When siblings matter: Commentary on Golub and Sobin's 'Statistical modeling of litter as a random effect in mixed models to manage "intralitter likeness"'.

ABSTRACT

Responsible data practices include acknowledging and properly dealing with data structures in which the assumption of independence of observations is violated. Golub and Sobin uncover a common statistical faux pas in neurodevelopmental research - inconsistent and inadequate handling of the clustered variability associated with the use of multiple littermates - and propose a simple mixed model solution in which litter is included as a random factor.