Factor scores in EFA with categorical... PreviousNext
Mplus Discussion > Exploratory Factor Analysis >
Message/Author
 Lns posted on Friday, January 16, 2015 - 6:03 pm
Can you please tell me the syntax or process to obtain factor scores for an EFA conducted with categorical (or rather, binary) indicators? FSCORES and FSCOEFFICIENT state that they will not run with an EFA.
 Linda K. Muthen posted on Saturday, January 17, 2015 - 11:23 am
If you have only one factor solution, for example, EFA 2 2 or EFA 4 4, you should be able to use the FSCORES option. An alternative is to use ESEM which is the same as EFA. See Example 5.24. It is an EFA if you remove the covariates and direct effect.
 Lns posted on Saturday, January 17, 2015 - 1:54 pm
Thank you for your reply. I am looking at a 2-factor solution. What should I do for that?
 Linda K. Muthen posted on Saturday, January 17, 2015 - 6:43 pm
TYPE = EFA 2 2;
 Lns posted on Saturday, January 17, 2015 - 7:18 pm
Thank you again. I am still receiving an error message I do not understand:
*** WARNING in SAVEDATA command
Factor scores cannot be computed for TYPE=EFA. Note that factor scores are
available with the use of EFA factors (ESEM). Request to save FSCORES is ignored.
*** ERROR in ANALYSIS command
OUTPUT options FSCOEFFICIENT and FSDETERMINACY and SAVEDATA SAVE option
FSCORES are not available when there are no latent variables.

Here is my syntax:
ANALYSIS:

! TYPE = COMPLEX efa 1 3 ;
TYPE = complex efa 2 2 ;
estimator = WLSMV;
ROTATION = geomin;
plot: type = plot2;

OUTPUT:

savedata: file is xyz.dat;
save = fscores;
 Linda K. Muthen posted on Sunday, January 18, 2015 - 6:41 am
Please send your output and license number to support@statmodel.com.
 Kasey Stanton posted on Thursday, January 04, 2018 - 12:27 pm
I am trying to use the following syntax to save factor scores from a bi-factor EFA. Have I got this setup correctly? It runs okay, but when I look at correlations for these factor scores with other variables they don't really make sense, so perhaps I messed something up in my dataset. I just wanted to rule out that my syntax was setup correctly first, however.

ANALYSIS:

ROTATION = BI-GEOMIN(ORTHOGONAL)
ESTIMATOR = WLSMV;

MODEL: fg f1 f2 BY idas1 idas5 idas9 idas10 idas27 idas38 idas48 idas59 idas73 idas82 idas6 idas36 idas39 idas53 idas69 idas77 idas3 idas14 idas23 idas44 idas47 idas63 idas2 idas33 idas81 idas25 idas84 idas96 idas58 idas83 idas98 idas100 idas113 idas64 idas70 idas92 idas94 idas108 (*1);

OUTPUT: STDY;

SAVEDATA: FILE IS rapsfs.dat;
SAVE=FSCORES;
 Bengt O. Muthen posted on Thursday, January 04, 2018 - 4:23 pm
See UG ex 4.7.
 Kasey Stanton posted on Thursday, January 04, 2018 - 5:27 pm
Thank you for the quick response; I have checked out that example. Also, I am noticing that when I run this syntax I still get SRMR instead of WRMR as a fit statistic even though I am using a WLSMV estimator. Did I miss something else here? Thanks again!
 Kasey Stanton posted on Thursday, January 04, 2018 - 5:44 pm
And nevermind; I re-ran it and was able to get the correct statistic. I apologize for the extra post, and thank you again!
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: