Growth factors as predictors in Regre... PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
 Anton Rainhardt posted on Tuesday, September 22, 2015 - 8:04 am
Dear Muthens,

i have this following situation. I have a dataset of four measurement occasions. In the first three occasions the same test was introduced. This three timepoints are equidistant. The fourth data aqcuisition was some years later. The time distance to the fourth measurement occasion is much longer, and another test, though a similiar psychological construct, was conducted.

At first i tried to fit a model on the data of the first three measurement occasions. On theese, a linear model fits perfect to the data.
To take this special fourth measure into account, i thought about using the information of the slope and intercept mean and variation (of the first three measurements) as a predictor for the fourth measurement occasion in a regression.

So i wrote a model:
i s | y1@0 y2@.1 y3@.2;
4th_var ON i s;

But the model fit decreased drastically.

So i was wondering, if this is the false solution for this problem?
Are there other, elegant solutions for this problem?

Thank you very much!
 Bengt O. Muthen posted on Wednesday, September 23, 2015 - 5:18 pm
I don't think there is a solution to the problem of including a different test in a growth model unless some items are in common.

You can do the regression you show but it doesn't tell you about growth for the 4th time point. The misfit may be due to the first 3 y's having direct effects on the 4th.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message