Posted by Andrew Douglas on December 23, 2015
Last week, we wrote about data from Public Policy Polling (PPP) that allowed us to simulate a ranked choice voting (RCV) Republican primary in New Hampshire, showing Donald Trump in a neck and neck race when head-to-head with Senator Marco Rubio, despite leading Rubio 28% to 11% in the first choice totals trumpeted by the media.
This week, a new poll from PPP allows us to conduct a similar simulation, but this time with data from Republican voters nationwide. These simulations are possible because PPP releases data on the relationships between voters’ first and second choices, and reports voter choices when the field is limited to different sets of two, three, and four candidates.
As in New Hampshire, the results of our simulation illustrate how Trump’s large lead in most polls is deceptive: if elections were held today, a sizable portion of Republican voters would be unlikely to vote for Trump, but their support is currently divided among other candidates in the crowded field. While Trump may well win the nomination, he will need to keep making inroads with these voters to be successful.
The chart and table below break down the results of our simulated national RCV Republican primary. Again, Trump has a strong lead among voters’ first choices, with nearly double the support of the second place candidate, Senator Ted Cruz. By the third round of tabulation, when the field is reduced to just three candidates (Trump, Cruz, and Rubio), Trump’s lead is still considerable. But, when Rubio is eliminated, the overwhelming majority of his supporters back Cruz, rather than Trump, putting the pair in a dead heat in the final round, with 50.6% supporting Trump, and 49.4% supporting Cruz.
While this process of transfers allows RCV to produce outcomes that are more representative of the will of the majority of voters, the systems used in most American elections mean that a candidate can easily win even when opposed by a majority within the electorate, so long as their opponents’ votes are divided among a number of other candidates. As a result, there is often pressure on struggling candidates to drop out of races like this year’s GOP primary so that similar candidates will inherit their support. RCV would wrest these decisions away from party insiders and their backroom deals, and place them in the hands of voters on Election Day.
The PPP data makes the simulation of an RCV election possible because it includes more information about the full array of voters’ preferences than other polls. Beyond the simulation, the data reveals some other interesting relationships. For example, although support for Jeb Bush is languishing in the single digits, when placed in a head to head competition with Donald Trump, nearly 37% of voters back Bush, to Trump’s 63%. This is not far off the performance of Rubio, who falls to Trump 41% to 59%, despite the fact that Rubio has largely superseded Bush as the viable “establishment” candidate in the race. Unsurprisingly, there is little overlap between supporters of Bush and supporters of Trump: just 5% of those listing Bush as their first choice list Trump as their second choice candidate, while only 7% of Trump’s supporters list Bush as their second choice.