2
$\begingroup$

I'm pretty new to Latent class analysis, and I obviously still have a lot to learn.

I have a data set with 10 000 observations and 20 variables. I'm trying plot the probabilities of each of my identified classes in the latent class analysis I did using the poLCA package in R (http://cran.r-project.org/web/packages/poLCA/poLCA.pdf).

What I'm trying to do is something like Figure 1 in Quek et all. 2013 (http://www.researchgate.net/publication/258441959_Concurrent_and_simultaneous_polydrug_use_latent_class_analysis_of_an_Australian_nationally_representative_sample_of_young_adults)

The problem is that every time I run the poLCA() function, my plot changes, so do the the probabilities of each class and of course this is because the results of the poLCA() changed.

Is this normal that doing two or more times the same thing with the same data I get different results ? I'm feeling very uncomfortable here, so if someone could explain me what's happening here, I could maybe live with myself again :)

Improve this question
$\endgroup$

2 Answers 2

Reset to default
5
$\begingroup$

Yes: it is perfectly normal. The algorithm used to find the ML estimation of a LCA can stop to a local maximum of the likelihood depending on the starting values. (This is a quite general problem in statistics) Unless you fix the starting values, poLCA automatically generates them at random each time you run it. The usual way to avoid this problem and find the global maximum likelihood solution is to try several starting values, that is, to repeat the function call several times and choose the one associated with the highest likelihood. This is simply done in poLCA via the "nrep" option.

Note that, in general, the higher the number of variables, and the number of modalities for variable in a LCA model, the higher the number of repetitions needed to be reasonably sure of the result.

Besides, I am assuming you do not have identifiability issues, that is, the number of independent parameters of your model is lower than the number of distinct configurations of the variables. If different estimations of the model are associated to the same likelihood value, the model you specified is not identifiable.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Hi ruggero, thank you for your answer.Do you have any scientific references/articles for this? $\endgroup$
    – leakciM
    Commented Mar 16, 2015 at 20:50
  • $\begingroup$ Collins, Linda M., and Stephanie T. Lanza. Latent class and latent transition analysis: With applications in the social, behavioral, and health sciences. Vol. 718. John Wiley & Sons, 2013. $\endgroup$
    – ruggero
    Commented Mar 18, 2015 at 15:39
0
$\begingroup$

I know this post is old, but I had this same question and just figured it out. Assuming you use the "nrep" option to find the global maximum likelihood, the poLCA package will assign your observations to different classes when using a different seed number. However, the composition of those classes is the same as with the previous seed - they are just numbered differently. For example, if I set my seed=1234, and then run the exact same code with the seed=5678, LCA class 2 for seed=1234 might now be class 5 for seed=5678.

Improve this answer
$\endgroup$

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.