Probing & plotting interaction
Message/Author
 SherrieW posted on Sunday, August 21, 2011 - 2:51 pm
I need help to plot continuous latent variable interactions.

I used LMS in Mplus to find the interaction:
ANALYSIS:
type = missing;
TYPE = RANDOM;
ALGORITHM = INTEGRATION;

Model: m1 by a1 a2 a3 a4;
m2 by a5 a6 a7 a8;
m3 by a9 a10 a11 a12;
m4 by a13 a14 a15 a16;
m5 by a17 a18 a19 a20;

m1xiv | m1 XWITH iv;

dv on iv (a1);
dv on m1xiv (a3);
dv on covariates;
dv on m1 m2 m3 m4 m5;

And created constraints to probe the interaction:

MODEL CONSTRAINT: NEW(ss1-ss10);
ss1=a1+a3*1;
ss2=a1+a3*2;
ss3=a1+a3*3;
ss4=a1+a3*4;
ss5=a1+a3*5;
ss6=a1+a3*10;
ss7=a1+a3*15;
ss8=a1+a3*20;
ss9=a1+a3*25;
ss10=a1+a3*30;

But, now, how do I plot the interaction? Where do I find this information?
 Bengt O. Muthen posted on Monday, August 22, 2011 - 10:02 am
Take a look at how the interaction is expressed on slide 164 - for the Topic 3 handout of 5/17/2010.
 Sher posted on Tuesday, August 23, 2011 - 10:45 am
How do I find the mean of the interaction (z) and plus or minus one standard deviation around the mean of z?
 Sher posted on Tuesday, August 23, 2011 - 5:32 pm
Correction: How do I find the mean of the moderator (z) and plus or minus one standard deviation around the mean of z?
 Bengt O. Muthen posted on Tuesday, August 23, 2011 - 5:44 pm
For an observed moderator, do a Type=Basic run.
 Sher posted on Tuesday, August 23, 2011 - 6:16 pm
The moderator is latent. Is there a way to find the mean, SD for a latent moderator?
 Linda K. Muthen posted on Wednesday, August 24, 2011 - 1:01 pm
See TECH4.
 Sher posted on Wednesday, August 24, 2011 - 2:42 pm
I tried TECH4 and received the following message:

*** WARNING in OUTPUT command TECH4 option is not available for TYPE=RANDOM.
Request for TECH4 is ignored.
 Linda K. Muthen posted on Wednesday, August 24, 2011 - 3:02 pm