Re: [ECOLOG-L] AIC, data-dredging, and inappropriate stats

[Eric Rexstad]

Regarding Dr. Robertson's comment regarding the use of fit statistics 
along with information-theoretic model selection metrics. 
Recall that model selection metrics provide a relative measure among 
competing models (i.e., "is model A preferable to model B"), but they do 
not inform about whether the models (selected or not) fit the data.  It 
is plausible that the 'best' model in a candidate set produces rubbish 
predictions.  So some measure of absolute fit (distance between the data 
and the predictions made by a model) is also needed when assessing 
whether the 'best' model is also a 'good' model.

--
Eric Rexstad
Research Unit for Wildlife Population Assessment
Centre for Research into Ecological and Environmental Modelling
University of St. Andrews
St. Andrews Scotland KY16 9LZ
+44 (0)1334 461833
The University of St Andrews is a charity registered in Scotland : No SC013532