Is there a way to deal with errors such as this?
Thanks to Jonathan Judge and Ben Bolker for their replies. It appears that the short answer to my question is "No." That being so, I will go with the strategy of wrapping the calls in try() and simulating new data if the error occurs. If this risks biasing the simulation results, well, at least I have a good excuse to tell the referees! :-) Perhaps later I will re-run the simulations using mixed_model() from GLMMadaptive and compare the results. Thanks again. cheers, Rolf
Honorary Research Fellow Department of Statistics University of Auckland Phone: +64-9-373-7599 ext. 88276