I am a user of LISREL8/PRELIS2. I have a question about using it with correlation matrix. When a correlation data is analysed with WLS method, the hypothesized model must be 'scale-invariant' (the term defined by Cudeck (psychological bulletin V.105 N.2, 1989)) to obtain correct chi-square and standard error? If this condition is not required, where can I find the reference concerned with this issue, which may be internationaly published, so easily accessed to?
Thanks in advance.
Best regards, Choi, Young Sung
======================= PhD Candidate in KAIST 373-1 Gusung-dong Yusung-Gu Taejon, R. of Korea Email : email@example.com