matched pair analysis?
On Sun, 20 Nov 2011, Ryan King wrote:
I'm looking to analyze pairs of competing individuals, where the outcomes are success / failure in the match and each individual has predictors that I think of as random effects. It's easy to analyze win/loss vs phenotype*(zi - zj), but that doesn't properly condition on the fact that exactly one of the two has to win. Do any of the R glmm packages have an option to explicitly handle matched 0/1 pairs like that?
Are there rematches and multiple rounds? Have you looked at the BradleyTerry2 package? Cheers, David Duffy.