About model select & difference between cor and cor_auto in ggmmodselect/thresholded EBICglasso about bootnet

[SachaEpskamp]

Hi! These are two very outdated blog posts so you might not want to rely on these too much. As for the choise of estimator we wrote this paper:

https://psycnet.apa.org/record/2022-06515-001

And another useful paper might be:

https://psycnet.apa.org/record/2022-52634-001

If you want to compare samples the estimator ideally is the same, and it is best to use something like a multi-group analysis or permutation test (e.g., NetworkComparisonTest). Also, I don't recommend using the threshold in EBICglasso anymore nor do I recommend using polychoric correlations as input (see above paper for simulations showing this doesn't work very well).

The best updated recommendations can be found in our new textbook:

https://www.routledge.com/Network-Psychometrics-with-R-A-Guide-for-Behavioral-and-Social-Scientists/Isvoranu-Epskamp-Waldorp-Borsboom/p/book/9780367612948

[Brooks0303]

GET IT!
THX for your reply. Ill read these papers and book to find more suggestion.
And thanks in advance if there are more confusion😀