Hi,
I ran PCA with Varimax rotation to my related dimensions. While total variance, communalities and KMO and Barlett's test are all within accepted thresholds, and the number of factors generated are consistent with literature, majority of factor loadings generated low values (.1, .009, .2, .6..etc) and sometimes can't find factors variables loading together.
How to deal with this problem even though my variables are taken from literature and I don't want to delete any. Total data set is 245 samples.
Many thanks in advance
Hi Eias Al Humdan ;
I think, you can use principle axis factor for factor extraction method, if your variables doesnt satisfy normality assumptions. If your data is categorical, WLSMV or ULSMV estimation method can be used (In Mplus). The most important thing is rotation method. You must use oblique techniques for related factors.
Good luck.
I had a similar problem. Once you read my paper, you would know how I handled the factor loading and defended it:
Article Korean Patient-Perceived Satisfaction Scale of Community-Bas...
Hope that this can be of help to you.
Hi Claire, I wish I could generate such high and consistent loadings. You've already deleted 12 items for communalities issues. My model is just perfect in terms of all prerequisite tests prior to factor analysis. Even my communalities are all above .6. The low factor loading scores and the fact that almost all converge onto the first component is my problem.
Claire, really thank you for sharing your wonderful work, it's useful in somehow.
Eias
Eias, try exploratory sem techniques. Your scale may has unidimensional but multi-component construct.
Thanks a lot Abdullah for your suggestion. Is exploratory sem available in spss?
Factor (http://psico.fcep.urv.es/utilitats/factor/Download.html) or Mplus software can be used for this purpose. And this pdf can be efficient for you.
https://www.statmodel.com/download/EFACFA810.pdf
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix