Individual participant data meta-analyses should not ignore clustering
Interviewer effects when investigating abuse were not compatible with effect modification but instead with confounding