I performed a principal axis factor analysis on four items and extracted one factor. I would like to use the factor values of the extracted factor for further analyses. To calculate the factor values I chose the regression method in SPSS. The calculated factor values now have a mean of 0 and both negative and positive values occured, eventhough all variables were positively correlated. Therefore I assumed that they were mean centered by SPSS. Am I right with assuming that?
Thank you for your time!
Kind regards,
Tobias
Thank you for your answer!
How would I then interpret a factor value of 0? Average on each variable? Average of this new factor?
Typically factor scores are standarized post-extaction.
It would not seem feasible to pre-center presumed factor before conducting PCA of any FA.
Hope this helps
Matt
My experience is with spectral data, so it may not apply to your data, but usually the very first thing that happens to spectral data is to mean-center it. From there I believe the individual points in the spectra (the absorbance at each wavelength or frequency) would be weighed by the variance of all data in the training set at that point. I think the goal here is to "amplify" the importance of the points in the spectra that are correlated (either positively or negatively) rather than parts of the spectra that have no significance and reflect only noise. You'll really have to peek under the hood of how that data was processed to know for sure, but I agree with your assumption, that the data was mean-centered.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS