Measuring primary turnout in aggregate analysis

Research exploring participation in primaries by way of aggregate data faces the problem of finding an appropriate denominator on which to calculate turnout. Decisions need to be made on whether, and how, to divide either registered voters or the voting age population into partisan subsets. In order to cope with this problem, numerous methods have been devised to measure turnout. This paper explores the merits and compatibility of four of these measures, determining that the measure based on the normal vote has theoretical and empirical advantages over the others.