Seminar Notice<DIV>
	<DIV></DIV>
	<BR>
<DIV>&nbsp;Statistical Laboratory</DIV>
<DIV>&nbsp;Iowa State University</DIV><BR>
<DIV>
		DATE AND TIME:&nbsp; <b>Monday, January 31, 2005, 4:10 p.m.</b></DIV><BR>
<DIV>
		PLACE: <b><font color="#990000">319 Snedecor</font></b></DIV><BR>
<DIV>
		SPEAKER: <b><font color="#000099">Hwan Chung</font></b>, Department of Statistics, The Pennsylvania State University, University Park, PA</DIV><BR>
<DIV>TITLE:&nbsp; Latent-Class Logistic Regression</DIV><BR><BR>
<DIV>ABSTRACT</DIV><BR><BR>
<DIV>In the traditional latent-class (LC) model, multiple categorical responses 
are assumed to be independent within categories of a latent classification 
variable. This model has recently been extended to incorporate categorical and 
continuous covariates as predictors of class membership through multinomial 
logistic regression.&nbsp; Routines for maximum-likelihood (ML) estimation are 
currently available in Mplus (Muthen &amp; Muthen, 1998) and Latent Gold 
(Vermunt &amp; Magidson, 2000).&nbsp; In many examples, however, the likelihood 
function exhibits unusual features, causing ML estimates and their associated 
standard errors to behave erratically.&nbsp; In this talk, we explore a variety 
of theoretical and practical issues surrounding the use of the LC logistic 
regression model, including Bayesian alternatives to ML estimation.&nbsp; We 
illustrate these issues with an example from adolescent substance use: tracking 
changes in marijuana use and attitudes among American high school seniors from 
1977 to 2001, using data from Monitoring the Future.</DIV><BR>
<DIV>COFFEE:&nbsp; 3:45 p.m., 104 Snedecor Hall</DIV><BR>
	<DIV>
		</DIV>
</DIV>