|
Activity Number:
|
299
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2004 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #301603 |
|
Title:
|
Bayesian Hierarchical Models for Process Dissociation Procedure in Memory Research
|
|
Author(s):
|
Jun Lu*+ and Dongchu Sun and Paul L. Speckman and Jeff Rouder
|
|
Companies:
|
University of Missouri, Columbia and University of Missouri, Columbia and University of Missouri, Columbia and University of Missouri, Columbia
|
|
Address:
|
146 Middlebush Hall , Columbia, MO, 65211-4100,
|
|
Keywords:
|
Bayesian hierarchical model ; probit model ; linear mixed model ; process dissociation procedure
|
|
Abstract:
|
Memory serves multiple functions in different ways. Many multiple memory theories consist of two primary components: (1) the "recollection," which refers to a conscious recall of an event in the past, and (2) the "automatic activation," which reflects the unconscious activation of previously encountered events. These two components are typically correlated. Such correlation makes it difficult to estimate the parameters of the automatic and recollective components. Thus, we propose a Bayesian hierarchical model to get stable estimates under a Process Dissociation Procedure (PDP) experiment. The model features two probit links. Each probit link has a form of linear mixed model with additive components that reflects the participant effects and the item effects. We assume correlations across the effects between two linear models. Two sets of prior distributions are proposed to model the correlation: the shared-component prior and the Wishart prior. Bayesian computation can be done via MCMC. Simulation studies are conducted to evaluate the performance of the Bayesian approach. We also provide the derivation of Bayes Factors and an alternative memory experiment design.
|
