I'm in the middle of developing a psychometric scale and running a few factor analyses here and there.

I already have a theory in mind and the scale was developed with 4 factors. I ran a confirmatory factor analysis with this theory and concept and got relatively good fit (e.g., CFI & TLI > 0.9, SRMR < 0.05, RMSEA < 0.07, etc.). Out of curiosity, I ran an exploratory factor analysis just to see if it'll factor things into the conceptualized factors. Oddly enough, the EFA only showed me 3 factors rather than the 4. I ran a CFA according to what the EFA gave me, just to compare with the conceptualized CFA, but this new CFA with the EFA information showed relatively poor fit (e.g., CFI & TLI < 0.9, SRMR > 0.1, RMSEA > 0.1, etc.).

So my question is, what does it mean when the conceptualized CFA model is showing better measures of model fit than what the EFA model is giving me? Is this at all anything I should be concerned about?

  • $\begingroup$ You should extract four factors in the EFA. Fewer factors will always have a worse fit. $\endgroup$ – Jeremy Miles Dec 11 '19 at 15:09
  • $\begingroup$ Can you show us the screeplot? $\endgroup$ – Erik Ruzek Dec 12 '19 at 15:59
  • 1
    $\begingroup$ I managed to figure it out, it was because there were cross-loaded items that I did not properly cross-load in the CFA that was resulting in the poor fit. $\endgroup$ – ssjjaca Dec 12 '19 at 20:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.