Our Impact


Finding Out What Voters Really Think About Presidental Debates

With the 2020 presidential campaign season underway, a new cycle of debates has been set in motion — complete with perpetual analysis as to what transpired in each. But the accuracy of polls and post-debate surveys are limited by their dependence on voters’ stated responses, which are inaccurate reflections of their true positions.

Once commentators begin declaring winners, voters reinterpret their own perceptions of the debate. Even a survey took immediately after the debate (before the spin begins) is colored by a voter’s perception of who won, as they reframe the debate in their mind.

Therefore, understanding how voters ‘actually’ experience debate is critical because it tells us much more than the debate’s winners and losers. To unveil the “why’s” behind the “what’s,” researchers need a granular, second-by-second analysis of voter response. Only this level of analysis can facilitate the accurate measurement of each argument and exchange.

Historically, the only way to get such data was through dial testing. A select group of ‘undecided’ voters turns a dial during the course of the debate to indicate their agreement/disagreement with a candidate’s statements. Typically, their collective response is represented as a graphic overlay (popularly called the ‘worm’) superimposed over a telecast.

While dial testing has a long tradition, it suffers from numerous flaws. First, measurement is limited to undecided voters since supporters subconsciously game their response by cranking the dial down whenever an opponent speaks and up when their candidate speaks. But elections are seldom won by independent voters alone and ignoring supporters leaves a critical gap in understanding the true dynamics of the debate.

What’s worse, even undecided voters are rarely “fair” referees. If you analyze their response patterns, you will see that many begin cranking their dials before a candidate has even made their argument. In other words, undecided voters are seldom ‘truly’ undecided… they have their leanings… and those leanings influence their dial behavior.

And, at the end of the day, dials are still a stated response; which means that we’re gauging the “rational” journey of voters during the debate. But as we know from recent neuroscience discoveries, our judgments as voters are often shaped by emotional dynamics, of which we often lack self-awareness. Consequently, it is critical to measure both the rational and emotional dimensions of the debate experience.

Fortunately, there is a range of tools today that better measure voter’s second-by-second experience during the debates. These biometric analytical tools afford new opportunities to more accurately measuring voter response on a continuous basis; including a voter’s underlying emotional journey. This represents a significant breakthrough that can dramatically improve post-debate analysis.

Such measures allow voters to experience a more ‘natural’ viewing experience, allowing them to be more absorbed in the debate content without continuously ‘thinking’ about how to respond. Since all voters can be accurately measured, whether undecided or supporter, deeper insights about the larger dynamic become available, including, for example, a comparison of levels of enthusiasm observed by supporters of different candidates. And, where dial testing treats the entire debate as a level playing field, biometrics unveil the ‘true’ dynamic, situating responses within the larger tide of the debate as voters typically become progressively more bored as the debate continues.

MediaScience incorporated such measures in our analyses of debates in the 2012 and 2016 elections, and in both cases, the results provided a trove of insights that couldn’t be accessed by other means.

Consider, for example, the 2012 vice presidential debate between Joe Biden and Paul Ryan. Analysis of dial data reveals little about the overall experience of the debate. Both candidates have numerous key moments throughout. But biometric data (specifically galvanic skin response) demonstrates not only how the audience becomes progressively more bored (situating key moments within a larger context), but, more importantly, how the entire debate has a single pivot – the point at which Biden tells Ryan, “What, now you’re Jack Kennedy.” Until then both groups of supporters evenly tracked in the debate, but after that Democrats were far more energized. With dial data, that key pivot presented as one of many key moments, masking its significance.

presidential debates insights

Biometric analysis during the 2016 election cycle was even more revealing, given the emotional dynamics associated with the Trump candidacy.

Our analysis of voter facial expressions revealed a high level of negative response by Democrats whenever Hillary Clinton spoke. A similar pattern was not replicated among Republicans who generally responded well to Trump throughout the debate. We framed this as a significant ‘enthusiasm gap,’ highlighting Democrat’s discomfort with their candidate.

Undecided voters were surprisingly engaged throughout the debate (an unusual finding) and responded best to policy discussions (where supporters typically became bored). This reflected an undecided voter different from that seen in the 2012 election, where voters were mostly undecided due to apathy. In our 2016 sample, undecided voters were genuinely grappling with whom to vote for and responded well to a discussion that could help influence their decision. On policy discussions, Trump had something of an emotional advantage among such voters; typically responding with strong definitive positions (‘in my first hundred days…’). Clinton, by contrast, was typically evasive or overly technical, resulting in a lower emotional response.

But perhaps the most dramatic difference in the debate experience was across gender lines. A key part of the second presidential debate, for example, discussed the Entertainment Tonight sex tape recordings. For women, this was the most engaging part of the entire debate. Men, however, demonstrated remarkably low response levels.

Content from The Bipartisan Press. All Rights Reserved.