ordbetareg v0.8
- add plot_heissfunction to examine how scale components
change given levels of a factor
- add prior_manualoption to allow for a complete
over-ride of priors in the model (enables cutpoint modeling)
- add posterior_epred_ordbetafunction to allow for
predicting the different components of the scale (bottom, middle and
top)
- update vignette with sections on power analysis and modeling
cutpoints
- assorted bug fixes to multivariate/missing data modeling
ordbetareg v0.7.3.1
- removed transformrfrom imports because of many
dependencies
- fixed bug if the intercept was specified and centered due to
brmscode change
- fixed bug that prevented the make_stancodeoption from
working. changed that option toreturn_stancode
- fixed pp_check_ordbetafor multivariate DVs
ordbetareg v0.7.2
- glmmTMBno longer has standard errors from the- marginaleffectspackage. Updated vignette to reflect
this.
ordbetareg v0.7.1
- Removed dependency on fauxasfauxwas
taken off CRAN. This also means thesim_ordbetafunction
cannot generate correlated covariates.
- Fixed a bug with the transformoption to theslopesfunction inmarginaleffectsin the
vignette.
- marginaleffectsnow supports standard errors for
ordered beta regression models in glmmTMB, so vignette was updated to
reflect this.
ordbetareg v0.7.0
- Allow intercepts to receive separate priors from main effects
(permitting zero-ing out intercepts).
- Changed log-likelihood calculation to match brmsdocumentation and permit point-wiseloocalculation.
- Added animated density plots for continuous responses.
- Improved plot formatting and added theme and label options.
- Updated vignette to include information about glmmTMBas an alternative for estimation and new plot functions.
- Added distribution rordbetaanddordbetafunctions.
- Removed legacy 0 + Interceptfunction code. Now accepts
any kind of formula.
ordbetareg v0.5.0
- Fixed bug disabling extra priors for phi regression/modeling.
- Added pp_check_ordbetaregfunction for accurate
posterior predictive plots for discrete and continuous outcomes.
- Updated vignette and documentation.
ordbetareg v0.3.0
- Fixed bugs relating to processing of bounded outcomes.
- Allow for user-specific bounds to be set for normalization of
outcome to 0/1 (issue #2).
- Enable multiple imputation and multivariate responses with
brms.
- Fix documentation issues (issues #4 and #5).
ordbetareg v0.2.1
- Added RNG code for the induced_dirichletprior on
cutpoints to allow for thesample_prioroption to work withbrm, allowing for Bayes factors and other post-estimation
tools with thebridgesamplingpackage (issue #1).
- Fixed error in terms of correctly accounting for reverse transform
of the ordered cutpoint vector.
- Made the ldpffunction more stable by using logs
instead of direct multiplication.
- Fixed typos in the vignette.
ordbetareg v0.2