Latent profile analysis/latent class analysis, tidyLPA: How to implement maximum likehood with robust standard errors, and check if it is needed.

I've started to use tidyLPA for a project of mine and have a question about assumptions for conducting LPA. As part of my project I have been running some multiple regressions and have found some mild non-normality but, and this one concerns me more due its magnitude, heteroskedasticity. I was wondering if this is something I should be concerned about for conducting my LPA as I am using the same data and the same variables. Additionally, there was also univariate non-normality on many of the variables being used, though the multivariate non-normality was not that severe in the regressions.

As such, I'd like to ask what sort of assumptions should I be checking for when running an LPA and how would I check these in tidy LPA? And if they are violated how would I deal with them? I have read in Spruk et al. (2020) that maximum likehood with robust standard errors can help address violations of normality (and I'm guessing heteroskedasticity) in LPA but I have not found any guides on how to imperilment this in tidyLPA or if Tidy LPA already does use these robust std errors.

Any feedback or resources would be very helpful.

Kind regards,

Source: Spurk, D., Hirschi, A., Wang, M., Valero, D., & Kauffeld, S. (2020). Latent profile analysis: A review and “how to” guide of its application within vocational behavior research. Journal of Vocational Behavior, 120, 103445. Redirecting

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.