Activity Number:
|
666
- Prediction and Calibration
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 3, 2017 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #324953
|
View Presentation
|
Title:
|
Model Assessment on Longitudinal Binary Data with Time-Dependent Covariates
|
Author(s):
|
Tan Li* and Katrina INTERNATIONAL Epnere
|
Companies:
|
Florida International University and Florida International University
|
Keywords:
|
longitudinal binary data ;
time-dependence ;
multilevel logistic regression ;
Markovian model ;
model assessment ;
HIV risk
|
Abstract:
|
Longitudinal binary data with time-dependent covariates is usually seen in social science and epidemiology. One of commonly used solution is using multilevel logistic regression while decomposing the time-varying covariates to between-subject effect and within-subject effect. Recently a Markovian model of first order (MARK1 model) is published which can avoid the decomposition of the time-dependent covariates. The purpose of this paper is to assess the two models' performance based on parameter estimation and errors of hypothesis tests using a simulation study under various situations with different values of parameters, as well as a real dataset from an NIH funded social work study on HIV risk behavior.
|
Authors who are presenting talks have a * after their name.