Confirmatory Factor Analysis on a Polychoric Correlation Matrix

[CBF]

I am doing factor analyses on survey items that use a four-point ordinal scale. When I used PROC FACTOR to run my EFA, I was able to run it on the polychoric correlation matrix, which is appropriate for ordinal data. I would like to treat these data as ordinal for the CFA I am running on a second sample. My understanding is that the best way to do this is to use the Asymptotic Distribution-Free (ADF) method. In PROC CALIS, it seems that METHOD=WLS achieves this. However, if I try to use the polychoric correlation matrix as the data set, I get an error that METHOD=WLS will not accept data of TYPE=_CORR_. Does anyone know a way around this problem? I tried stripping the type variable from the matrix data set, and that did not work. 

This is the code that doesn't work. Any help would be greatly appreciated!

PROC CALIS DATA=POLYCHORIC MODIFICATION METHOD=WLS ;
	FACTOR ONE ===> &F1VARS, 
		   TWO ===> &F2VARS,
		   THREE ===> &F3VARS;
RUN;

[sbxkoenk]

Usage Note 22558: Performing factor analysis on binary or ordinal data
https://support.sas.com/kb/22/558.html

Usage Note 22529: Can PROC CALIS analyze categorical data?
https://support.sas.com/kb/22/529.html

SAS/STAT User's Guide
Appendix A: Special SAS Data Sets
https://go.documentation.sas.com/doc/en/pgmsascdc/v_044/statug/statug_appssds_toc.htm

For method=wls (same as method=adf) , you can work with inwgt= data set!

SAS/STAT User's Guide
The CALIS Procedure
https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/statug/statug_calis_details01.htm

Maybe @jiltao can give some more info.

Koen

[sbxkoenk]

Also @CatTruxillo might be able to help you out.

A Gentle Intro to SEM, Part 3: Measuring Latent Variables with Confirmatory Factor Analysis
by @CatTruxillo 
https://communities.sas.com/t5/SAS-Communities-Library/A-Gentle-Intro-to-SEM-Part-3-Measuring-Latent...

Koen