AMOS: covariance matrix only input based on ordered catagorical data - possible?

#1
Hello,

I am attempting to use AMOS to run a number of structural equation models. Several of my manifest variables are ordered categorical, a few (intended to be exogenous) are even binary. Unfortunately I only have access to secondary summary data (i.e. a covariance matrix with means and SD). I can run the models "normally" without identifying the data as ordered categorical though I don't believe it's acceptable to do so. I am attempting to allow non-numeric data and re-code as ordered categorical so that might use Bayesian estimation. Does anyone with experience using matrix input only without raw data have any thoughts on this? I'm not sure that it is even conceptually sound but am unsure how to proceed. Any help would be greatly appreciated!!
 

spunky

Doesn't actually exist
#2
Does anyone with experience using matrix input only without raw data have any thoughts on this?
for most purposes, in the case of discrete/ordered categorical data, you *MUST* supply the full dataset to any SEM software for it to run. the covariance-as-input trick is only appropriate if your data is continuous.

still, i guess you could always try and give it the covariance/correlation matrix and hope for the best? here's an article that may be helpful. maybe your data fits one of the models the authors tested and it won't be super horrible to analyze the covariance/correlation matrix as if it came from continuous data.

Rhemtulla, M., Brosseau-Liard, P., & Savalei, V. (2012). How many categories is enough to treat data as continuous? A comparison of robust continuous and categorical SEM estimation methods under a range of non-ideal situations. Psychological Methods. Advance online publication. doi: 10.1037/a0029315