Luo Wenshu posted on Tuesday, April 10, 2012 - 7:08 pm
Dear Prof. Muthen,
I are running two level path analysis. The model has a very good fit. However, when I compare the path or correlation coefficients, I see big differences between Standardized (STDYX) and non-standardized solutions in the correlation coefficients as well as their signficance levels. why does this happen? Which solution should I choose?
Raw and standardized coefficients have different sampling distributions. Because of this, their significances can differ. Raw and standardized coefficients are in a different metric. This is why their sizes vary. Which you should report depends on what is common practice in your field.
Luo Wenshu posted on Tuesday, April 17, 2012 - 8:20 pm
Thank you very much, Linda.Could you answer me a follow-up question? I am running path analysis (two level). does the standardized solution (STDYX) is the result of estimation by using correlation matrix for both within and between levels, rather than covariance matrix?