Robust Bayesian T-Test.
README
Robust Bayesian T-Test (RoBTT)
This package provides an implementation of Bayesian model-averaged t-tests that allows users to draw inference about the presence vs absence of the effect, heterogeneity of variances, and outliers. The RoBTT packages estimates model ensembles of models created as a combination of the competing hypotheses and uses Bayesian model-averaging to combine the models using posterior model probabilities. Users can obtain the model-averaged posterior distributions and inclusion Bayes factors which account for the uncertainty in the data generating process. User can define a wide range of informative priors for all parameters of interest. The package provides convenient functions for summary, visualizations, and fit diagnostics.
See our manuscripts for more information about the methodology:
- Maier et al. (2022) introduces a robust Bayesian t-test that model-averages over normal and t-distributions to account for the uncertainty about potential outliers,
- Godmann et al. (2024) introduces a truncated Bayesian t-test that accounts for outlier exclusion when estimating the models.
We also prepared vignettes that illustrate functionality of the package:
Installation
The release version can be installed from CRAN:
install.packages("RoBTT")
and the development version of the package can be installed from GitHub:
devtools::install_github("FBartos/RoBTT")
References
Godmann, H. R., Bartoš, F., & Wagenmakers, E.-J. (2024). A truncated t-test: Excluding outliers without biasing the Bayes factor.
Maier, M., Bartoš, F., Quintana, D. S., Bergh, D. van den, Marsman, M., Ly, A., & Wagenmakers, E.-J. (2022). Model-averaged Bayesian t-tests. https://doi.org/10.31234/osf.io/d5zwc.