Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Robin Jeffries <rjeffries@g.ucla.edu> |
To | statalist@hsphsun2.harvard.edu |
Subject | st: GSEM References |
Date | Thu, 27 Mar 2014 08:22:01 -0700 |
Hello, I just started learning the SEM analyses technique recently in an attempt to verify that our data supports the theoretical information-motivation-behavior model. Most literature I've found on testing the validity of this model involve fully continuous data and therefor rely on goodness of fit statistics such as CFI and RMSEA. However, most if not all of my data is categorical. Some fit nicely into latent factors, others do not and/or need to enter the model unmodified and not treated as a continuous measure. I understand why GSEM does not provide the same type of GOF statistics that SEM does, (It's a similar concept between logistic and linear regression). However, I can't seem to find any literature that does a similar type of analysis so I can continue to learn 1) how to interpret the SEM results and 2) how to determine if the model that seemingly fits (no convergence problems, all paths significant) is actually doing a good job. Is there some sort of ROC curve that can be created? I have also read briefly in this listserv archives, that you can treat all variables as continuous just to get the measures of fit? Are there any references to indicate that this is a valid technique? Any suggestions on resources to how to interpret/use/learn GSEM is appreciated. Thanks, -Robin * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/faqs/resources/statalist-faq/ * http://www.ats.ucla.edu/stat/stata/