Good Day,

I am planning a study that will I analyze using Latent Class Analysis. I have three tests that will be used as inputs to the LCA and I wish to create a 2-class model.

I have some pilot data that I can use and I am looking for resources to calculate the power required for a larger study. Does anyone have any open source /free resources? (e.g. online calculators, R etc)



1 Answer 1


I don't think this kind of calculator exists so far, but you can automate it a little with LatentGold software, as explained in this Gudicha's paper - it describes how you can calculate power and gives an example code on p.34.

LatentGold is not open source, you can try to reproduce the Gudicha's steps with the other software.

  • $\begingroup$ @Maxim: How do we determine what a class is in LCA from the data. I have done some reading and only found that "finding patterns of response". But that is vague. What Mathematical/Statistical tools are we using to determine the existence of these patterns? Knn ( nearest neighbors)? Sonething else? $\endgroup$
    – MSIS
    Apr 14, 2020 at 15:57
  • 1
    $\begingroup$ @MSIS LCA is a model-based approach, which makes it very different from nearest neighbor and alike heuristics. It means LCA has an explicit model which is fit to a data with iterative estimators such as maximum likelihood. The models vary within LCA, but a common one looks a bit like logit ordinal regression. I'd advise to begin with very intuitive Alan McCutcheon's (1987) Latent Class Analysis, Sage green series. $\endgroup$
    – Maksim
    Apr 16, 2020 at 7:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.