I would like to have the unrotated solution for the factor analysis for categorical data. I stored the matrix of polychoric correlation from Mplus and in R I could compute the same eigenvalues from this corr matrix. Then I read the corr matrix into SPSS and run factor analysis there, but the varimax loadings between SPSS and Mplus differ a lot (using ULS in both, rotation with and without Kaiser normalization). Any idea what else could be different? Currently I'am interested in the loadings of the first factor only. Could I just run EFA 1 1 to get loadings of the unrotated solution?