#latent class analysis

[statsineducation]

Slides, papers, example data, and example Mplus inputs for my talk on Latent Class / Profile analysis for the Research Methodology Center in October 2016. 

[statsineducation]

Notes from Methods Colloquium 9/11/2015

If you follow the link, you’ll find a folder that contains the powerpoint slides from the 1-hour talk introducing LCA. The folder also contains an example dataset (SPSS and .dat formats), Mplus input, and Mplus output for the example data from Pew Research. 

I have a few other pieces of information about LCA here on this website. See HERE for sample size guidelines, and HERE for more information about Entropy.

[statsineducation]

Latent Classes and Entropy

When you run a latent class analysis (LCA) or latent profile analysis (LPA), one of the indicators of model fit you should examine from the output is entropy. The entropy statistic is printed in the Mplus output.

It is an estimate of how distinct the identified groups / classes/ clusters are from one another. In other words, it assesses the extent to which the groups you identify in a latent class analysis are different from one another. 

ROT: Entropy values greater than .80 indicate a good separation of the identified groups (Ramaswamy et al., 1993)

Ramaswamy, V., DeSarbo, W. S., Reibstein, D. J., & Robinson, W. T. (1993). An empirical pooling approach for estimating marketing mix elasticities with PIMS data. Marketing Science, 12(1), 103−124.

[statsineducation]

Latent Class Analysis Sample Size

Latent Class Analysis (and Latent Profile Analysis) requires samples sizes of at least of 250, though even that number can still be unreliable. Generally, a sample size of 500 seems to be a consensus for best practices. 

  Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14, 535–569.