Bi-factor EFA questions PreviousNext
Mplus Discussion > Exploratory Factor Analysis >
 Itzik posted on Tuesday, July 29, 2014 - 2:58 pm
Hi all.
I have a couple of questions regarding bi-factor EFA:

1) Does the non-orthogonal bi-geomin rotation allows correlation between the specific factors themselves, or also between the specific factors and the general factor?

2)What is the meaning of high NEGATIVE loadings on the specific factors. Is it some kind of suppression? Should specific factor scores consider these somehow?

3)Are all-high loading on the general factor (0.7 and above) imply a robust unidimensionality? (allowing to use the total score of the scale)

Thanks a lot!
 Bengt O. Muthen posted on Tuesday, July 29, 2014 - 4:38 pm
1) The non-orthogonal rotation focuses on allowing correlations among the specific factors. Correlations with the general factor can also appear but are typically not useful.

2)If all loadings for a specific factor are negative they can simply be changed to positive and the same estimated model is obtained.

3) Perhaps. You may want to look at factor determinacy to see how well estimated factor scores correlate with the true factor.
 Itzik posted on Wednesday, July 30, 2014 - 12:01 am
Thank you for your quick response.
About negative loadings - its actually that not all loading in a factor are negative, but only some of them. How can I interpret it?

 Linda K. Muthen posted on Wednesday, July 30, 2014 - 5:24 am
Factor loadings are regression coefficients. In the factor model, the items are regressed on the factors.You would interpret it as a negative regression coefficient.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message