Model syntax for Growth Model, possib...
Message/Author
 Heather Rickels posted on Tuesday, February 07, 2017 - 4:22 pm
I am trying to model growth for 4 time points in Grades 6, 7, 8, and 11. The means seem basically linear, except that Grade 11 is lower than expected. I'm running into problems specifying a model.

What would be the difference in interpretation between these two examples? Is either considered piecewise?

Heather

1) MODEL: IW SW | M6@0 M7@1 M8@2 M11*;
IW WITH SW;

2) MODEL: IW SW | M6@0 M7@1 M8@2 M11@5;
[M11*];
IW WITH SW;
 Bengt O. Muthen posted on Wednesday, February 08, 2017 - 2:31 pm
In 1), the mean and covariance structures are both influenced by the free time score at M11. It is easily interpreted in a growth model framework. In contrast, in 2) the mean and covariance structures deviate from that of a growth model.

You could also consider

2) MODEL: IW SW | M6@0 M7@1 M8@2 M11@5;
[M11*];
IW WITH SW;

You can see how the different choices influence the residuals for the means and the residuals for the variance-covariance matrix.
 Heather Rickels posted on Thursday, February 09, 2017 - 7:48 am
Dr. Muthen,

Thank you so much for your quick response. That helps a lot. I will probably go with #1 then, as it will be easier for me to interpret and explain the results.

Just to clarify the second part of your message. When you said...

You could also consider

2) MODEL: IW SW | M6@0 M7@1 M8@2 M11@5;
[M11*];
IW WITH SW;

...was that model supposed to be slightly different than my #2? Or were you suggesting to compare how my #1 and #2 affect the residuals, etc.?

Thanks again,
Heather
 Bengt O. Muthen posted on Thursday, February 09, 2017 - 6:16 pm
Yes, compare how your #1 and #2 affect the residuals and fit.