@EikoFried I have a big dataset (n=4600) with all >400 MMPI items. Would that be useful for network modeling for you?
-
400 items means 400*399/2 = 79,800 parameters to estimate. We'd need n=200,000 for that probably ;)
Well, it's a start. You could use hierarchical clustering/etc to reduce the items to a more manageable number first.
-
-
I could do all sorts of things!
-
But if it were my project, I'd do parallel analysis & exploratory graph analysis and see how many item communities to extract.
-
That N would allow for co-occurrence network, to see how often high scores in one var occur w high scores in another https://osf.io/8mj84/
-
W "co-occurrence netw" you mean visualizing cor matrix as network (i.e. no model)? I meant regularized part cor net https://arxiv.org/abs/1607.01367
-
I developed that approach to link network analyses with intra-individual methods. Makes sense e.g., for mixture distributions & co-morbidity
-
Interesting, will check. But there are many network models for intra-ind network data, e.g. R-packages mlVAR graphicalVAR gimme
-
Because parameters are inter-individual averages, even on the within-level, + general tendency (r, beta) can hide mixture distribution 2/2
-
Cool. Is OSF link the best description of the model, or is there better place to start? Thanks
- 4 more replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.