Amos can imputated factor scores, and it also provided ''factor score weights'', but when I use those weights to calculate the factor scores, I find that is not same as the imputated one.Why?
Also, the basic pre-condition of the Regression Method to estimate factor scores is requiring all the variables and factor values are standardized, but the final estimations of factor values' variances are far from 1.
Hello xu,
There is nothing that restricts computed factor scores using the regression method from yielding a variance (across cases) different from 1 in AMOS. Have a look at example 6 from Arbuckle's manual for AMOS: https://www.ibm.com/docs/en/SSLVMB_26.0.0/pdf/amos/IBM_SPSS_Amos_User_Guide.pdf (p. 126).
Good luck with your work.
David Morse Thanks a lot. Yes, the calculation of factor scores using the regression method can not promise the variance is near 1, not only in Amos but in all the software if do not make any constraints. I think it is a deficiency in the regression method since the formula of this method is based on the pre-condition that requires all the observed variables and factors to be standardized.
Also, could you please tell me your opinion on my first question? That really puzzled me for a long time.
Best Wishes
Thanks, everyone. I got the answers. The answer to the first question is Amos just centralized the variables, not standardized them, before weighting the sum of those variables. The answer to the second one is that since the regression model does not consider the error term, the variance of the estimated latent variable is not exactly 1.
- Badges
- Science topic