The following is a guest post from Paul-Henri Gurian, professor emeritus in the Department of Political Science at the University of Georgia.
Earlier this year my colleagues and I published a research article on the impact of divisiveness in presidential campaigns. (“National Party Division and Divisive State Primaries in U.S. Presidential Elections, 1948-2012”, Gurian, Burroughs, Atkeson, Cann and Haynes, Political Behavior). Our research measured the impact of national party division on the national popular vote in the general election. We also measured the impact of a divisive state primary on the general election results in that state. We found that, relative to what would otherwise be expected, the impact of a divisive state primary was limited but the impact of the national party division was potentially much larger.
Our research indicated that the impact of a divisive state primary is rarely more than 1% in the general election in that state. Applying the results of this research to the 2016 primaries we found that the primaries in only a few states indicated an advantage to either party of more than 1%. For example, we found that in South Carolina and Texas, where Clinton did quite well in the primaries and Trump did poorly, the expected general election advantage to Clinton was just a shade over 1%. Similarly in West Virginia and Washington state, where Clinton did poorly but Trump did well, the expected advantage to Trump was about 1.2%.
Looking at the potential swing states we found that none of them exceeded a 1% advantage to either candidate. (We were not able to estimate the impact in Colorado.) Our analysis indicates that Clinton would receive an advantage of between .50 and .88% in Virginia, Ohio, Florida, Iowa and Georgia, but less than that in the other swing states. The analysis indicates an advantage for Trump in none of the swing states. This may be partially because Clinton had only one opponent throughout the campaign, while Trump had multiple opponents throughout most of the campaign; thus in most states Clinton's share of the Democratic primary vote was greater than Trump's share of the Republican primary vote.
Because national party division is difficult to measure, we estimated its impact in two different ways. We used aggregate popular vote (1972-2012) as one measure. Similar to our measure of divided state primaries, we compared the proportion of the national aggregate popular vote for Clinton in the Democratic primaries to the aggregate popular vote for Trump in the Republican primaries.
Clinton received the majority (55.2%) of the aggregate Democratic primary vote, while Trump received only a plurality (45.0%) of the Republican primary vote. In other words, among primary voters there was a larger pool of Republican voters who supported candidates other than Trump than the pool of Democratic voters who supported a candidate other than Clinton. Using this measure of national party division, we estimate that Clinton will receive 2.4% more of the national popular vote than would otherwise be expected in the general election.
We also estimated the impact of national party division using an alternative measure, comparing the percent of delegates supporting the two nominees at their conventions (1948-2012). Contrary to the popular vote measure, the convention vote measure indicates that in 2016 the Republican party is more united. Trump received a larger majority (69.8%) of delegate votes at his convention than Clinton did in hers (59.7%). This suggests that Trump will receive 1.2% more of the general election vote than would otherwise be expected.
Part of the explanation for the difference may be the fact that Democratic rules allocate delegates proportionately, while Republican rules usually allocate disproportionally more delegates to the winner of the primary. Clinton received 55.2% of the aggregate popular vote and 59.7% of the convention vote. (The additional votes for Clinton were largely because of super-delegates.) Trump, on the other hand, received only 45.0% of the aggregate popular vote but received 69.8% of the delegate vote at the convention. Again, this is likely because of the “winner-take-more” rules in the Republican party.
The convention vote measure is more “conservative” in the sense that it almost always indicates a smaller advantage to the more unified party than the aggregate popular vote measure does. From 1972 to 2012, the aggregate popular vote has indicated an advantage greater than 6% for one party or the other in 5 of the 11 elections; the convention vote measure has not indicated an advantage of that magnitude in any election during that time period.
Comparing the two measures to subjective observations, the convention vote measure seems to be a better fit. For example, in 2004 President Bush ran unopposed while Senator Kerry won all but three primaries; the convention measure indicates a small advantage for Bush while the popular vote measure indicates a large advantage. Both estimates for 2016 are relatively small compared to other elections since 1972. This is not surprising since both parties are somewhat divided. (Compare, for example, the 1972 and 1984 elections, when the Republicans were very united while the Democrats were severely divided.) To summarize, our research suggests that the impact of national party division will be modest (though it is unclear which party would benefit more), and a small advantage to Clinton in several swing states.
The Electoral College Map (8/3/16)
Filling Nomination Vacancies That Don't Exist
The Electoral College Map (8/2/16)
Follow FHQ on Twitter, Google+ and Facebook or subscribe by Email.