This post was prompted by an graphic by The Economist examining if the United States had a parliament with Proportional Representation. By examining the platforms and support bases of five major presidential candidates, it was able to divide US politics into five major “parties:”
Based on 2016 final overall primary results, and 2016 final overall congressional election results, we were able to recreate this thought experiment, but with results derived from total primary and general election voting:
In addition, county-by-county primary results could be factored into congressional election results, leading to an accurate image of the level of support for each “party” per congressional district. With this, we were able to approximately model the results if the election had been conducted under Alternative Vote:
The “Conservatives” suffer the most in this arrangement, whilst the “Christian Coalition” gains. This is likely because in almost all cases the “Conservatives” will have less support than the “Christian Coalition”, and thus will most likely lose out and have their excess votes moved towards the Coalition.
Additionally, the Social Democrats lose out at the expense at the Liberals. This is likely because supporters of Sanders are generally sparse outside of his core areas, whilst Clinton has broad support, meaning that in most cases Clinton would lose out from redistribution. Finally, Trump’s party loses out marginally, perhaps due to the fact that he is unlikely to be a second-choice, being such a divisive figure.
In addition, the election can also be simulated with Single Transferable Vote, using the States as districts:
Single Transferable Vote is a “halfway” between Alternative Vote and Proportional Representation: it is largely proportional in large states, while it resembles Alternative Vote in smaller states. Thus, we see elements of both AV and PR in the STV results.
For more specific results for both, including district-by-district results for AV, and state-by-state results for STV, use the spreadsheet below:
In our redistribution of votes, we assumed that:
- Of Sanders voters, the majority (80-85%) would gravitate to Hillary second, but certain others would gravitate towards Trump. This is largely due to the fact that under a split 5-party system, Sanders would not be urged to support Hillary in the general election, meaning that the 90% figure of Sanders-to-Clinton voters would likely be lower.
- Of Clinton voters, the majority would gravitate to Sanders (60-70%), but a substantial minority would vote for Kasich second
- Of Kasich voters, the vast majority would vote for Ted Cruz second, the same would apply for Ted Cruz voters.
- Trump voters would support a mixture of Cruz (50%), Sanders (20%) and Kasich (20%) for their second choices.
- Third parties (Libertarian, Green, etc.) were not a relevant factor. We ran models that included them in redistribution, but they failed to make a significant difference in results. Furthermore, the large number of parties in this situation means that the urge to vote “outside” of traditional party opinion would be less acute.
For questions about methodology, please do not hesitate to contact us.