0

I plan to extract the latent variables' factor scores after conducting Confirmatory Factor Analysis (CFA). I will use these factor scores as explanatory variables for my next statistical procedure - multinomial logistic regression.

I learned that I can get the factor scores through data imputation (see: https://www-01.ibm.com/support/docview.wss?uid=swg21600240). I was planning to use the Bartlett Method. However, the only options available in AMOS are Regression imputation, Stochastic regression imputation, and Bayesian imputation.

Questions

  1. Is there a way to obtain the factor scores in AMOS by using the Bartlett method? How?
  2. If it is not possible, will the factor scores produced by the different data imputation methods vary significantly among each other?
  3. Can you please provide me with some reading materials regarding this topic?
Preston Botter
  • 2,611
MarBorj
  • 21
  • get the factor scores through data imputation What does that mean? Do you want to say that when supplying the program with original data matrix it could compute factor scores? (That's so). – ttnphns Jun 24 '18 at 09:45
  • @ttnphns After doing CFA, I want to save the factor scores to the original data file or to a new data file. Here's where I learned that I can use imputation to do this: https://www-01.ibm.com/support/docview.wss?uid=swg21600240 – MarBorj Jun 24 '18 at 09:57

1 Answers1

0

Is there a way to obtain the factor scores in AMOS by using the Bartlett method? How?

I'm not sure if Bartlett scores can be obtained using AMOS, although they can be obtained using SPSS. You should have access to SPSS, as AMOS is a SPSS module. Bartlett scores, as well as Regression and Anderson-Rubin scores, can be extracted in SPSS by clicking on the "Scores" option.

If it is not possible, will the factor scores produced by the different data imputation methods vary significantly among each other?

Can you please provide me with some reading materials regarding this topic?

Yes, Bartlett scores are very different from other types of factor scores$^1$. For more information on factor scores in general, I highly recommend reading DiStefano, Zhu, & Mindrila (2019) and this CV post (specifically the answer provided by @ttnphns).

If it is not possible, will the factor scores produced by the different data imputation methods vary significantly among each other?

Yes, factor scores obtained via imputation methods are different. See Merkle (2011) for more information.

$^1$ Briefly, an advantage of Bartlett scores is that they are unbiased. However, this is not always a desired property (e.g., when assigning scores to individuals for diagnostic purposes).

References

DiStefano, C., Zhu, M., & Mindrila, D. (2019). Understanding and using factor scores: Considerations for the applied researcher. Practical assessment, research, and evaluation, 14(1), 20.

Merkle, E. C. (2011). A comparison of imputation methods for Bayesian factor analysis models. Journal of Educational and Behavioral Statistics, 36(2), 257-276.

Preston Botter
  • 2,611