Hello!
I am validating an adapted tool which suposly has 4 factors. The EFA agrees with it, but the CFA show bad fit indices (to make it work I need to free 3 items). Why is this happening? is it correct to fere this much items (scale has 14 items).
CFA typically implies more restrictive assumptions than EFA (i.e., no or very few cross-loadings are typically allowed in a CFA model). If you have cross-loadings that are significantly different from zero (even if small), CFA tests of model fit may detect those as non-zero.
Also, conducting CFA after EFA on the same set of data makes little to no sense. See my Youtube video on this topic here: https://youtu.be/Qzdr1xaoApE
Christian Geiser I know is non-sense, but is an analysis we are doing previously a mediation analysis and I want to ensure this new version of the scale works. Using the previous knowledge, I thought a CFA as a better solution to observe if the scale assess the factors as previous models. Since it does not work, I checked the EFA, where I found this issue.
I will check your video, thank you!
Uday Arun Bhale I added a comment here. At the beginin I was trying a CFA, since it didn't work, I planned to use an EFA... then this occur.
can you specify what is your interference of it didn't work?
Fundamentally EFA is to be done to explore and CAF to confirm the theory.
if possible share the results?
- Badges
- Science topic
- Similar topics
- Correction