Wirt Atmar wrote: >Gareth Russell writes: > > > <snip>
>Gareth's comments do allow me however an opportunity to expand a little bit >on my previous posting. I personally hold David Anderson and Ken Burnham in >very high regard, but I worry that the AIC is being oversold to the ecological >community -- for two different reasons. > >The first is that there is a high degree of arbitariness to formulation of >the AIC. Cost-of-complexity penalties are common in engineering equations, but >the penalty rate isn't something assigned by God or Mother Nature. It depends >more on the whims of the equation's author at the time he first wrote the >equation than anything else. > > > > Things are not quite as bad as this. One thing Burnham and Anderson don't make claear is that AIC is a measure of predictive ability of a model: it finds the model that minimuses the predictive loss. Hence, there is a solid theoretical underpinning for it. The problem, in ecology at least, is that few ecologists are actually interested in building models that will be used for prediction. Instead they're more interested in explanation (i.e. finding out which factors explain the abundance of blue retractable ballpoint pens in their native habitats). The complexity that is needed for the final model will depend on a host of factors (amount of data, how much is already known about the system etc.), and it's difficult to see how to develop a model selection criterion that will take these into account in an easy way (in principle it can be done, by defining a loss function). >And that brings me to my second concern. All models are not equal in their >value to us. The equations of Shannon, Boltzmann, Clausius, Kepler and >Einstein >represent fundamental understandings of the governing rules of the universe. >And in that regard, they represent a deep human understanding, which is of >course the primary goal of science. Indeed, Einstein's E = mc^2 was such a >triumph >because it doesn't even require a scaling factor to relate such previously >disparate qualities as mass and energy. Due to earlier careful measurements, >we >had already gotten the units correct. > >This is a qualitatively different condition than sequentially running through >every conceivable polynomial model, attempting to choose the best solely by >means of some mechanical metric such as the AIC. If that's done, in the end >nothing has been learned, and the question becomes: was it even worth the >effort? >I would have a terrible time calling this scattershot procedure science. > > > I would hope that most of us would agree. At the very least, one should look at a range of plausible models, and find the ones that make substansive sense. Don't ask a statistician to do your thinking for you! Bob -- Bob O'Hara Department of Mathematics and Statistics P.O. Box 68 (Gustaf Hällströmin katu 2b) FIN-00014 University of Helsinki Finland Telephone: +358-9-191 51479 Mobile: +358 50 599 0540 Fax: +358-9-191 51400 WWW: http://www.RNI.Helsinki.FI/~boh/ Journal of Negative Results - EEB: www.jnr-eeb.org
