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.
-
Read our paper:
Schönbrodt, F. D. & Wagenmakers, E.-J. (submitted). Bayes Factor Design Analysis: Planning for compelling evidence. Retrieved from http://ssrn.com/abstract=2722435.
-
Read the additional manual.
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:
- Add a new file with 4 functions:
sample.functionselect.functionBF.test.functionfreq.test.function
Replace "function" with the test name, e.g.: sample.ANOVA etc.
- Add the new "type" (e.g., "ANOVA") to the
print.BFDAfunction inR/1-Simulation.R - Add the new "type" (e.g., "ANOVA") to the
BFDA.sanityCheckfunction inR/10-SanityChecks.R - 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 me know. The preferred workflow would be that you fork the Github project, implement the test, and send me a pull request. If everything works as intended, we can add the test to the main project.
