On January 6th, Public Policy Polling released its latest poll of the New Hampshire primary and made clear that the most important action in this Trump-filled Republican nominating contest will come from the Granite State. The national (and most state level) polls currently show Donald Trump far out in front, Ted Cruz and Marco Rubio in double digits, and the rest of the pack far below. Meanwhile in Iowa, Cruz and Trump are locked in a battle far ahead of everyone else, with only Rubio potentially lurking but well behind. New Hampshire is a somewhat different story. The PPP survey shows a remarkable breakdown of support that presents an opportunity, but also a challenge, in solving the Trump riddle. And the dynamic that could take Trump down just might be a decades old political science theorem known as Duverger’s law, so long as the collective action problem currently gripping the GOP race is resolved.
Duverger’s law is part of rational choice theory. In its simplest form it goes like this: single-ballot, non-proportional representation, plurality-winner elections favor the two-party system. That is, where voters are allowed to choose only one candidate (single-ballot voting), where the first place finisher retains all the spoils of the election (non-proportional representation), and where the victor is the candidate who receives the most votes on the first and only ballot (plurality-winner elections—in contrast to a majority requirement backed by a runoff), the election process converges toward a two candidate race. In other words, these circumstances mean that first place is all that matters, and a two-party system will result from such incentives. There are two theoretical explanations for why Duverger’s law works in practice. On a macro level, donors and potential candidates must make important investments in their choice of party, of money and career prospects respectively, and thus they are drawn to the parties that give them the best chance of immediate dividends. More importantly for individual elections, Duverger’s law supposes that voters are rational actors who vote to maximize their preferences rather than simply voting for their first choice at all costs.
Such “strategic” voting—as opposed to “sincere” voting—works in the following manner: in an election between candidates A, B, and C, where voter V has the preference A>B>C, she will initially tell pollsters that she supports candidate A; however, if the election draws near and it becomes clear that candidates B and C have significantly more support than A, V will choose not to “waste” her vote and will instead vote for B in order to prevent her least preferred choice. Under this dynamic, even a third party with significant support will nonetheless drop significantly in the actual vote as it becomes clear its candidate cannot win. Indeed, this exact phenomenon has been documented in the strongest third party challenges of the last half century. George Wallace in 1968, John Anderson in 1980, Ross Perot in 1992 and 1996, and Ralph Nader in 2000 all received significantly fewer votes than public opinion polls would lead one to believe. In fact, in 1968 George Wallace experienced an unequal decline from polls to election results depending on location. In the solid South where the Alabama governor was actually the strongest candidate (Wallace in fact won Georgia, Alabama, Mississippi, Arkansas, and Louisiana), a post-election survey found that only 4% of Wallace supporters voted for their second choice between Nixon and Humphrey, while in all other states 17% of Wallace voters told the pollster that they voted strategically.
As you can see, seemingly nothing about Duverger’s law directly relates to American presidential primaries. Yes, primaries are single-ballot elections, but they are only sometimes non-proportional and are plurality-winner contests in theory but often not in practice. Under the Republican Party’s rules for 2016, states may award their delegates in a “winner-take-all” fashion only on or after March 15th, and even after that date they are not obliged to do so. Thus, most states will award their delegates proportionally based on a candidate’s share of the vote. Moreover, while there are no runoffs, it is far from true that primaries are simple plurality-winner elections. Media proclamations of momentum arising from exceeding expectations are often more important in primaries (especially in the early states) than actual delegate counts. In 1992, for instance, Iowa Senator Tom Harkin received little credit for winning his own state’s caucuses, while Bill Clinton was pronounced the “comeback kid” in New Hampshire despite finishing second. Even though Harkin won more delegates, Clinton was better positioned for the nomination. Thus, the incentives for sincere voting in presidential primaries are far higher than in the general election. Voting for a losing candidate may still confer delegates or attention that will aid them down the line. Nevertheless, in this strangest of political seasons, we just might see Duverger’s law flower in the cold New Hampshire winter.
