Hello everyone,
If there are Mplus experts here, would you please help clarify my doubts on the output? I am looking at a very simple model, with "POSPROB" as dependent variable and "PERIOD" as independent variable. A 2nd level moderator "RATING". So I used the below syntax:
I get the output as shown in the picture. My doubt is, why Mplus does not show whether the main effect of PERIOD on POSPROB in the within level is significant or not? I tried other ways (e.g., by adding the command in the within level again) but the model could not run. In that case, it is obvious that the 2-level moderation is significant, but I cannot tell whether the main effect is also significant. Any help would be much appreciated! Thanks a lot!
data: file = all data-chainonly.dat;
variable: names = No ID Rating Ratings Rooms
Exp Month PosProb Senti Period;
missing = all(999);
usevar = ID PosProb Period Rating;
Cluster=ID;
WITHIN = Period;
BETWEEN = Rating;
DEFINE: Center Rating (Grandmean);
ANALYSIS: TYPE = TWOLEVEL RANDOM;
ALGORITHM=INTEGRATION
MODEL: %WITHIN%
Slope | PosProb on Period;
%BETWEEN%
PosProb on Rating;
Slope on Rating;
PosProb with slope;
Output: SAMPSTAT STDYX;
The Level 1 effect (slope) is a random slope in your model and the intercept is a random intercept. This means that the Level 1 regression coefficients are not fixed effects but random effects that can vary across Level 2 units. Therefore, you do not get a fixed intercept or slope estimate at Level 1, but only the residual variance estimate.
You are regressing the intercept and slope on your Level 2 predictor in the between part of your model. The Level 2 regressions are both significant indicating that differences in slopes across Level 2 units can in part be "explained" by the Level 2 predictor. In addition, the Level 2 predictor also accounts for differences in the intercept of the Level 1 regression across clusters. To me, it all looks good.
- Badges
- Science topic