Can I perform Principal Component Analysis or Factor analysis on data with 0.33 Cronbach’s Alpha from Reliability analysis?
If no or yes, add reference.
Hello Ikenna,
The facetious answer is, of course you can; the data won't protest! A more serious response is that Cronbach's alpha could be low when: (a) you have a multi-dimensional factor structure for your measure (rather than unidimensional); and/or (b) the individual items/variables simply do not work cohesively together (and the measure is in need of modification or improvement). In his paper (Coefficient alpha and the internal structure of tests, 1951, Psychometrika, 16, 297-334, doi:10.1007/bf02310555), Lee Cronbach points out that coefficient alpha can quantify the dominant factor among the items/variables.
So, yes, you may, and probably should, explore the factor structure of the measure. This may convince you that it is or isn't functioning according to the stated purpose (with the population you're using).
Good luck with your work!
Yes you can, if your data is multivariate. However, you need to take care of the coefficient alpha accordingly.
Yes, you can. Carrying out factor analysis has nothing to do with your CA coefficient.
Thank you all, I did apply it to PCA with about 6 components extracted. I notice that the result in FA is not encouraging even after data transformation.