Multilevel, mixed, and hierarchical models

Stata provides a complete array of features for estimating and analyzing multilevel models (also called mixed models or hierarchical models) on continuous, binary, and count outcomes.

Explore Stata 12’s resources on multilevel, mixed, and hierarchical model capabilities.

Overview of linear multilevel model capabilities in Stata

Overview of estimators for multilevel models with count and binary outcomes

List of multilevel, mixed, or hierarchical model capabilities

Introduction to multilevel mixed-effects models with examples (from the Stata manual)

(Stata also provides commands for fitting logistic and Poisson multilevel models.)

See Stata’s dialog box for xtmixed

The user-written package generalized linear latent and mixed models (gllamm) provides a full range of generalized linear model (GLM) estimators for multilevel processes, including estimators with latent variables. See the GLLAMM website at http://www.gllamm.org for documentation, extensive discussions, and resources.

Stata Press publishes a 562-page book on multilevel modeling using Stata.

Public training course on multilevel/mixed models

Stata Journal articles

“metandi: Meta-analysis of diagnostic accuracy using hierarchical logistic regression” by Roger M. Harbord and Penny Whiting (abstract)

“Fitting mixed logit models by using maximum simulated likelihood” by Arne Risa Hole (abstract)

“Estimating variance components in Stata” by Yulia V. Marchenko (abstract) (full article)

“Reliable estimation of generalized linear mixed models using adaptive quadrature” by Sophia Rabe-Hesketh, Anders Skrondal, and Andrew Pickles (abstract) (full article)

“Review of Multilevel and Longitudinal Modeling Using Stata, Second Edition, by Sophia Rabe-Hesketh and Anders Skronda” by Nicholas J. Horton (abstract) (full article)

Stata is also a complete integrated statistical package that provides everything you need for data analysis, data management, and graphics. To learn more, click here.