As the presidential race enters its final ten weeks, campaigns will increasingly focus on turning out their supporters and persuading those who remain uncertain. For better or worse, those who follow polling tend to focus most closely – and often more narrowly than the campaigns themselves – on those who remain in the “undecided” column in poll questions.
SurveyMonkey’s Election Tracking provides a unique vantage point to follow the undecided, as well as those who lean one way or another but remain less than certain about their choice. Our data confirm the most obvious facets of the undecided – they tend to be politically independent with major reservations about Hillary Clinton and Donald Trump – but also adds some critical nuance, including a slightly more Republican than Democratic tilt.
According to a recent review conducted by Politico’s chief polling analyst Steven Shepard, polls show those undecided in the 2016 presidential race to be “primarily independent, younger and strongly aligned against both politicians.” Shepard generalized patterns across multiple polls because “the sample sizes of undecideds in each poll are small.”
The very large samples regularly reached by SurveyMonkey’s Election Tracking face no such limitation. Just this past week, for example, we interviewed 2,410 voters undecided on a two-way choice between Clinton and Trump, 1,313 undecided on a four-way choice that also includes Libertarian Gary Johnson and Green Party candidate Jill Stein, and 1,067 that were undecided on both questions.
Our results generally confirm Shepard’s observations, but add a different wrinkle with respect to age. Contrary to the findings on other polls, younger voters surveyed in our tracking are as able to choose between Clinton and Trump as older voters. In fact, over the past week, voters under the age of 35 were actually slightly less undecided (9 percent) than those older (10 percent). The pattern is similar – 4 percent undecided among 18-35 year-olds, 5 to 7 percent among those older – on the four-way choice that includes third party candidates.
These patterns are not unique to the past week. Younger voters were also slightly more decided in all of the data we collected during August (not shown in the table above).
Why? The answer may be a combination of online interviewing and the unique way SurveyMonkey asks voters to choose. Rather than presenting “undecided” as an explicit option, we instruct respondents at the beginning of the survey to feel free to “skip ahead to the next question” when they “do not have an opinion.” We classify those who skip the vote questions as undecided. The combination – choosing a candidate absent an explicit undecided option and the lack of involvement of a live interviewer – may provide the same sort of nudge toward a decision as casting a real ballot.
Consistent with the higher undecided percentage found by other polls, however, we find younger voters are far more uncertain than their elders about the candidates they lean to supporting. In a follow-up question asked in the first three weeks of August, just 42 percent of 18-to-24-year olds who had chosen between Clinton, Trump, Johnson and Stein were absolutely certain they would support their selected candidate, compared with 68 percent of those over 65. Certainty about the vote choice increases in near linear fashion as we move from the youngest to the oldest age groups.
Like the other public polls, we also have more undecided voters among political independents, particularly among true independents who do not “lean” to either party. On the two-way choice between Clinton and Trump, for example, nearly a third (30 percent) of non-leaning independents are undecided. The undecided percentage falls to single digits among those independents who lean to a party, nearly matching the levels reported by partisans.
As we reported previously, true independents are also the most likely to select a third party candidate when offered. Nearly half of non-leaning independents opt for either Johnson (who leads with 30 percent) or Stein (13 percent) in a four-way contest with Clinton and Trump. Nevertheless, even with these additional choices, the undecided percentage remains highest among non-leaning independents (15 percent), signaling that many of these less politically engaged Americans may not cast a ballot.
The greater uncertainty and indecision among both younger voters and political independents is entirely consistent with the unusually high unfavorable ratings the major party candidates receive from both groups. As we reported previously, most voters who dislike both Clinton and Trump with equal intensity initially identify as independents (66 percent) and nearly half (45 percent) lean to neither party.
Aside from the independent skew of the undecided, these results also reveal a more subtle but consistent pattern: Republicans are slightly more likely to be undecided than Democrats.
Considering the numbers in a different way, the composition of the voters who are undecided confirms this. Republicans and Republican-leaning independents make up a slightly larger share of the undecided than Democrats and Democratic-leaning independents, whether we focus on those who are undecided on the two-way question featuring just Trump and Clinton (29 percent Republican vs. 22 percent Democrat), the 4-way choice that also includes Johnson and Stein (32 vs. 24 percent) or the smaller pool of those undecided on both (30 vs. 21 percent).
Should every undecided voter cast a ballot, these results suggest an opportunity for Trump to narrow Clinton’s lead, if he can coax the Republicans among these voters to support him.
However, both previous research and our data suggest that being completely undecided correlates with non-voting. In our interviews conducted earlier in previous weeks, for example, undecided voters said the chances of their voting in the general election were 50–50 or less far more often (26 percent) than those supporting Clinton (8 percent) or Trump (7 percent).
So the slightly greater Republican skew of the totally undecided may be early signs of lower than usual turnout on the horizon among Republican partisans.