Take the 2-minute tour ×
Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It's 100% free, no registration required.

I am testing a model which involves three variables. I am testing a partial mediation model. In SPSS, when I calculate bivariate correlation among these three variables, the maximum correlation (say between B&C) I get is -.530 . However, using AMOS and testing for discriminant validity, I get the correlation as -.814. This is very high and is creating problems for discriminant validity. Other two correlations between A & B and between C & A remains more or less the same (-.314 and .359 vs -.369 and .429) I am not able to understand why it is happening and what I should be doing to control it.

share|improve this question
1  
The problem is probably because you are estimating 2 different models and perhaps don't realize it, and almost certainly not due to differences in software. I'm not able to comment further without more detail on what you are trying to model. –  D L Dahly Dec 29 '12 at 9:53
    
The correlations between two variables should not change at all, and I am sure that AMOS and SPSS both calculate correlations correctly. Therefore, what you are reporting are not pairwise correlations but something else; but it's unclear what. Can you post your code? –  Peter Flom Dec 29 '12 at 12:31
    
Thank you. I am now able to now pinpoint why it is happening in AMOS. One of my variable (affective commitment) shows a better fit with a two factor solution (although in literature it is mostly used as one factor but two factor observations are reported as well) and when I use it as two factor the correlation with another variable (cynicism) increases to -.814 as reported above. However, if I model it as one factor the correlation between the two decreases to -.469, although the overall model fit reduces significantly. What I should do to take care of this problem? –  RaviM Dec 29 '12 at 14:39
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.