Super-populations are seldom explicitly defined, but doing so may be a good exercise for avoiding unit of analysis errors. Many orthopedic papers present statistical analyses of joints or cells and of phantom measurements, and errors in analyses of hips, knees, hands, feet, shoulders and elbows appears, according to a recently published systematic review5, to be a surprisingly common problem; of the 142 reviewed papers 42% involved unit of analysis errors. What populations are involved in these studies? What relevance have the reasoning on super-populations on the statistical analysis? Three examples of common study situations will be discussed. First, an experimental unit, and the unit of analysis in the statistical analysis, is usually defined as the smallest amount of experimental material that can be independently assigned toatreatment. Itcanbepatientsinclinical studies,animalsin in vivo studies and cell cultures in in vitro studies6. If independent subjects (or animals) contribute multiple observations to a sample, and within-subject variance is less than between-subject variance, observations are not independent. This violates the independence assumption of random sampling.