I performed an EFA with principal component as extraction method and varimax rotation on 11 items of a scale. The EFA resulted in a single factor. To confirm the model fit CFA was performed, which does not suggest a good model fit. Can I do the following:
Go back to EFA and force the extraction to 2 factors and see if any of the items cross loadings are >0.3 and remove them from the scale. Then rerun the CFA to find model fit
Regards
Please refer to the literature about the dimensionality of the scale i.e one or two and only then you can go for forced factor determination in EFA. you can also improve the results by keeping the communality value at .70.
CFA relies on Maximum Likelihood extraction, rather that PC, and typically uses oblique rotation (correlated factors) rather than Varimax. So try rerunning your EFA with specifications that match CFA.
Thank you everyone for your kind help.
I used Maximum Likelihood as Extraction Method and Promax Rotation. The extracted factors remained same, however, the CFA indicates a PERFECT FIT this time.
Kind Regards
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS