Marker selection by Akaike information criterion and Bayesian information criterion

We carried out a discriminant analysis with identity by descent (IBD) at ea ch marker as inputs, and the sib pair type (affected-affected versus affect ed-unaffected) as the output. Using simple logistic regression for this dis criminant analysis, we illustrate the importance of comparing models with d ifferent number of parameters. Such model comparisons are best carried out using either the Akaike information criterion (AIC) or the Bayesian informa tion criterion (BIC). When AIC (or BIC) stepwise variable selection was app lied to the German Asthma data set, a group of markers were selected which provide the best fit to the data (assuming an additive effect). Interesting ly, these 25-26 markers were not identical to those with the highest (in ma gnitude) single-locus lod scores. ((C)) 2001 Wiley-Liss, Inc.