- The stage 1 analysis in
bmasem_stage_1()
now additionally returns a pooled partial correlation matrix for network analysis.
- Added option to marginalize the random-effects in
type = "RE"
models (longer runtime, larger ESS and ultimately more efficient)
- If fixed-effects, then set no moderators
- Removed equality test on real numbers in Stan
- Some instantiate-forced changes
- Added meta-analytic SEM predictor matrix
- Export model-implied matrix
- Ensure asymptotic variance of log-correlation matrix is symmetric
- Fixed bug in
bmasem_stage_2()
function. Theacov_mat
is now correctly ordered based on lavaan object
- Pooled object can also be analyzed using bayesianmasem package with
bmasem_stage_2()
function
- Now allows for pooling correlation matrices
bmasem_stage_1()
function, pooled matrix can be analyzed in a different software
- Added covariance matrix analysis methods
- Added parameter equality constraints for residual correlations
- Package works and has basic tests
- Added a
NEWS.md
file to track changes to the package.