But first, the findings. On the top line of the PPP poll, Trump continues his sizeable, but smaller than the national average, lead at 29%. The remarkable dynamic picked up in the poll, though, is in the placement of candidates two through six. There, PPP finds Rubio at 15%, Chris Christie and John Kasich at 11%, and Jeb Bush and Cruz at 10%—no fewer than five candidates in double digits, four of whom are separated by a mere point (the remaining six candidates total 13% with 2% undecided). These basic findings have been corroborated by the Monmouth, ARG, and NH1/Reach surveys released in the past few days. It is truly a mad scramble in the middle.
What makes the PPP poll so remarkable, however, is its wealth of data contained in the cross tabs. You see, PPP also asked a host of other identifying questions and opinions of its participants, such as what the voter’s gender and ideology are, and whether they support criminal background checks to purchase a gun. PPP then produced tables comparing an answer to one question to the answer to another—for instance, the gender breakdown on the gun issue. Revealingly, one of these cross tabs compares a voter’s favorable/unfavorable opinion of each candidate with whom the voter will ultimately vote for; that is, PPP allows us to see to what degree Ben Carson voters, for example, view Carly Fiorina, Rubio, Bush, etc. favorably.
These candidate support versus candidate favorability cross tabs are extremely illuminating. See if you can spot the pattern. Bush voters’ favorable/unfavorable views of Christie are 50% to 22%. Similarly, Bush supporters place Kasich at 58/7 and Rubio at 51/29. That same metric for Trump, however, is 9% favorable and 87% unfavorable! For Cruz it comes to 28/43. Christie voters view the top contenders this way: Bush 57/27; Kasich 63/17; Rubio 52/21; Cruz 29/38; and Trump 19/73. Kasich and Rubio have generally similar breakdowns to varying degrees. On the other hand, Trump supporters view Cruz favorably 60/24, while Bush and Kasich have very high negatives. 69% of Cruz supporters like Trump and have, for the most part, high levels of disdain for the other four. This split is reflected in the cross tabs of candidate support and candidate second choice support (voters were also asked to reveal their second choice candidate in addition to their top preference). This analysis demonstrates, for instance, that 30% of Christie voters’ second choice is Rubio, while 18% chose Bush. Bush voters prefer Christie for second at a 20% clip, with 18% for Kasich, and 12% for Rubio. Meanwhile, 33% of Cruz supporters’ second choice is Trump, and the converse situation sits at a considerable 37%.
So to summarize, the New Hampshire polling currently reveals a clear frontrunner with double the support of the second place candidate, followed by a cluster of four candidates with meaningful support, all of whose supporters detest that frontrunner and like each of the other candidates in the cluster. The front runner on the other hand, is well liked only by the supporters of a fifth member of the 10-15% group. To be extremely clear, this does not demonstrate that there are two “lanes” of support—the mythical Establishment-outsider split. There are meaningful ideological differences between Bush, Christie, Kasich, and Rubio, and there is no “Establishment” voting bloc. Rubio, for his part, polls much better among Trump and Cruz supporters than those of the other three. What is does make clear, however, is that there is a clear ideological and attitudinal spilt among the top candidates that is manifesting itself in the way voters view them. A lot of people like Trump, and a lot of people dislike him. It just so happens that those who view him unfavorably in New Hampshire are evenly distributed among four different candidates.
This terrain is simply a starker crystallization of the scenario pundits have pointed to for the last few months—that Trump’s rise presents a classic collective action problem. In September, Nate Cohn of the New York Times identified this dynamic: “every group argues that someone else ought to do the work of taking him down.” In October, Perry Bacon of NBC News echoed this sentiment; there is “a strong incentive to have Trump attacked, but not a single major part of the GOP motivated to do it.” Indeed, while Trump is reviled by party leaders and disliked by many non-Trump supporters, it is to no one’s individual electoral advantage that they be the one to take him on. Even more problematic for those who dislike Trump, the unusually large horde of candidates this election cycle has worked to dilute the anti-Trump vote, such that it exists, which, combined with Trump’s strong plurality, has created an incentive for struggling candidates to remain in the race and continue to cause that dilution. Why, after all, 5% might put you only a small surge away from second.
All of which brings us back to Duverger’s law. Now that the tallies have both elevated and balanced four anti-Trump challengers in New Hampshire—a state whose unusual Republican ideological makeup and “open” primary (open, that is, to independents in addition to registered Republicans) often elevate both quirky and moderate candidates (Ron Paul and Jon Huntsman finished second and third, respectively in 2012)—there is the potential for Duverger’s law to decide the outcome, or succumb to that very collective action problem. Consider the ways in which the unique structure of the Republican race conforms to a non-proportional, plurality election. First, voters are potentially less invested in elevating their preferred candidate than they are in stopping Trump. In party primaries, voters usually have favorable views of most candidates, even if they do not ultimately support them. Here, however, New Hampshire supporters of Bush, Kasich, Christie, and to a lesser extent Rubio have overwhelmingly negative views of Trump and Cruz. Meanwhile, Trump is blowing the rest of the field away in every state except Iowa, and the Hawkeye state has him as a clear co-favorite with the also disliked Cruz. Second, the benefits of finishing a strong second or third are likely to be muted in this chaotic and crowded race. For instance, even if Kasich surges and finishes in third with a surprising 17%, he’s just not going to get the usual oxygen from such a result, especially if Trump laps the field, another contestant edges him for second, and the other candidates are bunched closely behind. The only way a losing candidate will get a considerable boost out of New Hampshire is to significantly separate himself from the rest of the field. Third, because of New Hampshire’s more moderate electorate, the state is perceived as the last, best opportunity to mount a challenge to the surging Trump and Cruz.
In this situation, the fact of proportionality is diminished, the perception of plurality importance is enhanced, and the likelihood of strategic voting greatly increases. Duverger’s law suggests that under these circumstances, candidate support should consolidate (although this remains an imperfect environment for true application of Duverger’s law and so support should still be split among more than two candidates. Cruz supporters in New Hampshire, for instance, will continue to have a strong incentive to vote for their candidate even if he is unlikely to win, especially if he secures a victory in Iowa). Christie, Bush, Kasich, and Rubio supporters will have the strong incentive to vote strategically for the candidate among this group that has the best chance of beating, or finishing a strong second to, Trump. A modified approach of Duverger’s theoretical underpinnings thus suggests that as the election draws near, support for one of these candidates will increase at the expense of the others. Strategic voting may thus elevate a candidate currently languishing far behind Trump into a true challenger.
The only hitch in this plan, however, is the remarkable balance between the non-Trump candidates. Again, Duverger’s Law depends upon voter rationality and knowledge. Even assuming that New Hampshire Republicans have sufficient incentive to vote contrary to their sincere preferences and that they are generally aware of the current polling standings (certainly saturated media coverage of the horserace helps), they still have to know which candidate among the Kasich, Christie, Bush, and Rubio crowd is actually ahead. Right now, those four candidates are separated by a mere 5% in the PPP poll and 4.5% in the Real Clear Politics polling average. If in the days leading up to the February 9th primary, there is no greater clarity about who in fact is in second place, Republican voters open to voting strategically will have no mechanism by which to coordinate. There will be no way for them to know which alternative to unify behind. Already the candidates are paying heed to this problem, as the Christie, Rubio, Bush, and Kasich foursome is running stinging attack ads against each other rather than against Trump. Thus, in the weeks ahead, a key question will be whether any of these four can gain an edge in the Granite state polling. If so, the movement to that candidate may be swift and powerful. If not, the Republican faithful may just tear each other apart while Trump uses their collective inaction to his advantage.