I have two growth models, one modeling growth across one phase and one modeling growth across another phase. each has a separate intercept and slope. i need to test whether the intercepts are significantly different from each other. this is how i specified it:
Look at the User's Guide. Note that i1, and i2 are variables. You have to say which parameters for these variables you are focusing on - I assume the means. Then you use the parameter labeling approach in the Model command:
I have a follow-up question to this post. Is there a way to test whether covariates have an effect on the change in intercepts? In other words, I want to know if the difference in means of the intercepts from time 1 to time 2 varies as a function of my model covariates.
I don't think I was clear enough regarding this question. The model in question includes two separate growth models - one modeling growth across 4 time points in middle school and one modeling growth across 4 time points in high school. I1 is the last time point in middle school and I2 is the first time point in high school - these two intercepts represent functioning immediately before (I1) and immediately after (I2) the transition to high school. In addition to examining these 2 growth trajectories, I performed a model test to determine if the difference between the mean of I1 (m1) and the mean of I2 (m2) is significantly different from zero. Having found this difference to be significant, I now want to determine if a set of covariates predict this difference, but I do not know how to model this. How would I do this?
You can test this by regressing i1 and i2 on a covariate or set of covariates in a series of analyses. Note that in these regressions there are two possible sources of mean differences -- intercepts and slopes.
Step 1. Regress i1 and i2 on x for example to see if the covariate has signficant effects.
Step 2: Regress i1 and i2 on x where the two regression coefficients are held equal.
Do a difference test to determine whether the regression coefficients are the same or not. If the coefficients are the same, this means that the covariate does not predict the difference. If they are different, this means that the covariate does predict at least part of the difference.