Jen posted on Saturday, November 15, 2014 - 12:32 pm
Hope you are well. I have a basic multigroup growth model. I am adding a few covariates (that I'll regress i s & q on). This model converges and fits very well when I do not center. I would ideally like to center the covariates to aid in the interpretation of i, s, and q (specifically, I would like to say if there is a significant slope and quadratic in each group at the mean of all covariates). However, when I center any of the covariates (or all of them), suddenly there are many "NaN" values in my output (for the Loglikelihood, AIC, BIC, and many coefficients). Is there a reason I should expect this in the multigroup context?