Alpha level in Bonferroni Multiple Comparisons test

[Ansuyah Magan]

Hi
I would like to know the alpha level that Stata uses when doing a Bonferroni multiple comparisons test. Does it use 0.05 by default?
thanks
Ansuyah

[Marcos Almeida]

Welcome to the Stata Forum / Statalist

According to the Stata Manual,

The Bonferroni adjustment (see Miller [1981]; also see van Belle et al. [2004, 534 – 537]) does
this by (falsely but approximately) asserting that the critical level we should use, a, is the true critical level, α, divided by the number of tests, n; that is, a = α/n. For instance, if we are going to perform six tests, each at the 0.05 significance level, we want to adopt a critical level of 0.05/6 ≈ 0.00833.

Hopefully that helps.

[Carlo Lazzaro]

Ansuyah:
welcome to the list.
As an aside to Marcos' helpful advice please note that, in case of 6 multiple comparisons the statistical significance is at 0.05 (even if the revised critical level is 0.05/6≈ 0.00833).
Put differently, you cannot report that one or more multipe comparisons are statistically significant at 0.00833 level.
For some concerns about Bonferroni's adjustment, please take a look at: https://www.ncbi.nlm.nih.gov/pubmed/9553006.

[Ansuyah Magan]

Thank you Marcos and Carlo for your explanations. The Pubmed link was particularly useful.

Kind regards
Ansuyah