Also, are the standardized regression weights equivalent to the correlation between a factor and a manifest variable?
I write this question with reference to an example on p138-142 of the following document: ftp://ftp.software.ibm.com/software/analytics/spss/documentation/amos/20.0/en/Manuals/IBM_SPSS_Amos_User_Guide.pdf.
Here are illustrative figures and a table:
Best Answer
Yes, if the factors are uncorrelated. This is also true in exploratory factor analysis - if you do an orthogonal rotation (or if you don't rotate) then you have one loading matrix. If you do an oblique rotation, the factors are correlated, and then you have a pattern matrix (which is correlations) and a structure matrix, which is regressions.
If you think about CFA in terms of regression, it becomes clear why.
When you do regression, the regression weights are given by:
$$ \beta = X^{-1}Y $$ Where X is the correlation matrix of the predictors (if we're doing everything standardized). If the predictors are uncorrelated, then $X^{-1}$ is an identity matrix, and so $\beta = Y$. Same with CFA, if your factors are uncorrelated, then regressions are correlations. If not, then they're not.