|
|
| Message/Author |
|
|
| rgrove posted on Wednesday, September 05, 2018 - 10:22 pm
|
|
|
Hi,
I was wondering whether you could help me with some syntax for a model. We have longitudinal data at four time points measuring alcohol use. We also have four other variables that are measured at the same time points (depression, anxiety, conduct and hyperactivity). I want to no only look at the direct effects of each variable, but also estimate the moderating effects of these four variables on alcohol use over time. For example, does anxiety moderate the effect of depression on alcohol use over time? Would it be most appropriate to use be latent growth modelling in this situation or a model that includes interaction variables. Any help would be much appreciated! |
|
|
|
|
You can use Define to create interactions between these time-varying covariates and regress your DV on them at each time point.
If you specify a growth model for your DV, perhaps instead you want an interaction between a covariate and the growth slope using XWITH. |
|
|
| rgrove posted on Wednesday, September 12, 2018 - 8:21 pm
|
|
|
Hi Bengt,
Thanks for getting back to me. I have used the define term to have a look at interaction terms. Would this need to be done for all four time points? Please see syntax below.
DEFINE: CENTER T1_GAD T1_PHQ T1_SDconduct T1_SDhyper (grandmean);
T1IntPHQ_Hyper = T1_PHQ * T1_SDhyper;
T1IntPHQ_Conduct = T1_PHQ * T1_SDconduct;
T1IntGAD_Conduct = T1_GAD * T1_SDconduct;
T1IntGAD_Hyper = T1_GAD * T1_SDhyper;
MODEL:
ALC BY T1quantity* T3quantity T5quantity T7quantity;
ALC@1;
ALC ON T1_GAD T1_PHQ T1_SDconduct T1_SDhyper;
ALC ON T1IntPHQ_Hyper T1IntPHQ_Conduct T1IntGAD_Conduct T1IntGAD_Hyper;
With the growth model, I'm still unsure as to how to specify the interaction terms. I have the following model:
DEFINE: CENTER T1_GAD T1_PHQ T1_SDconduct T1_SDhyper (grandmean);
MODEL:
i s | T1quantity@0 T3quantity@1 T5quantity@2 T7quantity@3;
s ON T1_GAD T1_PHQ T1_SDconduct T1_SDhyper;
Would I then specify T1_GAD XWITH s to get the interaction terms here?
Thanks again for all your help |
|
|
|
|
Q1: Yes
Q2: Yes |
|
|
| Back to top |
|
|
