BEAR Short Courses

New Series (starting Fall 2008):

BEAR Short Courses

Sun-Joo Cho, UC Berkeley

Bayesian Analysis of Explanatory Item Response Theory Models using WinBUGS

Session I: Introduction to Bayesian inference and computation and WinBUGS
October 28th, 2008/2-4 pm/Tolman Hall, Room 2515

The first part of the workshop will introduce Bayesian inference and computation. Topics include item response theory (IRT) models as a nonlinear mixed effect model and their likelihood, prior distribution, posterior distribution, and posterior simulation. Based on these topics, the use of WinBUGS will be introduced to build Markov chain algorithms using the Gibbs sampler. Topics include the basic programming routine (e.g., indexing and looping), the model specification, the initial value setting, the monitoring convergence, and the output analyses in WinBUGS. The Rasch model will be used to illustrate these topics.

Session II: Using WinBUGS for Explanatory Item Response Theory Models
November 25th, 2008/2-4 pm/Tolman Hall, Room 2515

The second part of the workshop will illustrate the use of WinBUGS for a selection of explanatory IRT models from De Boeck and Wilson (2004). Possible models include the partial credit model, the latent regression Rasch model, the linear logistic test model (LLTM), Rasch testlet model, mixture IRT models, and multilevel IRT model. The use of WinBUGS for a simulation study will be also discussed. Participants are encouraged to bring a laptop computer with WinBUGS 1.4 (downloaded from http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/contents.shtml#install).

January 2009

Betsy J. Feldman, UC Berkeley

Workshop on Structural Equation Modeling: “A Colorful Introduction” (Revised)

Goals

The goal of this workshop is to give a general introduction to structural equation modeling (SEM). It is not possible to cover all possible topics in a week—or even in a semester—but this should give you enough background to think about how SEM might be used to address some of your research problems and to allow you to ask further questions. The workshop will take place over two 3-day sessions: in January, 14 – 16 (the week before spring semester classes begin) and in mid- to late-March (exact dates tba). Each day will comprise two roughly-3-hour sessions: lecture in the morning and hands-on work with Mplus in the afternoon. Mplus distributes a free “demo” version that has all of the features of the full program but allows a limited number of variables. A full version of Mplus will also be installed on one of the computers in the QME lab.

During the first session, in January, we will review multiple regression, discuss path analysis and confirmatory factor analysis, and learn how these components go together to become a structural equation model. During the second session, following a brief review of SEM from the first session, we will cover more complex SEM & longitudinal models. The remaining curriculum will depend, in part, on what special topics or research problems interest you. Note that I will be at Berkeley and available for questions through the rest of this school year and the summer, at a minimum.

Instructor

Betsy Feldman
b.feldman@berkeley.edu

Text

TBA

Software

Mplus 5.1, demo version
Download at http://www.statmodel.com/demo.shtml
Download the user guide at http://www.statmodel.com/ugexcerpts.shtml

Day 1

Reading

Data inspection
Regression
TBA
Mplus User Guide Ch. 1 – 2

Morning
Know your data!
Multiple Regression Review
Introduction to path diagrams

Afternoon
Introduction to Mplus

Day 2

Reading

TBA
Mplus User Guide Ch. 5

Morning
Path Analysis
Confirmatory Factor Analysis

Afternoon
Practice with Mplus: User guide examples from chapters 3 and 5

Day 3

Reading

TBA

Morning
Structural Equation Models

Afternoon
Practice with Mplus: User guide examples from chapter 5

Closing remarks, final questions, evaluation, and discussion of next session

Thank You for visiting, come back soon!