| Message/Author |
|
|
|
|
| I am very new with Mplus. I am running EFA(Factor Analysis) and somehow confuse with which output is having the factor loadings. It is GEOMIN ROTATED LOADINGS or the FACTOR STRUCTURE. Mplus vr6.12 |
|
|
|
|
| The factor loadings are in Geomin Rotated Loadings. |
|
|
|
|
Hi Prof Linda,
How do i estimate/calculate the factor scores for each of the factors Mplus or the summated scales for each factor. Also how do i calculate the reliability of each factor. |
|
|
|
|
You cannot get factor scores with TYPE=EFA. You would need to run the EFA via ESEM. Use Example 5.24 without covariates and the direct effect. This is the same as EFA.
For continuous factor indicators, you can look at factor score determinacy to assess reliability. |
|
|
|
|
Hi Prof Linda,
Thankx for your response. In ESEM with categorical(ORDINAL SCALE) how can i get or access the reliability. my scale is ordinal on 5 likert scale. The determinacy function is not working for categorical scale. Is there any means to get the reliabilities.
secondly should i read the factor scores from STDYX Standardization or MODEL RESULTS for categorical. Also why SRMR is not reported for categorical in ESEM.I have a very large sample size so Chi-Square model text is not good for me. I then limited to few model test. |
|
|
|
|
You can look at the information function given with the PLOT command. Also, see the IRT literature.
Factor scores are based on the unstandardized results. Use CFI and the other fits statistics. |
|
|
|
|
Drs. Muthen,
I am doing an EFA with binary data and have obtained a loading greater than 1 for one of the ten indicator variables. I checked the residual variance and it is positive. I reviewed the Joreskog paper about loadings greater than 1, but he specifies standardized loadings. I assume that EFA reports unstandardized loadings because in the CFA, my standardized loadings are all less than 1. An earlier post by Linda on the discussion board said that loadings over 1 were possible for categorical factor analyses because the loadings refer to regression coefficients rather than correlations. Is there a reference in an article or book that I could use for that statement?
Thanks,
Mary Mitchell |
|
|
|
|
| See our FAQ on Standardized >1. There are 2 papers mentioned there which you can refer to. It happens when the predictors, in this case the factors, correlate highly. Standardized or not doesn't matter. |
|
|
|
|
I have a quick question about the factor loadings shown in the output for a ESEM. I'm getting the MODEL RESULTS but I'm not sure if the estimates are the rotated factor loadings or not. The rotation method should be GEOMIN but this was not mentioned in the results.
I used imputed data sets with categorical variables and a cluster variable.
So part of my syntax is
ANALYSIS: TYPE = COMPLEX;
ESTIMATOR = WLSMV;
MODEL: f1-f10 BY y1-y52 (*1);
Are the estimates in the model results rotated factor loadings?
Thanks |
|
|
|
|
| Please send your full output to Mplus Support along with your license number. |
|
|
| Back to top |
