When Polls Fail: Rethinking the GOP Presidential Race with Ranked Choice Polling

Posted by Nathan Nicholson on July 21, 2015
In a flashback to 2012, Donald Trump is riding a burst of media coverage to the top of the Republican primary polls. But as Princeton's Sam Wang, writing for The New Republic, points out, all Trump's front-runner standing shows is that traditional polling can't tell us very much about what voters really think--especially in such a crowded field.

We know that at this instant, more voters seem to favor Trump than any other candidate. But with over a dozen contenders jockeying for position, that actually doesn't get us very far in assessing the race as a whole. We don't know the breadth of Trump's support: how many Bush, Walker, or Rubio supporters would happily vote for Trump if their favorite candidate couldn't win, and how many wouldn't vote for him no matter what? Wang notes that favorability ratings imply a very low ceiling for Trump, with a base of core support that doesn't have much room to grow--but horserace polls can't tell us that. And, just as importantly, they can't tell us much about what Trump's base of support really looks like. Who would Trump's supporters back if Trump weren't in the race? Could another candidate tap into Trump's base with the right positioning? Does, say, Scott Walker have a big cache of potential supporters whose affinity for Walker is "hidden" beneath a slight preference for Trump?

Without this information, Trump's plurality support is little more than an interesting talking point, at least when it comes to learning who's likely to actually win the race. As Wang reminds us, a string of marginal candidates (remember Herman Cain and Michele Bachmann?) took a turn at the top of the Republican primary polls in 2012, each generating a flash of media attention before slipping into irrelevance. Ultimately, those polls failed to communicate those candidates' key failing--that despite their ability to snatch a slim plurality lead, the majority of primary voters found them unacceptable.

But pollsters could very easily generate a wealth of new information about the primary field--by allowing respondents to rank their choices rather than just report a single favorite. Results of a ranked choice poll would clearly communicate which candidates have broad-based support, which candidates are marginal, and which candidates could hold onto a shot at victory as the field narrows. In particular, candidates who held onto steady second- and third-choice support, regardless of media-driven churn at the very top of the field, would shine through as the real contenders. And analyzing voters' backup choices would provide a trove of information about different ideological factions within the primary electorate and how the candidates appeal to those factions. 

Ranking choices would be easy to implement into polling--it's used to elect actual candidates in many U.S. jurisdictions and around the world, with heralded results.

In a demonstration of how ranked choice polling could reshape how we think about crowded races, Democracy For America conducted an internal poll on the Democratic primary field in November 2014 in which respondents ranked up to three candidates. The results page allows viewers to pick apart these rankings by eliminating candidates and seeing the results change to reflect the new state of the field. Take a look for yourself how much more information this method conveys about the candidates, the voters, and the horserace than a traditional poll.

Beyond polling, there's a strong case for using ranked choice voting (RCV, or instant runoff voting) to actually elect winners in presidential primary contests. Under RCV, voters rank candidates in order of choice, and a winner is determined by repeatedly eliminating the last-place candidate and reassigning votes for the loser to those voters' next-ranked choice until one candidate has a majority of votes. Because voters' "backup" choices count, RCV picks winners with both a strong core of support and proven broad appeal. That's exactly the kind of winner a presidential primary should produce.

Whether Donald Trump is a flash in the pan or a serious contender, his rise is a reminder that we need smarter polls--and, ultimately, smarter voting systems--to tell us where he and his opponents really stand in the minds of voters.
Show Comments
comments powered by Disqus

Join Us Today to Help Create a More Perfect Union