1
$\begingroup$

I have a dataset with almost 300 cases and 8 variable. Some are binary, some are ordinal variables with 4 to 5 categories. I am performing a hierarchial cluster analysis and latent class analysis in R.

For the former I use gower distance and hclust function in R. For the latent class analysis I use poLCA package. The change in agglomeration coefficient after clustering and local minima of AIC after latent class analysis suggests the same number of latent groups (4) which is also in line with my pre-study beliefs.

However, after latent class analysis, the BIC have a perfect linear correlation with number of possible latent classes when interpreted graphically. There is no local minima present with up to 10 classes. How should I interpret this finding?

$\endgroup$
0
$\begingroup$

From my experience, this type of result (i.e., no minima for BIC) tends to occur when the model get lost into over-segmentation (I can't explain "why"). However in your case I find it surprising that AIC (By the way you should rather use AIc or CAIC, especially if relatively small sample size) does not suffer from the same issue.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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