Statistical Software

Here are some of my personal favorites, which are freely available on the Internet


LEM can fit latent class models, survival models, and lisrel-type models for categorical data. The author of the program is Jeroen K. Vermunt.
A manual
is also available. This software is used in my course: Seminar: Selected topics in measurement theory. Sample LEM programs for
mixture and latent class signal detection models are available in my 2002 MBR article and on my homepage.

aML — aML is now free!

aML allows one to fit multilevel multiprocess models typically used in econometrics and biostatistics. It can be used to fit multivariate signal
detection models with and
without sample selection, as well as mixture and latent class signal detection models. A sample program for fitting
a multivariate signal detection model with sample
selection is given on my homepage.


Mixor can be used for ordinal logistic regression with random effects; the author is Donald Hedeker. There is also software for regression,
poisson regression, and survival analysis with random effects, as well as a list of publications involving use of the software.
Click on download and directions will follow.


WinBUGS performs Bayesian inference Using Gibbs Sampling, and can fit many types of complex models. A users manual and relevant
references are provided. Lately I've been using the open version, OpenBUGS, some sample programs are given on my homepage.


R is an environment for statistical computing and graphics; it is similar to the commercially available software S-PLUS. It can be used for
many types of statistical analysis and, thanks to contributed packages, can fit many types of advanced models.

Next are commercially available software that I like and also use in some of my courses

Latent Gold

Latent Gold is software for latent class and mixture models, including models with continuous latent variables. Version 4.5 includes
syntax that allows one to fit a wide variety of models, including the unequal variance SDT model, latent class SDT, mixture SDT,
multivariate SDT, and even multivariate models with sample selection. Examples can be found in many of my articles, I use this
software extensively in my research.


Mplus is software for structural equation modeling. A nice feature is that it can also be used for latent class models, and combined
structural equation/latent class
models. Some sample programs for using Mplus to fit latent class and mixture signal detection models 
can be found on my
homepage and on my presentations page. This software is used in my course: Latent Structure Analysis.


HLM is software for hierarchical linear and nonlinear modeling. The latest version is very user friendly. This software is used in my
course: Multilevel a
nd longitudinal data analysis, along with Mplus.


Stata is a general statistical package. It offers many types of analysis and models, including many add-ons. One I find of particular interest
is GLLAMM, which is a Stata program to fit generalized linear latent and mixed models. Note: TC students, faculty, and staff can obtain
this software at a significant discount because the computing center at Columbia has an arrangement with Stata to distribute the
software through their Gradplan program; for details click here