Anonymous posted on Tuesday, May 21, 2002 - 6:13 am
Dear Linda & Bengt,
I was wondering about the meaning of the estimated means from the residual output option in a GGMM run when covariates are present. Do the estimated means referr to the case when the covariates are "0"? Is there any difference between categorical and continuous variables? Thank you for your input.
bmuthen posted on Tuesday, May 21, 2002 - 10:14 am
With continuous outcomes, the estimated means are the "marginal" means, that is evaluated at the sample means of the covariates. With categorical outcomes, the estimated thresholds are presented. These thresholds are from the regressions of outcomes on covariates, so conditional on the covariates, analogous to an intercept in regular regression.
thank you for your quick response. Does it make sense to plot the curves based on the estimated means from a model with covariates? I would like to avoid to plot the trajectories for each combination of the covariates.
bmuthen posted on Tuesday, May 21, 2002 - 11:21 am