I have panel data regarding NPO and their financial structure. I would like to perform an exploratory factor analysis with those data to gather them into 5 factors. How best can I do it if I want 5 factors for the whole data (not depending on years)? Is it ok to perform EFA on the whole data without taking into account the time effect? What if I have missing variables for some observations some years? Thank you
Hi Fanny, Jon's answer is right in that more confirmatory approaches are probably better than exploratory. It'd and old one, but a great one - check out Longitudinal Factor Analysis by Tisak and Meredith in Statistical Methods in Longitudinal Research: Principles and Structuring Change
Volume 1 in Statistical Modeling and Decision Science (1990).
Cheers
mark
Doing confirmatory factor analysis assumes that you already have a 5-factor model that you can specify. If so, then the previous answers will lead you in the right direction.
Alternatively, if you truly do want to do exploratory factor analysis, then you will need to do this once for each separate time period. Otherwise, the matching variables that appear at each time will dominate your results.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS