|
Activity Number:
|
58
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 3, 2008 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #302123 |
|
Title:
|
Multilevel Modeling Technique and Bootstrap Variance Estimation in Longitudinal and Cross-Sectional Complex Survey Data Sets
|
|
Author(s):
|
Alomgir Hossain*+ and Punam Pahwa and Bruce Reeder and Bonnie Janzen
|
|
Companies:
|
University of Saskatchewan and University of Saskatchewan and University of Saskachewan and University of Saskatchewan
|
|
Address:
|
Health Science Building, Community Health and Epidemiology, Saskatoon, SK, S7N5E5, Canada
|
|
Keywords:
|
Multilevel technique ; bootstrap ; longitudinal ; quasi-likelihood ; pseudo-likelihood
|
|
Abstract:
|
Two most commonly used techniques to analyze longitudinal and cross-sectional complex survey data are: (i) standard regression techniques based on the quasi-likelihood for parameter estimation and the bootstrapping approach for variance estimation and (ii) multilevel technique based on pseudo maximum likelihood. Our study objectives are to explore the standard regression parameter estimation based on quasi-likelihood techniques, bootstrapping approach for variance estimation by using the longitudinal data from Canadian National Population Health Survey (NPHS) and weighted and un-weighted parameter estimation by using the cross-sectional Canadian Heart Health Datasets (CHHD), collected in1986--92, after controlling PSU level. The longitudinal NPHS began in 1994/95 with a sample size of 17,276 and data being collected every two years.
|
