|LCA Stata Plugin:|
|Users’ Guide:||LCA Stata Plugin Users’ Guide (v. 1.2.1)|
Note: The feature of the plugin that triggered matrix size requirements has been removed. Users should not see that problem in this or future releases.
The LCA Stata plugin was developed by the Methodology Center to allow Stata users to perform latent class analysis (LCA). The plugin makes it possible to pre-process data, fit a variety of latent class models, and post-process the results without leaving the Stata environment. Compatible with Stata for Windows. Features include
- simple model specification,
- multiple-groups LCA,
- LCA with covariates (prediction of latent class membership),
- baseline-category multinomial logit model or binary logit model for prediction,
- posterior probabilities available in output,
- parameter estimates available in output,
- optional Bayesian stabilizing prior to handle sparseness issues in estimation,
- accounts for sampling weights and clusters,
- the ability to assess identification of models with covariates via multiple random starts,
- indication of which latent class is the best match for each individual, and
- the option to generate 20 random draws for each individual’s class membership based on posterior probabilities.
Read about latent class analysis.
LCA Stata Plugin (Version 1.2) [Software]. (2015). University Park: The Methodology Center, Penn State. Retrieved from methodology.psu.edu
Lanza, S. T., Dziak, J. J., Huang, L., Wagner, A. T., & Collins, L. M. (2015). LCA Stata plugin users’ guide (Version 1.2). University Park: The Methodology Center, Penn State. Retrieved from methodology.psu.edu
We maintain functions that enhance the functionality of the LCA Stata Plugin.
LCA Distal – for estimating the association between a latent class variable and a distal outcome using a model-based approach
LCA Bootstrap – for performing the bootstrap likelihood ration likelihood test
LCA Stata Plugin for Latent Class Analysis
In its simplest form, the LCA Stata Plugin allows the user to fit a latent class model by specifying a Stata data set, the number of latent classes, the items measuring the latent variable, and the number of response categories for each item. Multiple-groups LCA can be run using the GROUPs option; users can examine measurement invariance across groups by adding the measurement option. Additional parameter restrictions can be provided as well.
Continuous and categorical covariates can be included in the COVariates option in order to examine the relation between each covariate and the probability of latent class membership. Prediction can be modeled using a baseline-category multinomial logit model or a binary logit model.
A Bayesian stabilizing prior can be invoked when sparseness is an issue for parameter estimation. Random starting values can be generated by the program, or the user can provide starting values.
An empirical demonstration of PROC LCA (which includes the same functionality) appeared in Structural Equation Modeling:
Lanza, S. T., Collins, L. M., Lemmon, D. R., & Schafer, J. L. (2007). PROC LCA: A SAS procedure for latent class analysis. Structural Equation Modeling, 14(4), 671-694. PMCID: PMC2785099 View article
The download package includes examples that you can run in the LCA Stata Plugin.