Comparing effect sizes of two longitudinal linear relationships

[dbt_healthpsych]

Is there a way to determine if one longitudinal linear relationship is significantly stronger than another? I was planning on using regression (moderated multiple regression with PROCESS macro for SPSS because I have two groups and covariates) and simply looking at the R squared values, but my adviser wanted me to make sure there was not a way to test this.

In model 1: x= change in variable A from time 1 to time 2, y= variable B at time 3.

In model 2: x= variable B at time 2, y= change in variable A from time 2 to time 3.

(change in variable A will be residualized change scores)

Any ideas you might have would be appreciated.

 

[CB]

Hi there :welcome:

I guess I would wonder here what it is that you're really trying to find out. Given that your two models estimate quite different things, it's sort of obvious that the two population slopes won't be exactly the same. (And that's all a significance test would be trying to establish - whether the two slopes are exactly the same in the population). So how have you come to be interested in making this comparison?

 

[ondansetron]

Hi there :welcome:

I guess I would wonder here what it is that you're really trying to find out. Given that your two models estimate quite different things, it's sort of obvious that the two population slopes won't be exactly the same. (And that's all a significance test would be trying to establish - whether the two slopes are exactly the same in the population). So how have you come to be interested in making this comparison?

Worth adding that the dependent variables are different, so you can't compare R-square values, either.