Party-list Proportional Representation and the Single Transferable Vote are Not Functionally Equivalent

Do party-list proportional representation (PR) and the single transferable vote (STV) produce different results? These reforms are hotly debated in American politics, but most answers to the question come from non-U.S. cases. This paper replicates and extends the best existing analysis of U.S. data. Gosnell (1939) simulated list PR elections from observed STV first choices in Cincinnati (1925- 37) and New York City (1937), finding that majority control of government did not change across the systems. Later work took this to mean that both kinds of PR were equivalent with respect to preference aggregation. Reappraisal and extension of the study do not support that view. All else equal, voters’ lower rankings changed majority control in five of 21 elections, one of which was in the original sample. These results offer fresh perspective on ballot exhaustion and ticket-splitting, research into which otherwise often frustrates reform advocates.