10 Time-varying covariates x 9 depend...
Message/Author
 EFried posted on Friday, March 09, 2012 - 9:42 am
Hello!

I have
* 9 dependent variables (9 categorial depressive symptoms, 0-4), "How do you feel at the moment."
* 10 time-varying covariates (life events 1-10: yes/no), "Did A happen to you within the last 3 months? Did B happen to you within the last 3 months" (etc).
* 5 measurement points, time interval 3 months. Both the dependent variables and the time-varying covariates where measured 5 times for each person.

My question is whether life events do have a systematic influence on depressive symptoms: do specific life events cause specific patterns of depressive symptoms.

(1) Is there any way in MPLUS to calculate ONE interaction test for this, that would conclude: yes, there is some kind of interaction between life events and depressive symptoms.
(2) How would I further elaborate this then? E.g. Life Event "1" leads to more depressive symptoms A, B and C, whereas Life Event "2" leads to symptoms B, E and G.

Thank you
Torvon
 Linda K. Muthen posted on Friday, March 09, 2012 - 3:36 pm
You can do the following where y1, y2, and y3 are repeated measures of y and x1, x1, and x3 are time-varying covariates:

y1 ON x1 (1);
y2 ON x2 (1);
y3 ON x3 (1);
 EFried posted on Monday, March 12, 2012 - 5:20 am
Linda, thank you. I'm using this code snippet in my GMMs already to include effects of time-varying covariates on my dependent variables.

However, what I want to know is whether

x1 affects y1, y2, y3 differently than
x2 affects y1, y2, y3

(1) How could I do such a thing?
(2) And how could I do it with the complications that x is a time-varying covariate, I have 5 measurement points, 10x and 9?

Thank you!
 EFried posted on Monday, March 12, 2012 - 10:13 am
I forgot a "y" there:

(2) And how could I do it with the complications that x is a time-varying covariate, I have 5 measurement points, 10x and 9y?
 Linda K. Muthen posted on Monday, March 12, 2012 - 5:05 pm
You would use script like:

y1 ON x1 (p10);
y1 ON x2 (p11);

MODEL TEST:
0 = p10 - p11;