LIPpp estimated effect size values, and model fit comparisons

[Jack serna]

Hi everyone! 

I want to share some new research just in case the authors/maintainers were not aware. Grimm published in Frontiers in Education, estimated effect sizes to describe the amount of improvement in model fit (see here: https://doi.org/10.3389/feduc.2021.613645). Would love to see this integrated in tidyLPA by default, but for now can calculate it after the fact. 

After some extensive research on LPA and mixture modelling as part of an analysis I will be proposing for a Master's thesis, it still seems unclear what the standard approach is to comparing models. tidyLPA helps tremendously by providing the suggestion based on AHP. Though, there are also people commenting about the replicability from the mclust approach, where mplus uses multiple starts. Then, the simulation studies with mixed results on information criteria and likelihood ratio tests. There is the promise that tidyLPA dev version will include OpenMX, combining mplus and mclust approaches. It's unclear whether that dev version is released as I don't see it in the changlog. Perhaps, I could loop tidyLPA estimate profiles 50 times to count replicated results? That's tiresome, but would that be recommended?