Proportional representation with districts is the most prevalent electoral system in the democratic world. Under this system, within the same country, districts often substantially vary in their number of seats (magnitude). We show that under these circumstances, how the legislature is divided to districts, how legislators are assigned to districts, and in the presence of malapportionment, how votes are channeled through districts, lead to different characterizations of a single electoral system. We propose that the unit of analysis based on which an electoral system is characterized – districts, representatives, or voters – ought to be selected specifically to fit the research question at hand. We empirically demonstrate the difference among these dimensions on a cross-section of districted countries and explore its implications for inference.