BFDA Bayes factor design analysis

Installation

The BFDA package is not on CRAN yet, but you can install the development version from Github:

library(devtools)
install_github("nicebread/BFDA", subdir="package")

For installations on Windows the package requires R version 3.3.1 or higher.

How to use the BFDA package?

1. Read our papers:

• Schönbrodt, F. D. & Wagenmakers, E.-J. (2018). Bayes Factor Design Analysis: Planning for compelling evidence. Psychonomic Bulletin & Review, 25, 128-142. doi:10.3758/s13423-017-1230-y. [PDF][OSF project with reproducible code]
• Stefan, A. M., Gronau, Q. F., Schönbrodt, F. D., & Wagenmakers, E. (2018). A Tutorial on Bayes Factor Design Analysis with Informed Priors. PsyArXiv Preprint

If you use this package to compute and report your design analysis, please cite it as:

• Schönbrodt, F. D. & Stefan, A. M. (2019). BFDA: An R package for Bayes factor design analysis (version 0.5.0). Retrieved from https://github.com/nicebread/BFDA

2. Read the additional manual.

Sequential Design

3. Look at published use cases of BFDA

We compile a (non-exhaustive) list of papers that user BFDA in practice: A list of published examples

Extending the BFDA package

If you want to implement a new test (e.g., a BFDA for regression or ANOVAs): The BFDA package uses a rather modular system for creating new tests. For new tests, you need to:

1. Add a new file with 4 functions:

• sample.function
• select.function
• BF.test.function
• freq.test.function

Replace “function” with the test name, e.g.: sample.ANOVA etc.

2. Add the new “type” (e.g., “ANOVA”) to the print.BFDA function in R/1-Simulation.R
3. Add the new “type” (e.g., “ANOVA”) to the BFDA.sanityCheck function in R/10-SanityChecks.R
4. Do a lot of testing!

Probably it is easiest to take one of the existing test implementations (e.g., R/t.between.R or R/correlation.R) and replace the name and the content of the functions.

If you implement a new test, please let us know. The preferred workflow would be that you fork the Github project, implement the test, do a lot of testing, and send us a pull request. If everything works as intended, we can add the test to the main project.