Jane Smith posted on Tuesday, March 04, 2014 - 7:30 am
I’m conducting an LTA with an interaction term following example 8.14 in version 7 of the user guide. In my model, the covariate (X) is measured at time 2, and thus C1 should not be regressed on X. Instead, I'd like to regress C2 on X. When I remove ‘C1 on X’ from the overall statement, I get an error message. I’ve reviewed example 8.13 in version 6 of the user guide and am able to remove ‘C1 on X’ without a problem. Is example 8.13 a more accurate parameterization to achieve an interaction effect with the covariate predicting C2 rather than C1?