Thanks for that. I sent the citation to the R list so that the poor user might have access to better advice. It would be interesting to test out how this works via a simulation study.