I have developed a scale based on literature review. There are many scales already present on these variables (knowledge management and ICT) and many studies in literature have given various factors as these are multidimensional scales. I have conduct CFA to confirm these factors. Please provide me strong citations based on area of social sciences for this process where we don't need EFA for scale development.
Read Partricia and Rangarjan (2013), A playbook for research methods. EFA is an instrument formulation technique used to find viariables than will then be subjected to a confirmatory analysis using hypothesis to test it's credibility. Ideally, EFA is done before CFA
Dear Amritanshu
any, and I do mean any, textbook on research methods in the social sciences or just in general will tell you when to use EFA and when CFA. Pick and choose.
In anticipation of Holger Steinmetz' answer: If your instrument is theory-based, you may opt for CFA right away.
Best
Marcel
Dear colleague, I agree with the Marcel's commentary.
Your reasons are sufficient to perform only CFA and not EFA.
Best wishes,
Cristiano.
Please refer to my publication on scale development. It is mentioning EFA and CFA on the items extracted through the review of the literature.
Here is the link of the article.
http://ajss.abasyn.edu.pk/admineditor/papers/V11I1-1.pdf
Thanks
Noman Soomro
having taken a quick glance at your settings for the EFA, I do not believe they reflect the current state of the art (e. g. PCA =! EFA, eigenvalue criterion, varimax rotation etc.). Furthermore, your sample size appears to be quite small.
As far as your CFA is concerned, the "ideal values" are on the lower end of what is considered to be a 'good fit'. Personally, I am unaware that the RMSEA is supposed to be between .05 and .08. As far as I know, the lower it is, the better the empirical data reflect the theoretical model.
Lastly, testing for convergent and discriminant validity within your model - using the fit indices - is new to me as well. Typically, convergent validity is achieved by having your scales correlate positively with a scale measuring a similar construct while discriminant validity is achieved by having your scales correlate negatively with a dissimilar construct.
Amritanshu Vyas
I come back to my original recommendation: please consult a textbook on when to choose CFA and how to do it 'properly'.
Best
Marcel
Marcel Grieger can you please refer me some books where i can find the difference between both types of factor analysis techniques
Amritanshu Vyas
there you go (random selection, most of my literature is in German):
Stevens, J. P. (Hg.). (2009). Applied multivariate statistics for the social sciences (5. Ed.). New York, NJ: Routledge.
Hair, J. F., Black, W. C., Babin, B. J. & Anderson, R. E. (2010). Multivariate data analysis: A global perspective (7. Ed.). Upper Saddle River, NJ: Pearson.
Field, A. (2018). Discovering statistics using IBM SPSS statistics (5. Ed.). London: SAGE.
Kline, R. B. (2011). Principles and practice of structural equation modeling (3. Ed.). New York, NJ: Guilford Press.
There is an infinite number of articles on both EFA and CFA that discuss this in greater detail.
Futhermore, there are at least three threads on RG that addresse this issue as well.
Best
Marcel
- Badges
- Science topic
- Similar topics
- Psychology