There is pressure to make surveys ‘ecologically valid’, in that they should reflect the environment in which the behavior occurred. We argue this is the case for some, but not for all, survey work.