Impact of small group size on neighbourhood influences in multilevel models

Abstract

Datasets in which there is a small to moderate number of groups, with the majority of very small group size (n<5), size may fail to find or even consider a group-level effect when one may exist and also may be underpowered to detect fixed effects.