NEWS

ordbetareg 0.8 (2025-02-03)

add plot_heiss function to examine how scale components change given levels of a factor
add prior_manual option to allow for a complete over-ride of priors in the model (enables cutpoint modeling)
add posterior_epred_ordbeta function 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

removed transformr from imports because of many dependencies
fixed bug if the intercept was specified and centered due to brms code change
fixed bug that prevented the make_stancode option from working. changed that option to return_stancode
fixed pp_check_ordbeta for multivariate DVs

glmmTMB no longer has standard errors from the marginaleffects package. Updated vignette to reflect this.

Removed dependency on faux as faux was taken off CRAN. This also means the sim_ordbeta function cannot generate correlated covariates.
Fixed a bug with the transform option to the slopes function in marginaleffects in the vignette.
marginaleffects now supports standard errors for ordered beta regression models in glmmTMB, so vignette was updated to reflect this.

Allow intercepts to receive separate priors from main effects (permitting zero-ing out intercepts).
Changed log-likelihood calculation to match brms documentation and permit point-wise loo calculation.
Added animated density plots for continuous responses.
Improved plot formatting and added theme and label options.
Updated vignette to include information about glmmTMB as an alternative for estimation and new plot functions.
Added distribution rordbeta and dordbeta functions.
Removed legacy 0 + Intercept function code. Now accepts any kind of formula.

Fixed bug disabling extra priors for phi regression/modeling.
Added pp_check_ordbetareg function for accurate posterior predictive plots for discrete and continuous outcomes.
Updated vignette and documentation.

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).

Added RNG code for the induced_dirichlet prior on cutpoints to allow for the sample_prior option to work with brm, allowing for Bayes factors and other post-estimation tools with the bridgesampling package (issue #1).
Fixed error in terms of correctly accounting for reverse transform of the ordered cutpoint vector.
Made the ldpf function more stable by using logs instead of direct multiplication.
Fixed typos in the vignette.