Very informative. I’m interested in in applying CFA to estimate the possible biogenic/anthropogenic origin of compounds of interest. Do you think this model can be modified to:
- Sum to unit (so they can be interpreted as percentages)
- Be non-negative
- Fix loadings instead of variances?
PS I found your blog post very interesting, especially as they cover a normally obscure topic of Bayesian CFA/SEM. I feel like these techniques could have many more applications in science if they weren’t so confusing at times