A tacit assumption in laboratory animal research is that animals housed within the same cage or pen are phenotypically more similar than animals from different cages or pens, due to their shared housing environment. This assumption drives experimental design, randomization schemes, and statistical analysis plans, while neglecting social context. Here, we examined whether a domain of social context—social dominance—accounted for more phenotypic variation in mice than cage-identity. First, we determined that cages of mice could be categorized into one of three dominance hierarchies with varying degrees of dominance behavior between cage-mates, and low levels of agonistic behavior in the home-cage. Most groups formed dynamic hierarchies with unclear ranks, contrasting with recent accounts of stable transitive hierarchies in groups of mice. Next, we measured some phenotypic traits, and found that social dominance (i.e. dominance hierarchy type and degree of dominance behavior) consistently accounted for some phenotypic variation in all outcome measures, while cage-identity accounted for phenotypic variation in some measures but virtually no variation in others. These findings highlight the importance of considering biologically relevant factors, such as social dominance, in experimental designs and statistical plans.