[Roaming through methodology. XXVI. The ecological fallacy and its less well-known counterpart, the atomistic fallacy].

Usually, individuals form the unit of observation and analysis in epidemiological studies, but such is not the case in the ecological study design in which determinants and diseases are related at the group level (school classes, companies, suburbs, countries). Ecological studies are subject to the ecological fallacy when they are used to make inferences on relationships between determinants and diseases at the individual level: the relationship on the group level may not reflect the relationship on the individual level. On the other hand, individual-level studies are subject to the atomistic fallacy, when they are used to make inferences on relationships between determinants and diseases at the group level: a relationship between a determinant and disease on group level may not be exclusively based on the relationship on the individual level.