This vignette overviews current features of the RoBMA R
package (Bartoš & Maier,
2020). The table is organized by substantive capability. Some
rows point to additional features not (yet) available in the
RoBMA R package but featured in the metafor
package (Viechtbauer,
2010) as reference points.
Green ticks mark available features. Cells marked
limited mean the feature exists but with narrower model
coverage.
| Topic | Feature | brma.norm | brma.glmm | bselmodel | bPET/bPEESE | BMA.norm | BMA.glmm | RoBMA |
|---|---|---|---|---|---|---|---|---|
| Model / Structure | Fixed- and random-effects models | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
Moderation (mods)
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Location-scale models (scale)
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Multilevel models (cluster)
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
| Model averaging | No | No | No | No | ✓ | ✓ | ✓ | |
| Inclusion Bayes factors | No | No | No | No | ✓ | ✓ | ✓ | |
| General multivariate / covariance structures | No | No | No | No | No | No | No | |
| Robust / sandwich inference | No | No | No | No | No | No | No | |
Refit / update
(update())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
| Priors | Default prior distributions | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
Empirical/informed prior distributions
(prior_informed())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Custom prior distributions
(prior(),
prior_factor(),
prior_weightfunction(),
prior_none())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
PET / PEESE prior distributions
(prior_PET(),
prior_PEESE())
|
No | No | No | ✓ | No | No | ✓ | |
Prior-only inspection (only_priors = TRUE)
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Weight-function shape
(wf_cumulative(),
wf_fixed(),
wf_independent())
|
No | No | ✓ | No | No | No | ✓ | |
Factor contrasts
(contr.treatment(),
contr.meandif(),
contr.orthonormal())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
UISD estimation
(estimate_unit_information_sd())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
| Publication Bias | Selection models | No | No | ✓ | No | No | No | ✓ |
| PET / PEESE models | No | No | No | ✓ | No | No | ✓ | |
| Model averaging over bias models | No | No | No | No | No | No | ✓ | |
| Bias-adjusted summaries / predictions | No | No | ✓ | ✓ | No | No | ✓ | |
| Trim-fill / fail-safe N | No | No | No | No | No | No | No | |
| Prediction |
Pooled effect / heterogeneity summaries
(pooled_effect(),
pooled_heterogeneity())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
Heterogeneity decomposition
(summary_heterogeneity(),
tau², I², H²)
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Fitted-value extraction
(fitted())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Prediction for new covariate values
(predict())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Posterior predictive response summaries
(predict())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
True effects / BLUPs / random effects
(true_effects(),
blup(),
ranef())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Estimated marginal means
(marginal_means())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
| Plots |
Posterior plots
(plot())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
Prior plots
(plot_prior())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Marginal means plots
(marginal_means()
+ plot())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Regression plots
(regplot())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Funnel plots
(funnel())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
zplot diagnostics
(as_zplot()/zplot())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Weight-function plots
(plot_weightfunction())
|
No | No | ✓ | No | No | No | ✓ | |
PET-PEESE plots
(plot_pet_peese())
|
No | No | No | ✓ | No | No | ✓ | |
Radial / Galbraith plots
(radial(),
galbraith())
|
✓ | limited | ✓ | ✓ | ✓ | limited | ✓ | |
Forest plots
(forest())
|
No | No | No | No | No | No | No | |
Baujat plots
(baujat())
|
No | No | No | No | No | No | No | |
GOSH plots
(gosh())
|
No | No | No | No | No | No | No | |
L’Abbe plots
(labbe())
|
No | No | No | No | No | No | No | |
| Residuals / Diagnostics |
Raw residuals
(residuals())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
Pearson / standardized residuals
(rstandard())
|
✓ | No | No | ✓ | ✓ | No | limited | |
Studentized residuals (LOO-PIT)
(rstudent())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Q-Q plots
(qqnorm())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Cook’s distances
(cooks.distance())
|
✓ | No | No | ✓ | ✓ | No | No | |
DFBETAS
(dfbetas())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
DFFITS
(dffits())
|
✓ | No | No | ✓ | ✓ | No | No | |
Covariance ratios
(covratio())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Hat values
(hatvalues())
|
✓ | No | No | ✓ | ✓ | No | No | |
Combined influence summary
(influence())
|
✓ | limited | limited | ✓ | ✓ | limited | limited | |
LOO / WAIC diagnostics
(check_loo(),
loo::pareto_k_ids(),
loo::pareto_k_table())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Moderator collinearity diagnostics
(vif())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
| Refit leave-one-out / permutation tests | No | No | No | No | No | No | No | |
| MCMC Diagnostics |
Posterior summaries (Rhat, ESS, MCMC error)
(summary())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
Trace plots
(plot_diagnostic_trace())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Density plots
(plot_diagnostic_density())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Autocorrelation plots
(plot_diagnostic_autocorrelation())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Diagnostic-plot wrapper
(plot_diagnostic())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
| Model Comparison |
Marginal likelihood
(add_marglik(),
logml())
|
✓ | ✓ | ✓ | ✓ | No | No | No |
WAIC/LOO
(add_waic(),
add_loo(),
waic(),
loo())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Bayes factors
(bf(),
bayes_factor(),
post_prob())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
WAIC comparison
(loo_compare(),
loo_weights())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
LOO comparison
(loo_compare(),
loo_weights())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
| AIC / BIC | No | No | No | No | No | No | No | |
| Reporting |
Plain-text interpretation
(interpret())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
Sub-model summary
(summary_models())
|
No | No | No | No | ✓ | ✓ | ✓ | |
Prior inspection
(print_prior())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
| Extraction |
Posterior draw extraction
(as_draws(),
as_draws_array(),
as_draws_df(),
as_draws_list(),
as_draws_matrix(),
as_draws_rvars())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
Coefficients
(coef())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Point-wise log-likelihood
(logLik())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Sample size
(nobs())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Variance-covariance matrix
(vcov())
|
No | No | No | No | No | No | No | |
Credible intervals
(summary(),
summary_heterogeneity(),
pooled_effect(),
pooled_heterogeneity())
|
✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | |
Model weights
(weights())
|
No | No | No | No | No | No | No | |
Simulated responses
(simulate())
|
No | No | No | No | No | No | No |