AERA 2005 Presentation

Each link below will be added by end of April 2005

1. PowerPoint Presentation of "Multilevel Generalized Models: A Comparison of Population-averaged and Unit-specific Modeling"

2. Paper version (contains more detail and examples)

3. Data used in paper and presentation

• Violence Against Women (Stata .dta file; text file)

• Monitoring the Future 2003 (Stata .dta file; text file)

• NELS88 Dropout  (Stata .dta file; text file)

• Simulated Y and X (Stata .dta file; text file)

Notes to be added --

Generalized linear mixed models (GLMM) are useful for modeling outcomes that take non-normal forms (e.g., dichotomies such as pass vs. fail, or infrequent counts such as number of worker absences in a one-month period) and have correlated observations (e.g., correlated observations may result from clustering such as observations from workers clustered by schools or repeated measurements taken from individuals). 

Below are my notes on running GLMMs in R (an open-source statistical package based on S-Plus) and Stata.

Links to be added soon

• GLMM in R 

• GLMM in Stata with gllamm

Last revised 25 April 2005.