In Amos, instead of using second order factor analysis, I simplified the model by using latent variables (first order) as observed variables. The mean scores are different from data imputation, regression method (from Analyze tab, with CFA of each latent variable). Which one should be used for further analysis?
Hi,
There is no universal advise for this, You can follow whatever give a good results. However, I would recommend to check data normality on the data, if one i.e. either imputed or mean is not normal, ignore it and follow a normal one.
Regards,
Hi,
You should use the imputed scores.
When calculating mean scores, one is implicitely assuming
Especially the latter assumption is quite strong and usually too strong. Actually, these strict assumptions are the reason to do factor analysis in the first place.
So, if you do factor analysis, by any means use the factor scores coming from factor analysis itself.
Sincerely
Thank you, Anwar and Daniel!
As for me, I found imputed scores make more sense than the means. However, when I come up with item parceling in Amos, I can't find any references for using imputed scores/data imputation to parcel/grouping observed variables within one latent variable. For parceling, they use the means or sum only. If you have any works referring imputed score/data imputation for parceling. Please advise. Thank you so much!
Thank everyone for the replies. We are also struggling with the same question.
I appreciate it if you could provide your sources for your suggestions.
One of my contacts @Alireza-Shabani-Shojaei (also available on Research Gate) has been doing this for years, he mentioned it as a standard procedure. So I think Massoud Moslehpour that's it
Though I found this in the context of parcelling containing some references https://shahroodut.ac.ir/fa/download.php?id=1111121196
It's not a commonly accepted method. It scales down your data, and not accepted in well-published journals. That is my Conclusion Fakhra Jabeen .
@Trang Nguyen, could you please explain how was your decision in this regard? I also face the same problem. By Imputed score you mean to say score from AMOS? If I run two-factor CFA then I get two imputed scores, then how could I make it in a single factor? Please provide some references too.
Massoud Moslehpour and Fakhra Jabeen Please help me regarding the imputed score.
Thanks to all.
- Badges
- Science topic
- Similar topics
- Publishing