|
Activity Number:
|
115
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, August 4, 2003 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Stat. Sciences
|
| Abstract - #300472 |
|
Title:
|
Hierarchical Models for Learning Effects in Response Time
|
|
Author(s):
|
Jun Lu*+ and Dongchu Sun and Paul L. Speckman and Jeff N. Rounder
|
|
Companies:
|
University of Missouri, Columbia and University of Missouri, Columbia and University of Missouri, Columbia and University of Missouri, Columbia
|
|
Address:
|
222 Maths Science Bldg., Columbia, MO, 65211,
|
|
Keywords:
|
Bayes hierarchical model ; Weibull distribution ; response time
|
|
Abstract:
|
Time to complete a task often serves as a dependent variable in experimental psychology. Repeated measures are often obtained, and psychologists are interested in comparing response time distributions across subjects and experimental conditions. These distributions are typically skewed, and interest extends beyond comparing means. Bayesian hierarchical models are used to analyze three-parameter Weibull response time distributions. We focus on modeling the effect of learning, that is, the change in distribution as a function of trial. A lognormal prior on the scale parameter is used with mean effects for subject and learning. Four types of learning effect models are introduced and compared: an independent normal prior, a linear model in trial number, a first-order autoregression with unknown autocorrelation coefficient, and a second order Gaussian intrinsic autoregressive process. Bayesian computation is done via MCMC. Results suggest that the second order intrinsic AR process provides the best model.
|
- 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 2003 program |
