One of the empirical laws of United States politics is called the “incumbent rule,” a term coined by pollster Nick Panagakis in a 1989 article, which means that most voters still undecided in late polls go on to vote for the challenger. As hard evidence, Panagakis cites 155 polls (mostly of state-level races), in 82% of which the challenger got most of the votes of the late undecided. Chris Bowers updated Panakis’ work considering polls as recent as 2004, and concluded that in presidential races “undecideds break overwhelmingly – better than 6 to 1 – in favor of the challenger.”
There are also those who claim that the “incumbent rule” has for unexplained reasons recently vanished from U.S. elections. However, such claim is ill-founded, as explained in a longer version of this article.
Although Secretary Hillary Clinton is technically a non-incumbent, the specific circumstances of the 2016 election strongly suggest that she will indeed be perceived as the incumbent by most voters. Not only is she the candidate of the ruling party. She was Secretary of State for the entire first term of the present administration, has the explicit support of the president and those nearest to him, and, moreover, was herself in the White House for eight years as First Lady of another administration. Furthermore, Secretary Clinton is a career politician running against Mr. Trump, who as a businessman who has never held public office, is the ultimate outsider. Hence, if as the evidence indicates, there is an “incumbent rule” there is little doubt that it applies to Secretary Clinton.
A key question is what percentage of voters remain undecided as of this writing, 2 weeks before the election. The answer depends on the specific poll considered. Some pressure hard undecided respondents to indicate toward which candidate they are “leaning,” and then treat “leaners” as fully committed. However, this approach risks crediting to a candidate respondents who are very likely to change their minds before the election. Polls conducted online, do not have this problem, because there is no human to put pressure on the undecided respondent.
Below we shall focus on a traditional and an online poll: the IBD/ITPP Poll – which was judged most accurate both by an analysis focused on the 2012 presidential election, as well as by another analysis that looked at the three most recent presidential elections – and the Google Surveys Election Poll, which was ranked second by the previously mentioned 2012 analysis. While the Google poll is conducted over the Internet, the participants are selected through certain criteria meant to produce a reasonably representative sample of the general population.
The October-24 IBD/ITPP depicts a tied race, with both Secretary Clinton and Mr. Trump at 41%, although Mr. Trump had previously enjoyed a 1-2 % lead (within the margin of error, but persisting over multiple measures). The 2 major third-party candidates earn about 10% combined, leaving about 8% in the other/not-sure category. Since the race, per this poll, is basically tied, it seems clear that whoever of the major candidates wins a greater share of the undecided will win the election. As discussed earlier, per the “incumbent rule” that person is expected to be Mr. Trump. Furthermore, third-party defections could work in his favor, because the Libertarian candidate is drawing appreciably more support (by more than 2-to-1) than the Green party candidate, and Libertarians are ideologically closer to Mr. Trump than to Secretary Clinton.
The October-20 Google survey in principle brings better news for Secretary Clinton. She is leading 38.7 to 33.7 but with a huge 19.2 % undecided. The Libertarian candidate draws 6.5% while “other” draws 1.9% (Google does NOT offer the Green Party candidate as an option, which works against Mr. Trump). If as predicted by Bowers’ analysis for presidential elections “undecideds break better than 6 to 1 in favor of the challenger,” then Mr. Trump would easily overtake Secretary Clinton. But even a closer split would put Mr. Trump over the top. For instance, let us assume for simplicity that all the undecided voters choose one of the 2 major candidates (with minor candidates holding on to 8.4%, which leaves a total of 91.6% for both major candidates). Then a split of the undecided of 12.2-to-7 (roughly 7-to-4 or 1.75-to-1) in favor of Mr. Trump would yield him victory: 33.7+12.2=45.9 for him versus 38.7+7=45.7 for Secretary Clinton.
To conclude, there is strong empirical evidence that supports the “incumbent rule,” which holds that late undecided voters generally “break” strongly for the challenger (by better than a 6-1 margin in presidential elections). Although Secretary Clinton is technically a non-incumbent, for reasons discussed above she in all probability will be perceived as a real incumbent. The incumbent rule combined with two polls judged to be most accurate recently predict defeat for Secretary Clinton.
Dr. Virgilio Rodriguez is a scientist and consultant based in Germany, with a decade of research experience in telecommunication, mostly in European universities, such as RWTH Aachen (Germany), Supelec (France) and Surrey (England). He holds a doctorate degree from New York University, and received postgraduate training in economics at the University of London.