Cross loadings in measurement instrument PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
Message/Author
 finnigan posted on Thursday, January 21, 2010 - 6:57 am
Hi there

I would like to ask about the impact of cross loadings in a multiple indicator growth model.

An set of measures under a CFA showed a five factor model.If there are cross loadings of items between the factors, what would be the implication of running a multiple indicator growth model using one of the factors. However this factor's items might load on another factor in the CFA,but that factor is not explicity included in the growth model?

For example, 10 items load on one factor , but 3 of those items might crossload on factor two. If one is looking to examine growth in factor one, using a growth model, what implications does the cross loadings have?
 Bengt O. Muthen posted on Thursday, January 21, 2010 - 10:17 am
Because those 3 items load on a second factor, using one factor implies that the residuals of those 3 items are correlated (they correlate beyond a single factor). See Modification Indices. You could include those residual correlations.
 Mi-young Webb posted on Tuesday, June 03, 2014 - 9:12 am
Hi how can I compute unique R-squared in cross-loading? Assume that y1 cross-load on F1 and F2. Mplus report one R-squared value for y1. How can I compute % variance in y1 explained by F1 and % by F2?
 Linda K. Muthen posted on Tuesday, June 03, 2014 - 11:02 am
Each dependent variable gets one R-square. It cannot be divided up because the covariates are correlated.
 Jenna Ward posted on Tuesday, April 12, 2016 - 8:12 pm
Why doesn't the sum of the squared factor loadings add up to the R squared for a cross loading indicator?
 Bengt O. Muthen posted on Wednesday, April 13, 2016 - 12:51 pm
Because the factors are correlated which adds covariance terms to the explained variance.
 Jenna Ward posted on Wednesday, April 13, 2016 - 1:10 pm
Thanks! Is it ok to square the factor loading and interpret that as the percent of variance accounted for by that factor in the cross loading indicator?
 Bengt O. Muthen posted on Wednesday, April 13, 2016 - 4:24 pm
No, only if the factors are uncorrelated (and have variance 1).
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: