Plotting Latent I * Manifest Predicto...
Message/Author
 Andrew Ribner posted on Sunday, January 20, 2019 - 11:25 am
Hi,

I've run a LGCM with a latent intercept, linear, and quadratic slope across 6 timepoints. I've added an interaction with the following code

I S Q | test1@0 test2@1 test3@2 test4@3 test5@4 test6@5;

IxM | I XWITH M;

I ON M
!covariates follow
;

S ON I (b1)
M (b2)
IxM (b3)
!covariates follow
;

Q ON I
M
IxM
!covariates follow
;

I can plot the effect of IxM on S with the following code:

MODEL CONSTRAINT:
NEW(LOW_M MED_M HIGH_M LM_I MM_I HM_I);

LOW_M = -1; ! -1 SD below mean
MED_M = 0; ! mean
HIGH_M = 1; ! +1 SD below mean

! simple slopes for each value of M

LM_I = b1 + b3*LOW_M;
MM_I = b1 + b3*MED_M;
HM_I = b1 + b3*HIGH_M;

PLOT (PLM_I PMM_I PHM_I);
LOOP (I, -3, 3, 0.1);

PLM_I = LM_I * I;
PMM_I = MM_I * I;
PHM_I = HM_I * I;

What I would like to do is plot the effect on the growth over time rather than the growth parameter (i.e., see the effect on the overall trajectory rather than on the slope). Is that possible?

Thanks!
 Bengt O. Muthen posted on Monday, January 21, 2019 - 12:52 pm
Yes, this is possible. To see how i and s influence the outcomes Y over time, see our Topic 3 Short Course video and handout, slides 96-98.

For effect calculations with growth, see the article

Feingold, A. (2018). Time-varying effect sizes for quadratic growth models in multilevel and latent growth modeling. Structural Equation Modeling: A Multidisciplinary Journal. DOI: 10.1080/10705511.2018.1547110