Glossary of economics research
Results of search for Bonferroni criterion follow:
Bonferroni criterion:
Suppose a certain treatment of a patient has no effect. If one runs a test of
statistical significance on enough randomly selected subsets of the patient
base, one would find some subsets in which statistically significant
differences were apparently distinguished by the treatment.
The Bonferroni criterion is a redefinition of the statistical signficance
criterion for the testing of many subgroups: e.g. if there are five subgroups
and one of them shows an effect of the treatment at the .01 significance
level, the overall finding is significant at the .05 level.
This is discussed in more detail (and probably more correctly) in Bland and
Altman (1995) in the statistics notes of the British Medical Journal. Either
of these links should go there:
Llink 1.
Link 2; search for Bonferroni.
Source: British Medical Journal, statistics notes by Bland and Altman.
Contexts: statistics; epidemiology
Back to top