Methodology

In this report, we review existing evidence about fusion voting and its effects on political outcomes and provide an original analysis of how fusion has been used in New York and Connecticut. The limited use of fusion voting in contemporary elections makes it challenging to systematically study fusion voting and its impacts. While there are many historical accounts of fusion voting and minor parties in American politics, there is limited social science research on fusion voting, and we hope this report inspires more research interest in the subject.

In this report, we assess the core claims about fusion voting based on the limited available evidence and clarify the arguments of how fusion voting affects various outcomes of interest. We rely on academic research and commentary, our own quantitative analysis, and interviews with practitioners. The data used in the report comes from a variety of sources. Data on congressional elections comes from the MIT Election Lab, and data for state-level legislative races was collected from each state’s board of elections. Turnout data comes from the voting age population in each congressional district, obtained from the National Historical Geographic Information System and the U.S. Census Bureau.

We conducted 10 online interviews, focusing on the four states where fusion—disaggregated (full fusion) and aggregated (partial fusion, or dual-labeling)—is currently practiced: New York, Connecticut, Vermont, and Oregon. We interviewed expert practitioners across three areas of the political ecosystem: (1) third-party organizers, (2) affiliated interest group leaders, and (3) third-party elected officials. Interviewees were found through a combination of desk research, personal reference, and snowballing. One result of these limitations is that the interviewee pool over-represents members of the Working Families Party, self-identified progressives, and New Yorkers. However, our findings indicate many promising areas for future research, including expanding the study size with a larger sample and greater diversification.