1-1-1 with a Level 1 moderator PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Eunice Kim posted on Thursday, April 27, 2017 - 12:11 pm
I'm trying to investigate multilevel moderated mediation with variables assessed at the individual level. Basically, it is 1-1-1 mediation with a level 1 moderator (stage 1 moderator).

USEVARIABLES ARE
TeamID
x
w
m
y;

CLUSTER IS TeamID;
MISSING=ALL (999);

DEFINE: CENTER x w (GRANDMEAN);
inter1 =x*w;

ANALYSIS: TYPE IS TWOLEVEL RANDOM;

MODEL:
%WITHIN%
s1 | m ON x;
s2 | m ON w;
s3 | m ON inter1 (a2);
s4 | y ON m;
s5 | y ON x;

w (modvar1);

%BETWEEN%

[s1] (j1);
[s4] (f1);

s1 WITH s2 s3 s4 s5 x w m inter1 y;
s2 WITH s3 s4 s5 x w m inter1 y;
s3 WITH s4 s5 x w m inter1 y;
s4 WITH s5 x w m inter1 y;
s5 WITH x w m inter1 y;

m ON x (ab);
m ON w;
m ON inter1 (a3);
y ON m (bb);
y ON x;

w (modvar2);

MODEL CONSTRAINT:
NEW(walphahigh walphalow wmmhigh wmmlow wDiff);
walphahigh = j1+(a2*(modvar^.5));
walphalow = j1+(a2*(-1)*(modvar^.5));
wmmhigh = walphahigh*f1;
wmmlow = walphalow*f1;
wDiff= wmmhigh - wmmlow;


OUTPUT: TECH1 TECH3 CINTERVAL;
 Eunice Kim posted on Thursday, April 27, 2017 - 12:22 pm
I'm trying to investigate multilevel moderated mediation with variables assessed at the individual level. Basically, it is 1-1-1 mediation with a level 1 moderator (stage 1 moderator). The below syntax is correct?

USEVARIABLES ARE
TeamID
x
w
m
y;

CLUSTER IS TeamID;
MISSING=ALL (999);

DEFINE: CENTER x w (GRANDMEAN);
inter1 =x*w;

ANALYSIS: TYPE IS TWOLEVEL RANDOM;

MODEL:
%WITHIN%
s1 | m ON x;
s2 | m ON w;
s3 | m ON inter1 (a2);
s4 | y ON m;
s5 | y ON x;

w (modvar1);

%BETWEEN%

[s1] (j1);
[s4] (f1);

s1 WITH s2 s3 s4 s5 x w m inter1 y;
s2 WITH s3 s4 s5 x w m inter1 y;
s3 WITH s4 s5 x w m inter1 y;
s4 WITH s5 x w m inter1 y;
s5 WITH x w m inter1 y;

m ON x (ab);
m ON w;
m ON inter1 (a3);
y ON m (bb);
y ON x;

w (modvar2);

MODEL CONSTRAINT:
NEW(walphahigh walphalow wmmhigh wmmlow wDiff);
walphahigh = j1+(a2*(modvar^.5));
walphalow = j1+(a2*(-1)*(modvar^.5));
wmmhigh = walphahigh*f1;
wmmlow = walphalow*f1;
wDiff= wmmhigh - wmmlow;


OUTPUT: TECH1 TECH3 CINTERVAL;
 Bengt O. Muthen posted on Thursday, April 27, 2017 - 6:29 pm
You can't give a parameter label to a statement like this

s3 | m ON inter1 (a2);

because it does not refer to a parameter. The s3 variable has parameters on the Between level.

Typically, you won't find as many random slopes as your model includes; perhaps only one or two at most.
 Eunice Kim posted on Friday, April 28, 2017 - 2:06 pm
Thanks for your input, Bengt. I've corrected my syntax. Does this one look OK to you? I have a quick follow-up question for you. Among four random slopes I have, which one can be fixed? Is there any right way to choose fixed over random slopes?

USEVARIABLES ARE
TeamID
x
w
m
y;

CLUSTER IS TeamID;
MISSING=ALL (999);

DEFINE: CENTER x w (GRANDMEAN);
inter1 =x*w;

ANALYSIS: TYPE IS TWOLEVEL RANDOM;

MODEL:
%WITHIN%
s1 | m ON x;
s2 | m ON w;
s3 | m ON inter1;
s4 | y ON m;
y ON x;

w (modvar1);

%BETWEEN%

[s1] (j1);
[s3] (a2);
[s4] (f1);

s1 WITH s2 s3 s4 x w m inter1 y;
s2 WITH s3 s4 x w m inter1 y;
s3 WITH s4 x w m inter1 y;
s4 WITH x w m inter1 y;

m ON x (ab);
m ON w;
m ON inter1 (a3);
y ON m (bb);
y ON x;

w (modvar2);

MODEL CONSTRAINT:
NEW(walphahigh walphalow wmmhigh wmmlow wDiff);
walphahigh = j1+(a2*(modvar^.5));
walphalow = j1+(a2*(-1)*(modvar^.5));
wmmhigh = walphahigh*f1;
wmmlow = walphalow*f1;
wDiff= wmmhigh - wmmlow;
 Bengt O. Muthen posted on Friday, April 28, 2017 - 3:05 pm
There is no right way to choose which slopes are fixed or random - it needs to be explored.

To judge your setup, send the output to Support along with your license number.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: