Non-significant r2 in twolevel pathmo... PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Inga BEck posted on Monday, June 30, 2008 - 9:33 am
Hi!
I am a relative beginner of multilevel modeling in mplus and mow modeled a 2-level path model. So no factors, just sum scores plus one context variable. My model model looks like this:

Between:

W--> A_bw --> B_bw

Within:

A_wi --> B_wi

ICCs of the observed variabels range from .05 to .07. According to the indices, the model fits ok.

Interestingly, I got a non-significant R-square on the between-level regarding the A_bw variable (two-tailed p = .18). But A_bw is regressed on W, and this effectis (using stdyx)both moderately large (beta=.45)and significant (z-ratio 2.64, p<.01)

So what does a non-significant R-square on the between-level for A_bw acutally indicate? Does it indicate some source of misspecification in this model - despite the "good" fit of the indices and the theoretically expected estimates?

Thanks for your help,
Inga
 Linda K. Muthen posted on Monday, June 30, 2008 - 1:00 pm
A non-significant R-square indicates that the variance explained is not significantly different from zero.
 Inga BEck posted on Monday, June 30, 2008 - 2:47 pm
yes I see, but this stands in contrast to the significant effect w --> A_bw, don't you think so?

Best,
Inga
 Linda K. Muthen posted on Monday, June 30, 2008 - 2:57 pm
I don't think so. A coefficient can be significantly different from zero and still not explain a significant amount of variance. The tests are not the same.
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: