The correlations obtained at the end of an EFA in which an oblique rotation was used are different from the correlations that are obtained when the items from each factor are subsequently used to create subscale scores. Although the differences are usually not large, I wonder which set of correlations should be regarded as more valid and subsequently reported - and why. Can anyone shed light on this, please?
I am interested in this because a colleague and I are currently doing some research in which one set of our correlations falls a little above the criterion that I've seen as indicating that factors are correlated (i.e., .80), and the other set of correlations falls a little below that criterion. I'm not sure which way we should bounce in terms of interpretation.
It can vary according to the type of rotation method, while the reliable factor correlation is through an AFC.
Cristian Ramos-Vera. Thanks for your response, but I have to confess that, although I know that AFC can stand for "average frustrated chump", I don't know what an AFC is in this context. Could you explain, please?
At its most basic level, factor analysis is a reduction technique, a method of reducing many measures into fewer measures. The methodology works by placing scales or items that correlate highly with each other on one factor, while placing items that correlate at a low level with each other on different factors. Because one primary reason items correlate highly with one another is that they measure the same construct, factor analysis provides insights as to the common constructs measured by a set of scales or items. Because it helps answer questions about the constructs measured by a set of items, factor analysis is a major method of establishing the internal validity of tests, questionnaires, and other measurements. You can also think of factor analysis as a method of establishing convergent and divergent validity: items that measure the same thing form a factor (converge), whereas items that measure different constructs
Cristian Ramos-Vera. Thanks for getting back, but you haven't really answered my question. I'm aware of what factor analysis is (I use it quite often), but I don't know what AFC stands for. Can you help me with that, please?
The confirmatory factor analysis that allows to confirm the EFA and evaluate the correlations between dimensions and includes of SEM adjustment indexes of the proposed model
Dear Robert Trevethan
Do you mean EFA or CFA??
In CFA, Using the loadings for factor1 you'll figure out the formula for it; ie
Factor 1 = laodingA*A + loadingB*B + loadingC*C
The statistical software should give you the factor structure matrix which is the correlation matrix between the variables and the factors.
From Peter Samuels
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation