Multiple Comparisons: Bonferoni Test: Videos & Practice Problems

Bonferroni should always be used because it is the only method that controls Type I error under all conditions; methods such as Tukey or Holm are only historical curiosities and have no practical advantage.

Bonferroni is preferred when sample sizes vary greatly across groups and when heteroscedasticity is present; Tukey is only valid for one sample and cannot be used in standard ANOVA contexts.

Bonferroni is most powerful when the number of comparisons is extremely large and when effect sizes are tiny; alternatives like Holm increase conservatism and should be avoided for scientific studies.

Bonferroni is straightforward and safe when the number of planned comparisons is small or when strong control of familywise Type I error is required; alternatives like Tukey or Holm are preferable when many comparisons are made (to gain power) or when balanced pairwise confidence intervals are desired.