Arizona State University

Workshop:
Introduction to Latent Class Analysis with Mplus

Christian Geiser, PhD

Department of Psychology, Arizona State University

Dates and Location

Friday, April 22, 2011, 9:00 A.M.-5:00 P.M.

Saturday, April 23, 2011, 9:00 A.M.-5:00 P.M.

Location: Arizona State University, Main Campus at Tempe, AZ, Schwada Building (SCOB), Room 328 (http://www.asu.edu/uts/m_scob.htm)

Note: Registration closes on April 15, 2011!

Latent Class Analysis

Latent class analysis (LCA) is a more and more frequently used statistical method in different areas of empirical social science research. Important goals of LCA are to identify previously unknown subgroups of individuals (“latent classes”) that differ qualitatively (i.e., in kind) and/or quantitatively (i.e., in degree) with regard to certain attributes and to classify individuals according to their most likely latent class membership. For example, a marketing researcher may be interested in classifying individuals according to their kind of consumer behavior. A prevention scientist may ask whether individuals differ with regard to the kind or amount of risk behavior they show (e.g., alcohol or marijuana consumption). In cognitive psychology, a researcher might be interested in identifying subgroups of individuals that use distinct cognitive strategies to solve, for example, mental rotation tasks. Clinical psychologists may seek to classify patients into classes with different types of clinical (e.g., personality) disorders etc. LCA can also detect distinct latent groups and relate these groups to external variables.

LCA uses latent variables, thus allowing for (1) separation of measurement error from the “true” class membership of individuals and (2) assessment of the reliability of classification of individuals. Given that LCA focuses on the classification of individuals (rather than the classification of variables as, e.g., models of exploratory factor analysis), LCA is often seen as a “person-centered” statistical approach (as opposed to “variable-centered” approaches such as, e.g., factor analysis).

Overview of This Course

This course provides an introduction to LCA using the popular latent variable software Mplus (http://statmodel.com). Following a conceptual introduction and overview, issues of model estimation, interpretation, goodness-of-fit assessment, model selection, and presentation of results will be discussed. Special issues in LCA model fitting such as model identification, local solutions, boundary parameter estimates, and model fit assessment with sparse data will also be addressed. On the second day, more complex LCA models such as models with covariates and multigroup models will be discussed. Participants will be given time to analyze their own data and/or for doing supervised exercises in Mplus based on sample data sets that will be provided in the course. The number of participants is limited to 30.

Detailed Schedule

Time	Friday, April 22, 2011	Saturday, April 23, 2011
9:00-10:30	Introduction to LCA	LCA with covariates
10:30-11:00	Coffee Break	Coffee Break
11:00-12:30	Introduction to Mplus and application of LCA in Mplus	Multigroup LCA
12:30-1:30	Break	Break
1:30-3:00	Principles of parameter estimation and goodness-of-fit assessment	Overview of other LCA models and tips for the presentation of results
3:00-3:30	Coffee Break	Coffee Break
3:30-5:00	Special modeling issues: Identification, local solutions, starting values, boundary parameter estimates	Practical exercises in Mplus with own or provided data sets

About the Presenter

Christian Geiser obtained his PhD in psychology from Freie University Berlin, Germany, in 2008 and is currently an assistant professor of quantitative psychology at ASU. His main research interests are in structural equation modeling, multitrait-multimethod analysis, longitudinal data analysis, latent class modeling, and spatial abilities. In several of his research papers, Dr. Geiser has used latent class models to study substantive research questions, for example, regarding computer game preferences among adolescents and the use of different solution strategies in spatial cognition tasks. He has given numerous workshops on statistical data analysis and has recently published a German introductory textbook on data analysis with Mplus. More information can be found at his website: https://webapp4.asu.edu/directory/person/1546935

Registration and Lodging

Please register online at:

https://ggrp.asu.edu/opinio/s?s=1020

The workshop fee is $495 including electronic copies of the workshop materials. Payment is online via credit card (Master and Visa Card are accepted). A payment confirmation will be sent via email once the credit card payment has been processed. A withdrawal fee of $50 will be charged for all refunds until March 15. After March 15, no refunds will be granted. The maximum number of participants is 30. Complementary coffee and snacks will be provided during breaks. Participants must make their own arrangements for lodging and meals. Reasonably priced guest rooms within walking distance or an easy shuttle ride to campus are available, for example, at:

http://www.bestwesterntempe.com/BestWesternTempe.asp?id=29

http://www.qualityinn.com/hotel-tempe-arizona-AZ213

http://twinpalmshotel.com/

Materials

Participants who have registered for this workshop will receive the course materials (PowerPoint slides, sample data sets, input and output files) approximately one week prior to the workshop via email. Although computers will be available in the lecture room, participants are encouraged to bring their own laptop on which they have installed the demo-version of Mplus, which can be downloaded for free at http://statmodel.com/demo.shtml

Further Information