I calculated 4 factors as latent constructs in a Confirmatory Factor Analysis (I use AMOS). Now I am wondering if it is possible to extract some kind of a factor score like I know it from Exploratory Factor Analysis with SPSS to use those as independent variables in a multiple regression analysis.

Can I use the "Data Imputation" Function of AMOS? What are the risks?

THank you a lot in advance

  • $\begingroup$ See CFA factor scores AMOS and IBM's support on this topic. $\endgroup$ – Nick Stauner Jul 31 '14 at 3:20
  • $\begingroup$ thank you. my lecturer said, there would be problems with those calculated scores, because it includes missing values. Does anyone know more about those risks? $\endgroup$ – Anna0Sophie Aug 1 '14 at 6:35
  • $\begingroup$ I have the exact same problem AnnaOSophie, were you able to resolve this issue? If so, could you please share with me your solutions? Thank you! $\endgroup$ – Vianra Oct 26 '14 at 10:14

Factor regression, as this plug-in approach is known, is subject to measurement error bias. In linear regression $y=x'\beta + \epsilon$, if $x$ is measured with error, then the OLS estimates are biased towards zero (which is sometimes called attenuation bias). See Skrondal and Laake (2001) for a thorough study of the factor regression issue.

If you are working in AMOS, I don't see any reason why you shouldn't just expand your CFA to include regression onto your outcome of interest, making this a full structural equation model.

Imputation is out of the picture whatsoever here. I don't know why you are bringing this up.

  • $\begingroup$ Thank you StasK! The references are very useful! Just one more question, would the factor scores obtained from an EFA conducted in SPSS also suffer from attenuation bias as the factor scores obtained in the CFA in AMOS? $\endgroup$ – Vianra Oct 27 '14 at 20:12

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.