1-1-1 mediation in a balanced factori...
Message/Author
 Hadar Nesher Shoshan  posted on Wednesday, August 03, 2016 - 3:32 am
Dear Mplus team (I already wrote this massage, but couldn't find it in the forum).
I have a vignette study with vignettes nested (and totally balanced) within participants. I am interested in the 1-1-1 mediation, however because of the balanced structure, I have no between subject variance and cannot use the original Preacher syntax.
I tried to adjust the syntax, and would be grateful for your feedback before I continue to my more complex moderated mediation model.

ANALYSIS:
TYPE = TWOLEVEL RANDOM;
MODEL:
%WITHIN%
MEDIATOR ON IV(aw); !regress mediator on IV call the slope aw
DV ON MEDIATOR (bw); !regress DV on MEDIATOR call the slope bw
MODEL CONSTRAINT:
NEW (indw); !NAME THE INDIRECT EFFECTS
indw= aw*bw;
OUTPUT:TECH1 CINTERVAL;
 Bengt O. Muthen posted on Wednesday, August 03, 2016 - 2:28 pm
I don't understand what you mean by:

"because of the balanced structure, I have no between subject variance"
 Hadar Nesher Shoshan  posted on Wednesday, August 03, 2016 - 11:55 pm
The design was a full factorial 2X2,
I have 2 dummy variables, that were manipulated as 1/0. Every participant read 12 vignettes with the same combination of cues:
IV1-1 IV2-0
IV1-1IV2-0
IV1-1IV2-0
IV1-0 IV2-1
IV1-0 IV2-1
IV1-0 IV2-1
IV1-1 IV2-1
IV1-1 IV2-1
IV1-1 IV2-1
IV1-0 IV2-0
IV1-0 IV2-0
IV1-0 IV2-0

when I tried to fit the syntax of Preacher for 1-1-1 mediation with fixed slopes, I got the error massage that IV1 has no between person variance (which makes sense, because every participant has the exact same pattern of the IV), I hope it is clear now.
Thank you very much again
 Bengt O. Muthen posted on Thursday, August 04, 2016 - 6:46 pm
If the mediator and DV vary across subjects you want to model that variance on Between.
 Hadar Nesher Shoshan  posted on Thursday, August 04, 2016 - 11:12 pm
Thank you Prof. Muthen. Is there any example of how to model only the mediator and the DV with fixed slopes? I couldn't find anything like it in the literature.
 Bengt O. Muthen posted on Friday, August 05, 2016 - 3:03 pm
Fixed slopes are simply obtained by

mediator ON...

DV ON....
 Hadar Nesher Shoshan  posted on Wednesday, September 28, 2016 - 2:05 am
Dear Prof. Muthen,
Thank you very much. I have tried to adjust a moderated mediation syntax for the multilevel use. When I try to model the slopes on the between as you suggested I get an error massage.
*** ERROR in MODEL CONSTRAINT command
A parameter label or the constant 0 must appear on the left-hand side of a MODEL CONSTRAINT statement. Problem with the following:MED =
I would be grateful for your feedback.
The syntax is:

CLUSTER= CASE;
USEVAR= IV MOD MED DV INTER;
WITHIN= IV MOD INTER;
DEFINE:
INTER = IV*MOD;
ANALYSIS:
TYPE = TWOLEVEL RANDOM;
MODEL:
%WITHIN%
DV ON MED (b1);
DV ON IV (cdash);
MED ON IV (a1);
MED ON MOD (a2);
MED ON INTER (a3);

MODEL CONSTRAINT:
NEW(LOW_MOD HIGH_MOD
IND_LOWMOD IND_HIMOD
TOT_LOWMOD TOT_HIMOD);

LOW_MOD = 0;
HIGH_MOD= 1;
IND_LOWMOD = a1*b1;
IND_HIMOD = a1*b1 + a3*b1;
TOT_LOWEE = IND_LOWMOD + cdash;
TOT_HIEE = IND_HIMOD + cdash;

%BETWEEN%
DV ON MED;
 Linda K. Muthen posted on Wednesday, September 28, 2016 - 6:21 am
Try moving the following before MODEL CONSTRAINT.

%BETWEEN%
DV ON MED;
 Hadar Nesher Shoshan  posted on Wednesday, September 28, 2016 - 7:34 am
Dear Prof. Muthen,
Thank you very much! it did solve the error, but now I get another one saying:

*** WARNING in MODEL command
In the MODEL command, the following variable is an x-variable on the BETWEEN
level and a y-variable on the WITHIN level. This variable will be treated
as a y-variable on both levels: MED

Is it wrong to try to model the mediation only on the within (since my IV has no between level variance), and one path of on the between (since my mediator and DV have both variance between & within)
Thank you very much in advance!
 Bengt O. Muthen posted on Wednesday, September 28, 2016 - 12:56 pm
This warning can be ignored.