
Message/Author 


Hello, I hope that you can help me. I understand that model fit indices for twolevel random models are tricky because variance varies between subjects. What I would like to know is what this means for the tech3 output? In particular, I am interested in the correlation matrix given in the tech3 output. I can see that if I just put in the elements that I want in my model, without specifying pathways (as shown at the end of my question)the tech3 option gives me a correlation matrix for these. My maths is not great, but my understanding of such a matrix is that it relies on group variance. Given that this can't be reliably estimated, how should the correlation matrix in the tech3 output be understood? ANALYSIS: TYPE=TWOLEVEL RANDOM; MODEL: %WITHIN% S1W ON V; S2W ON A; S3W ON T2; S4W ON TA2; S5AP2 ON V; S6AP2 ON A; S7AP2 ON T2; S8AP2 ON TA2; %BETWEEN% W; AP2; S1; S2; S3; S4; S5; S6; S7; S8; OUTPUT: TECH1 TECH3; Thank you 


Tech3 gives the correlations between the parameter estimates, so that draws on information from all the data. 

Back to top 

