Regions of Significance with respect ...
Message/Author
 Steven De Laet posted on Wednesday, March 19, 2014 - 1:25 am
I was wondering whether the following is possible in Mplus:

I have an effect of X on Y with a dichotomous moderator Z.

I would like to define regions of significance: specific values of X at which the regression of Y on X is different between the two values of Z. In other words, I would like to examine the values of X for which the moderator Z and outcome variable Y are significantly related.

Thank you very much!
 Linda K. Muthen posted on Wednesday, March 19, 2014 - 10:39 am
See the loop plot of Example 3.18.
 Tess Tagge posted on Monday, November 13, 2017 - 7:29 pm
I am trying to probe latent categorical interactions using the Johnson-Neyman technique. I used the loop plot from example 3.18, but I am unsure where to find the specific values and 95% CIs.

My syntax is attached if that is helpful:

USEVARIABLES ARE GEN AGE DAS1 DAS2 AG CO EX NE OP PSC1-PSC19;
CLUSTER IS CPL;
CATEGORICAL ARE PSC1-PSC19;

ANALYSIS:
ESTIMATOR IS MLR;
TYPE IS RANDOM COMPLEX;

MODEL:
COMM BY *PSC1-PSC19;
COMM@1;
[COMM@0];

CGEN | COMM XWITH GEN;
CEX | COMM XWITH EX;
CAG | COMM XWITH AG;
CCO | COMM XWITH CO;
CNE | COMM XWITH NE;
COP | COMM XWITH OP;

DAS2 ON COMM (b1)
AG (b2)
CAG (b3)
OP (b4)
COP (b5)
CCO CEX CNE CGEN CO EX NE GEN AGE DAS1;

MODEL CONSTRAINT:
PLOT (lowAG highAG lowOP highOP);
LOOP(PSC, -3, 3, 0.1);
lowAG = (b1+b3*(-1))*PSC+b2*(-1);
highAG = (b1+b3*(+1))*PSC+b2*(+1);
lowOP = (b1+b5*(-1))*PSC+b4*(-1);
highOP = (b1+b5*(+1))*PSC+b4*(+1);

PLOT:
TYPE = PLOT2;

OUTPUT:
TECH1 TECH8 STAND SAMPSTAT CINT;
 Bengt O. Muthen posted on Wednesday, November 15, 2017 - 10:41 am
Send the output file and the gh5 file so we can look at the plots.