Multilevel path modeling problem PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Tanja Gabriele Baudson posted on Tuesday, December 17, 2013 - 7:21 am
Hello,

I want to predict teacher judgment accuracy from predictors at student (dummy coded) and at classroom level (continuous) and possible interactions. I am thus estimating an intercepts and slopes as outcome model.

Also, I have a variable indicating how well the teacher knows a child. As teachers differ in how well they know the kids of their class (for various reasons), my approach was to group-center it and to integrate it at the individual level:
[...skip usevar ...]
cluster = classno;

within = kennkind sex_kid lang_kid ses1 ses2 ses3;

between = nostudcl aa_berer;

missing = all(-99999);

DEFINE:
CENTER kennkind (GROUPMEAN);

ANALYSIS:
type = twolevel random;

MODEL: %within%
randslp | tjagdstu ON kennkind sex_kid lang_kid ses1 ses2 ses3;

%between%
tjagdstu ON nostudcl aa_berer;
randslp ON nostudcl aa_berer;
tjagdstu WITH randslp;

OUTPUT:
sampstat stdyx;

Have I missed something? (I am especially unsure if I have to estimate random slopes for all independent variables, e.g. randslp1 | tjagdstu ON kennkind; randslp2 | tjagdstu ON sex_kid, etc.).

Why is it that despite an ICC of .08 for the dependent variable, the level 2 predictors do not explain any variance?

Thanks in advance,
Tanja
 Bengt O. Muthen posted on Tuesday, December 17, 2013 - 3:47 pm
You cannot say

MODEL: %within%
randslp | tjagdstu ON kennkind sex_kid lang_kid ses1 ses2 ses3;

but instead you have to define a random slope for each of those 6 predictors separately. You don't have to define random slopes for all of them - in fact, I would explore the need for it for each at a time.

As for your last question, you may have the wrong predictors.
 Tanja Gabriele Baudson posted on Wednesday, December 18, 2013 - 5:35 am
Thanks, Bengt! I'm making progress ;)

Re. the prediction at Level 2, you are probably right. The question is exploratory, and the result makes sense, in a way.

I have now compared the model with random slopes for every predictor (which does not really explain anything) to one with direct prediction only:

ANALYSIS:
type = twolevel random;

MODEL: %within%
tjagdstu ON kennkind sex_kid lang_kid ses1 ses2 ses3;

%between%
tjagdstu ON nostudcl aa_berer;

Does this make sense to you?

Thanks a lot,
Tanja
 Bengt O. Muthen posted on Wednesday, December 18, 2013 - 10:40 am
Looks fine.

You don't need RANDOM in this case.
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: