Multiple testing
Bonferroni correction
Implemented
multiple-testing.bonferroniDefinition
Controls the family-wise error rate by multiplying each p-value by the number of tests.
Formula
p̃ᵢ = min(1, m · pᵢ)
Inputs and outputs
- p_values
- alpha
Returns: adjusted p-values, reject flags
Assumptions
No assumptions beyond valid, aligned inputs of the documented types.
Limitations
- Conservative when tests are positively correlated.
Python API
import evalsuite as es
es.adjust_pvalues(p_values, method="bonferroni")References
- Dunn, O. J. (1961). Multiple comparisons among means. Journal of the American Statistical Association, 56(293), 52–64.