedded experiments with alternative question forms. One objective of multiple imputation is to create analytic datasets corrected for response biases due to defects of the survey instrument, such as increasing rates of negative responses over the course of the interview induced by the respondents' learning to use skip patterns to reduce interview time. The imputation modeling task is particularly challenging because of the complexity of the questionnaire, the small sample sizes for subgroups of interests, and the existence of high-order interactions among variables. This talk presents our Bayesian modeling and computation effort in addressing these problems.
|
