[R] Sample size and stepAIC, step, or AIC

[ripley@stats.ox.ac.uk]

They use AIC as defined by its author, Akaike.  With good reason: `sample
size' is not a well-defined concept in general, and the R functions are
general.

Most people call the adjusted (I hesitate to call it `corrected') version 
`AICC'.  I believe the `clarified' 2002 edition of Burnham & Anderson uses 
AIC_c.


(It's easy to set up problems, e.g, a binomial glm, in which the number of 
observations relevant to each parameter differs radically.  They occur in 
real life too.)

On Tue, 4 Mar 2003, Martin Henry H. Stevens wrote:

> Do any R functions incorporate a sample sample size correction (e.g., 
> Burnham and Anderson 1998).

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595