Dear all
there are many methods to extract number of factors in EFA such as factors with eigen values exeed 1, and parallel analysis. my question is if I do both EFA and CFA for a tool, and the EFA showed 9 factors with eigen values more than one which is identical to the original tool, do i need to do parallel analysis? because that will make a different results. I will be appreciate your suggestions and hope if you could help me with resources>
thank you
Rana,
do I understand you correctly that you have two independent samples and that you wish to perform an EFA on the first sample and a CFA on the second sample? Or do you think about performing both analyses on the same sample?
Anhow, if you have a theory and you know which items are supposed to load on which factor, then I would perform just a CFA.
But if you need to perform an EFA, e.g., principal axis analysis, then I would report the results of the Kaiser criterion, the scree plot, and the parallel analysis. As these criteria will point to different solutions, you may argue with your theory and just keep 9 factors.
Please keep in mind that all criteria are subjective, and all have their strengths and weaknesses. Therefore performing a CFA would be much better in my opion.
Regards, Karin
Rana
Using CFA would be the better option if you have a firm theoretical model to test, but if you don't try looking at the resources that I mention below.
If you are performing EFA to identify a latent variable structure, the Kaiser Eigenvalue over one rule has been widely criticised by psychometric theorists for a long time and many advise using more defensible methods. Basing decisions on scree plots have similarly been discouraged.
Horn's parallel analysis (which you mention) and Velicer's minimum average partial technique are often recommended. These two techniques have shown the greatest accuracy with simulated data.
Whilst I agree that you should look at the outcomes from different techniques, it would be best to report a defensible method of selecting the number of factors to extract. Selecting the right method will depend on the exact specifications of your data and the aims of your work. I suggest looking at the work of Ruscio & Roche (2011) and Velicer et al. (2000) before choosing a way to decide on dimensionality. Both are extremely useful resources for considering this issue and I wish you the best of luck in your search.
Regards
Kevin
Ruscio, J., & Roche, B. (2011). Determining the Number of Factors to Retain in an Exploratory Factor Analysis Using Comparison Data of known Factorial Structure. Psychological Assessment, 24(2), 282-292.
Velicer, W. F., Eaton, C. A., & Fava, J. L. (2000). Construct Explication through Factor or Component Analysis: A Review and Evaluation of Alternative Procedures for Determining the Number of Factors or Components. In Goffin, R. D., & Helmed, E. (Eds.), Problems and Solutions in Human Assessment: Honoring Douglas N. Jackson at Seventy (Pp. 41-71). Boston: Kluwer.
Karin and Kevin
Thank you so much for you both, The resources which Kevin recommended seems helpful, I attend to use both CFA and EFA for a translated tool to Arabic. The results of Eigenvalues and scree plot seem fine but I was confused if I need more, so some researches mentioned that they are not enough while another didn't.
- Badges
- Science topic
- Similar topics
- Psychology