Greetings!
I am looking for a data reduction method for binary weighted variables. I did a factor analysis on a tetrachoric matrix, but it does not account for weights. It seems like Latent Class Analysis might be useful, but again, I don't know how to incorporate the weights, and I don't want to establish a number of factors a priori, rather look at the data (like a scree plot in FA) to decide how to split the variables into factors/ latent variables.
Specifics: I have 8 (yes/no) outcomes I want to create logistic models for. Rather than making 8 models, I think there should be a natural trend among some of the variables so I am hoping to reduce to a few factors/ latent vars to model. Thanks!
