LAWBL is to provide a variety of models to analyze latent variables based on Bayesian learning. For more information about the package, one can see here or here.
How to use this package in brief
- A design matrix Q is needed for PCFA, GPCFA, or PCIRM, but not necessary for PEFA
- Default setting can be used to minimize input (e.g., burn-in, formal iteration, maximum number of factors)
- To estimate PCFA-LI (when only a few loadings can be specified, e.g., 2 per factor), use m <- pcfa(dat=dat,Q=Q,LD=F)
- To estimate PCFA (with one specified loading per item), use m <- pcfa(dat=dat,Q=Q,LD=T)
- To estimate BREFA or FEFA (i.e., PFEA without partial information), use m <- pefa(dat=dat)
- To summarize basic information after estimation, use summary(m)
- To summarize significant loadings in pattern/Q-matrix format, use summary(m,what=‘qlambda’)
- To summarize factorial eigenvalues, use summary(m,what=‘eigen’)
- To summarize significant LD terms, use summary(m,what=‘offpsx’)
- To plot eigenvalues’ trace, use plot_lawbl(m)
- To plot eigenvalues’ density, use plot_lawbl(m, what=‘density’)
- To plot eigenvalues’ adjusted PSRF, use plot_lawbl(m, what=‘APSR’)
You are also encouraged to visit here for an online reference of all functions.
For examples of how to use the package, see
- For PCFA with continuous data: here
- For GPCFA with categorical and mixed-type data: here
- For PCIRM with dichotomous data and intercept terms: here
- For fully and partially EFA with unknown number of factors, please refer to the pefa() function.
If you would like to contribute an example to this website, please send your .Rmd file to me at email@example.com.