This page is not about the theory of exploratory factor analysis. I do not care too much about "theory", and I hate reading math formulations in textbooks and papers. What I want to know are how to perform an analysis and how to interpret the results. Specifically, when performing factor analysis, I want to know: which extraction methods to use? how to determine the number of factors to extract? which rotation method to use? how to calculate factor scores for subsequent analysis? how to interpret the output of SPSS? And, most importantly, what are the justifications of making certain decision.
This page is only a summary of ideas I've got so far. Or, I would say it is only a stack of various information. It is not my answers to the above questions. I do not think there are concrete answers to these questions because statistical literatures on these issues are disputed.
Terms and Their Relationships
Is PCA an Exploratory Factor Analysis?
Selection of Extraction Methods
Theoretical differences between PCA and CFA
PCA and CFA are different or not in practice?
Which Rotation Methods to Use?
How to judge if factors are correlated or not?
How To Decide the Number of Factors?
Which Matrix to Interpret?
- Harris (1985a, 1985b) has argued that factors should not only be scored, but also interpreted on the basis of the factor score coefficients rather than the pattern or structure coefficients (Grice, 2001a, p. 78).
- The book in office says only structure matrix is of interests.
- The webpage says both pattern and structure matrix should be interpreted.
Calculation/Estimation of Factor Scores for Subsequent Analysis
- Grice, J. W. (2001a). A comparison of factor scores under conditions of factor obliquity. Psychological Methods, 6, 67-83.