NEWS

---

hmetad 0.2.1

Bug fixes

• Additional arguments in type2_draws functions are now forwarded properly

Minor improvements and fixed

• Re-factored mean_confidence functions to call type2_draws/type2_probabilities functions for consistency

hmetad 0.2.0 (2026-08-21)

New features

• Added allow_negative_values argument to fit_metad, which models M-ratio on the identity (rather than the logarithmic) scale to allow negative values of M-ratio
  • Added ability to simulate with negative M-ratios. Note that the log_M arguments to simulation functions are renamed to reflect this
  • Added functions for empirical quantities (i.e., type1_probabilities, type2_probabilities, joint_probabilities, roc1, roc2, and mean_confidence), allowing for easy comparison with model estimates
  • Added functions for area under type 1 and type 2 ROCs (i.e., auroc1, auroc2, auroc1_draws, and auroc2_draws)
  • Added a by_response argument to roc2_draws and related functions, allowing for type 2 ROCs collapsed across type 1 responses

Minor improvements and fixes

• Added .stimulus, .response, .confidence, and .joint_response arguments to all draws functions, allowing users to specify the names of the corresponding columns
  • Confidence levels now match across the two type 1 responses in roc2_draws

hmetad 0.1.2 (2026-05-15)

Bug fixes

• Resolved error in posterior_predict() and posterior_epred() for models with multilevel effects on a single confidence level
  • cov_matrix() now works for scalar inputs
  • aggregate_metad() now removes rows with NA values prior to aggregation

New features

• metac2_parameters() function streamlines setting priors for confidence criteria

Minor improvements and fixes

• aggregate_metad() and fit_metad() now perform more thorough checks on the number of confidence levels, K
  • aggregate_metad() has increased efficiency

hmetad 0.1.1 (2026-04-20)

Bug fixes

• Fixed bug in linpred_draws_metad/linpred_rvars_metad where meta_c only used first draw

New features

• Added logit option to use Stan's multinomial_logit_lpmf/categorical_logit_lpmf

Minor improvements and fixes

• aggregate_metad() now preserves column types

  • aggregate_metad() and fit_metad() now infer K using the maximum confidence level (instead of the number of unique levels)

  • aggregate_metad() and fit_metad() now have more helpful errors/messages for invalid data arguments

  • Minor updates to package documentation

hmetad 0.1.0 (2026-03-16)

hmetad is now on CRAN!

hmetad 0.0.1