Very informative
I really appreciate your response.
First, thank you for pointing out the susceptibility to label-switching caused by the number of latent factors. There’s no need, actually, in my case for reduction to more than one factor, so I’ll probably reduce the number of factors to 1. I did that initially thinking that, because components 1 and 3 ar enot related to 0 and 2, that I would either have to use multiple factors, or remove some of these less correlated components. Since this is just an example, i’ll go with the latter.
@gBokiau my first question has to do with the intercepts you mentioned. I previously thought that, in this type of model, the intercept is fixed to 0, because the latent factors’ scale and location are kind of arbitrary. Is that not the case? If so, to clarify, do you mean I would need to add intercepts for each component weight in the calculation of the “p” parameter?
I’ll dig in further to your suggestions as well and let you know my progress. Thanks again for your help!