- Added a function for residual network analysis, doi: 10.1007/s11336-017-9557-x
- Updated citations
- Added the first CFA tutorial, to come: SEM, MIMIC, correlation structure analysis and modifying priors
- Now possible to obtain
data_list
from Stan, useful for modifying specific priors.
- Sort regression results by outcome variable
- Removed equality tests on real numbers in Stan
- Instantiate-forced changes
- Fixed mistake in calculation of factor residual variance for latent regression models, by transforming
simsem::findFactorResidualVar()
into Stan code. - Also removed 0 start values for latent regression coefficients.
- Ensure asymptotic variance of log-correlation matrix is symmetric
- Minor improvement on residual plots
- Fixed bug when
acov_mat
is supplied to ensureacov_mat
is correctly ordered based on lavaan object
- Users can now supply asymptotic variance matrix of correlation matrix as an input
- Added correlation structure analysis via matrix logarithm transformation
- Now using single Stan script for all models
- There are now parameter constraints for latent variable regression models
- Latent regression coefficient parameters are now standardized
- Eliminated the meta-analysis component, see bayesianmasem package
- Using only CmdStan via the instantiate package
- Permits equality constraints and fixed parameters for loadings, residual covariances and error variances.
- Code refactoring
- Log-likelihood now requested with minorbsem call
- So no more Rcpp
- JOSS paper added
- Added simple contribution guidelines in startup message and README.
- Added a Wishart-Wishart method as an alternative to WB-cond.
- Added log-likelihood computations for meta-CFA
- Added control over prior for factor standard deviation parameters
- Fixed error in implementation of RMSEA calculation in meta-analysis (marginal practical effect)
- Using huxtable to print tables instead of kableExtra
- Added dependent-samples MASEM (experimental -- CmdStan only)
- Fixed bug that caused target = "cmdstan" to fail when method = "none"
- RStan implementation now working and with tests
- Simpler installation
- Precursor to CRAN
- Fixed row order of variables printed
- Fixed an undiscovered error when user fits model with single chain
- Added Wishart-based fixed-effects meta-analytic SEM (also permits minor factors)
- First presentation of package
- CFA/SEM with minor factors assumed
- Random-effects meta-analytic permitting minor factor influences
- Added a
NEWS.md
file to track changes to the package.