Measurement vector PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
 Student 09 posted on Wednesday, June 24, 2009 - 7:38 am

my question of understanding refers to ypsilon in formula (1) in the Muthén (1994) SMR article, p. 379.

Here, ypsilon is labelled as a measurement intercept vector.

So, would it be correct that

a) ypsilon is the vector of the randomly varying intercepts of variables measured at the within-level

b) these intercepts represent the (estimated) group means of the variables measured at the within-level

c) but don't represent grand means?

Many thanks
 Bengt O. Muthen posted on Wednesday, June 24, 2009 - 8:56 am
The intercepts (actually "nu's") are randomly varying across the cluster units, that is, on the Between level. Their means are presented on the Between level; these are the grand means. So, for a given item

nu_j = nu + epsilon_j

where nu_j is the intercept conditional on being in cluster j and nu is the overall (grand) mean.
 Student 09 posted on Wednesday, June 24, 2009 - 9:16 am
thanks a lot!
 Student 09 posted on Thursday, June 25, 2009 - 11:56 pm

shouldn't the mplus approach to twolevel modelling also allow to model formative (as opposed to reflective) measurement models within- and between groups ?

If so, could you please suggest any references in this regard ?
 Linda K. Muthen posted on Friday, June 26, 2009 - 10:14 am
Yes, formative measurement models can be specified in a multilevel model. See the Topic 1 handout for how to specify a formative model in Mplus.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message