NEWS
metaBMA 0.6.9 (2023-09-13)
- Updated syntax for Stan arrays (thanks to @andrjohns)
- Dependency on newest version of rstan (>= 2.26.0)
- Updated references for tutorials on meta-analysis with model averaging
- Bug fixes for passing of arguments in plot_forest() (#23 and #24)
metaBMA 0.6.7 (2021-03-17)
- Transformation of effect sizes via transform_es()
- Sensitivity analysis via meta_sensitivity()
- New prior family: gamma distribution
- Scale of default prior for log odds ratio in meta_default() changed using the
scaling factor pi/sqrt(3) = 1.81 instead of 2.00 (thanks to Frantisek Bartos)
metaBMA 0.6.6 (2021-01-08)
- Upgrade to new rstantools folder structure
- New default prior for effect size: Cauchy distribution with scale=0.707 instead of normal distribution with scale=0.3
- New defaults in meta_default() wrapper function
- Faster CRAN checks and tests
metaBMA 0.6.4
- bugfix: argument "ci" for credibility intervals not working
- reference added: Gronau et al. (2020)
metaBMA 0.6.3 (2020-06-02)
- increased stability and precision of model-averaged posterior distribution and estimates (based on density approximations)
metaBMA 0.6.2 (2019-09-16)
- Moderator analysis: rename slope parameters "alpha" to "beta"
- Bugfix for meta_bma(): Only use H0 models for averaging of "d" parameter
- New tests: Scheibehenne (2017)
metaBMA 0.6.0
- new function meta_ordered() for order-constrained study-effects in random-effects meta-analysis
- table with estimates shows convergence statistics (Rhat, n_eff)
- meta_default(): new labels for effect = "d", "r", "z", "logOR"
- minor bugfixes and improvements
metaBMA 0.5.0
- Major refactoring (breaks compatibiltiy with previous versions)
- Possible to provide data frame via argument 'data'
- Removed arguments "d.par" and "tau.par": priors are now defined via d=prior(...), tau=prior(...)
- Possibility to fit random and fixed effects meta-analysis with moderators in stan (with JZS priors)
- Computation of log marginal likelihood with Stan samples and bridge sampling (via logml="stan")
- Improved numerical integration via integrate() [posterior distribution shifted to zero]
metaBMA 0.3.9 (2017-08-04)
- Updated citation for CRAN
- Added examples for meta_bma() and meta_random()
- Minor bug fixes
metaBMA 0.3.8 (2017-07-26)
- Data sets 'power_pose' and 'power_pose_unfamiliar' added
- Data set 'facial_feedback' added
- More informative description file
- Requirements for CRAN
metaBMA 0.3.0
- First stable version
- High-level functions meta_bma() and meta_default() perform model averaging for standard models (fixed, random + H0, H1)
- Plotting functions for averaged/random-effects/fixed-effects meta-analysis via plot_forest() and plot_posterior()
- Meta-analysis models are fitted by meta_fixed() and meta_random()
- Effect estimates of fitted meta-analysis models can be averaged by bma()
- Inclusion Bayes factor are computed by inclusion()
- User-specified and default prior functions are specified via prior() [can be plottet via plot(prior)]