MacKuen, Erikson, and Stimson’s classic article “Macropartisanship” extended the study of voting behavior from static analyses of American elections to the dynamics of partisanship between elections. This launched new frontiers of research, such as studying the effects of
presidential approval and economic indices on aggregate party identification. However, the Macropartisanship literature made an important oversight: Changes in partisanship between elections are usually from one partisan group to identification as an independent, or vice
versa. These are not the same switches a voter might make in successive visits to the voting booth, where the main question is: Democrat or Republican? A single measure of aggregate partisanship, like the original Macropartisanship, leaves out independents altogether. This has
important theoretical and empirical consequences that are increasingly evident in the era of polarization. Researchers focusing on aggregate partisanship require data and an empirical approach that allow going beyond theories like Key’s “switchers” and “stand-patters” models
of individuals’ vote choice. We conceive of Macropartisanship as a compositional variable and study how its components are affected by changes in economic sentiment and presidential approval.