Two wave 2-1-1 mediation PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Anne Richter posted on Thursday, March 05, 2015 - 1:35 am
Hi,
for my 2-wave 2-1-1 mediation, the following syntax is used:


USEVARIABLES = MT1 MT2 YT1 YT2 XT1;
CLUSTER IS TEAM_1;
..
!within = MT1 MT2
between = XT1;
!DEFINE: CENTER EmpjisT1 EmpjisT2 (GROUPMEAN); Question 2
!DEFINE: CENTER LEjisT1 (GRANDMEAN);
...
MODEL:
%within%
MT1 MT2 YT2;
YT2 ON MT1(b);
YT2 ON Yt1;
MT2 ON Mt1

%BETWEEN%
XT1 MT1 YT1;
MT2 ON XT1(a);

YT2 ON MT1(b);
YT2 ON XT1;

MODEL CONSTRAINT:
NEW(indirekt);
indirekt=a*b;

OUTPUT: CINTERVAL; ! question 3

I have some questions

1.Specifying the within variables:
I would have specified the MT1& MT2 as a within variable (level 1 variable), but get a error message. Is it correct to not specify it as a within only variabe?

2.Centering:
I would have centered MT1 and MT2 around the group means, however i get a error message ( GROUPMEAN centering can be applied to WITHIN variables only. Unable to center variable: MT1). Shall i center it around the grand mean then?

3.CI:
Demanding CI is not the same as bootrapped CI right? is it enough to interprete the indirect effect CI even though we know a*b is not normally distributed ?

Thanks a lot for providing advice! It is greatly appreciated!
Anne
 Bengt O. Muthen posted on Thursday, March 05, 2015 - 10:02 am
1. You mention MT1 and MT2 on the between-level - that contradicts your notion that they are within only variables. Specifying them as within variables means that they are not involved in between-level modeling.

2. First resolve the issue in 1.

3.CI is not the same as bootstrapped CI. If you are concerned about a*b being non-normal you can use Bayes.
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: