I think to avoid writing
Don’t give up!
Dealing with this now and it’s so frustrating. One reviewer said nice things about our theory/method but recommended rejection because “it’s unclear how this would be implemented in practical application.”
I’ll be leading a 4-day online (synchronous) seminar on SEM with categorical data June 18-21. Please share this announcement with anyone who may be interested.
This online statistical training with Wes Bonifay presents a comprehensive treatment of structural equation modeling (SEM) for binary and ordinal (categorical) outcomes.
At last week's NCME conference, I did a workshop on Bayesian latent variable modeling with a focus on the blavaan package. We discussed factor analysis, item response models, and two-level structural equation models. Slides and code are at the first bullet:
Reposting this now that the official version is online: https://doi.org/10.1177/10731911241234118
If you’re willing to exchange “confidence” for “credible,” I believe @winterstat.bsky.social has some Bayesian IRT examples from a recent paper
See also Freeman Dyson: “It is better to be wrong than to be vague.”
Good catch! We used the Beta[-1,1] prior, described in Merkle & Rosseel (2018) as "a beta distribution with support on (−1, 1) instead of the usual (0, 1)": www.jstatsoft.org/article/down...
2) Comparing to MI testing: “In fact, and in contrast to measurement invariance testing and related techniques … our similarity checking procedure is not tied to CFA or SEM at all, and can shed light on replication efforts involving any other model that can be estimated using Bayes’ theorem.”