Activity Number:
|
253
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 13, 2002 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Section on Bayesian Stat. Sciences*
|
Abstract - #300667 |
Title:
|
Hierarchical Modeling of Association Structures of a Set of Sequential Categorical Data: A Bayesian Approach
|
Author(s):
|
Getachew Dagne*+ and Hendricks Brown and George Howe
|
Affiliation(s):
|
University of South Florida and University of South Florida and George Washington University
|
Address:
|
13201 Bruce B. Downs, MDC 56, Tampa, Florida, 33612, USA
|
Keywords:
|
Hierarchical modeling; ; multilevel models; ; Bayesian inference; ; contingency tables; ; association models ; observational data
|
Abstract:
|
Etiologic and intervention research in prevention often rely on microcoded data of dyadic or group interactions to provide evidence of change due to development or intervention. These data have traditionally been collapsed into smaller or modest-size two-way contingency tables, followed by residual analyses. Such approach not only limits our ability to fit models, but also can introduce spurious findings. Instead of treating each dyad's or group's two-way contingency table independently, or collapsing the tables into single aggregate table, it is more efficient to analyze association structures in all dyads simultaneously using hierarchical models. We present here a Bayesian hierarchical models to analyze several two-way sequential behavioral data with random effects that allow different levels of variations across several behaviors. To illustrate this approach, we present an analysis of couple's interaction data from a recent study from George Washington University investigating how couples cope when one partner has become jobless.
|