Skip to content

Bonferroni simultaneous confidence intervals for multiple regression

4 messages · Cara Gormally, Dieter Menne, John Fox +1 more

Cara Gormally
# 
Hi,

I'm having no luck figuring out how to find Bonferroni simultaneous confidence intervals to obtain a family of estimates in R.  Does anyone know how to do this?
Thank you!
Dieter Menne
# 
Cara Gormally <cgormally <at> plantbio.uga.edu> writes:
confidence intervals to obtain a
Try package multcomp (no Bonferroni, but more powerful alternatives) or 
multtest on Bioconductor (Bonferroni-family, but only p-value).

Dieter Menne
John Fox
# 
Dear Cara,

You could use the confint() function, setting the level argument to 1 -
alpha/length(coefficients(model)), where model is the linear model that
you've fit and alpha is the complement of the level of confidence. If you're
interested only in a subset of say p coefficients, then use 1 - alpha/p.

I hope this helps,
 John

--------------------------------
John Fox
Department of Sociology
McMaster University
Hamilton, Ontario
Canada L8S 4M4
905-525-9140x23604
http://socserv.mcmaster.ca/jfox 
--------------------------------
Spencer Graves
# 
John Fox described one version of the standard Bonferroni 
correction.  I just got 51 hits from "www.r-project.org" -> Search -> "R 
Site Search -> "Bonferroni";  you may wish to review them.  More 
generally, a Google search for "Bonferroni" produced "about 347,000" 
hits, the fifth of which was the NIST / Sematech Engineering Statistics 
Handbook 
(http://www.itl.nist.gov/div898/handbook/prc/section4/prc473.htm), which 
is quite useful for many things. 

      However, if package multcomp or multtest (in Bioconductor) 
suggested by Dieter Menne will work in your application, they would 
probably be better than Bonferroni.  Bonferroni is like using a sledge 
hammer to open a can.  The sledge hammer would work, but a can opener 
would be preferable.  Bonferroni will work when other things won't, so I 
would use something like multcomp or multtest when I could and 
Bonferroni otherwise. 

      hope this helps.  spencer graves
John Fox wrote: