Polling the subconscious — Election summary

The polls got it wrong. Even the most famous and generally accurate ‘poll of pollsters’ got it wrong. Instead of a Clinton landslide, we witnessed a Trump victory. And the pitchforks are out.

Behind the pollsters is a growing firing line. Here’s a small sample to which and to whom blame is being assigned:

More will surely join the queue as the dust continues to settle.

The content that gets pinged around the echo chambers of the networked public generally follow two rules: it’s simple and emotional. The truth is far more complex.

Many Blacks, Latinos, gays, Jews, Muslims and other religious and ethnic minorities — men, women and maybe even ‘other’ (though I’ve yet to find data on this) — voted for Donald Trump. So did people of varying classes and levels of education. But in the face of complex problems, we seek simple answers.

That’s the overarching lesson we should be taking away from the election. We’ve allowed our ‘System One’ — the automatic, the emotional-based — to take over decision-making, when our System Two — the deliberate, the considered — should really have been involved in choosing who will govern one of the most complex systems in the world, the United States.

Brainsights can say this because it’s analyzed System One — the non-conscious brain — using neuro-technology as voters watched a range of political messaging, including the final presidential debate.

System One is largely responsible for decision-making, and what we’ve found helps to explain the discrepancy of the polls.

By conducting this analysis, and developing this tool, our aim is to encourage public opinion leaders, the media and all voters to consider System One; to be aware of their non-conscious biases when making decisions, particularly decisions as important as selecting the next President of the United States.

Measurement paradigms must evolve to account for our growing understanding of the role of emotions in making decisions. What has become glaringly obvious is that polling, which relies on explicit, stated response — the realm of System Two — is incomplete. Researchers know this, yet have persisted with traditional methodologies. Hopefully this latest embarrassment is a panacea to a measurement evolution.

Three reasons the final debate was ideal political communication for analysis

  • Breaking the echo chamber. The live nature of debates offers opportunities for voters to receive unfiltered information from candidates. It’s in the moment, it’s not fake news and it’s not targeted advertising. The same content is seen by both groups of supporters (and others). It’s an opportunity for candidates to reach out to ‘the other side’.

  • Convincing the undecideds. The third and final debate was likely the last big piece of political communication that most voters would view (indeed, more than 70 million people tuned in). And there may have been a lot of people (one poll says 13%) still unsure of who they’d vote for. It’s the final big opportunity for the candidates to pitch for the Presidency and command a large audience. The stakes were incredibly high.

  • A large dataset. A 90-minute piece of content provides us with a huge data set to understand patterns of non-conscious influence and activation. The candidates are side by side, the issues are (mostly) being handled one by one, and we can run statistical analysis to determine how persuasive specific messages are, as well as the persuasiveness of candidates and their policies.**

There are three major observations we have about the election and the issues summarized below. Two of them — the allure of strongmen and the attraction of the ‘rigged’ message — are drawn from our own data. The other — the corrosiveness of echo chambers and the complicity of the media — contains broader observations grounded in Brainsights’ deep understanding of media, gained through its research and its own experience. This helps to frame up the other two findings.

What happened?

The American Strongman

First, liberal democracies are not immune to the appeals of divisive and authoritarian strongmen. This may seem like a statement of the obvious: of course this is the case if a major political party nominee has built robust support based on a campaign rich in racism, sexism, xenophobia and deep distrust in economic, social and political institutions.

But while the commentary throughout the campaign largely focused on the apparently fringe segments of American society that Trump successfully galvanized with his divisive messaging, our data shows that even self-described liberal democrats found some of Trump’s appeal irresistible. Irresistible in the sense that our unconscious minds are drawn to his claims, even while our conscious minds reject it. It’s like some kind of hypnosis has taken control.

The data is pretty clear on this point: Trump was 22% more persuasive on his own, when he had solo time on screen speaking, than when he shared the screen with Clinton (See box to understand our definition of ‘persuasive’). A raft of his messages were also much more persuasive (see below). Yes, there was large-scale tune out when Trump rambled on incoherently, as he has in several moments throughout the debates. But when his message was more focused — and especially when it infused a sense of humanity (occasionally, it did) — he was significantly more persuasive than Clinton.

Measuring persuasion

Brainsights’ definition of ‘persuasive’ is a combination of attention, emotional connection, and encoding to memory as recorded in brain wave activity from the pre-frontal cortex, the source of human decision-making. Where this article mentions a statement or issue as plus or minus a certain percentage, that’s how much above or below it is from the average persuasiveness for both candidates for the entire final debate across all people analyzed. Our 60 participants were largely university-educated and culturally-diverse people — Americans, Canadians and those born elsewhere — who live in the (largely) socially progressive and multi-cultural city of Toronto, Canada.

To illustrate, consider a response to Chris Wallace in the final debate. Wallace pressed Trump on his characterization of Aleppo as having fallen. Referencing the city’s ruin and rubble, Trump successfully rebuked the moderator and eschewed the standard diplomatic and political definition of ‘fallen’ by asking: “What, do you need a signed document?” This was a great moment for Trump: it was 39% more persuasive than the debate mean, and held across ages, genders and nationalities.

This suggests that Trump’s position cut through the political-speak, while sticking up for the little guy. To many, this was subconsciously gripping. This is the hypnotic spell to which I’m referring. That’s the trap he and other strongmen set — it’s the bait on the hook. And though our rational minds would never admit it, our irrational minds went for the bait.

Of echo chambers and reality TV politics

Our emotions drive our actions and behaviour, despite the growing amount of available information at our disposal. Or, maybe, it’s because of it.

Such is the second troubling trend we’ve observed — the impacts of a changing media landscape on voter perceptions and citizenship. Better data and targeting capabilities have meant increasing media spend in digital channels; the ability to shape the news agenda and rally together followers have meant increased resourcing in social media for campaigns.

But a downside of this is the impact on broader social cohesion through the corrosive effect of ‘echo chambers’. The algorithmically ring-fenced media universe of each individual ensures that we’re consuming only the viewpoints we agree with and want, in a constant feedback loop of ever-refined ‘relevance’.

To illustrate, a Wall Street Journal interactive graphic shows the typical Facebook newsfeed of a ‘liberal’ and the typical Facebook newsfeed of a ‘conservative’ side-by-side. Users of the tool can select a topic like “Guns” or “Presidential Debate”, and peruse the sample feeds. They couldn’t be more different.

This narrowing information diet hardens our political biases. When The Economist investigated trust in various conservative and mainstream media outlets, it found 88% accuracy in predicting voter intention based on the news outlets one trusted or distrusted. This was a better predictor than party registration, race, education and gender combined.

What has resulted is information warfare, with battles waged on every screen and platform. When Brainsights posted its debate tool on its own Facebook feed, we were inundated with comments, most from apparent Trump supporters. The post only reached about 8000 people, but we had 28 screen pages full of comments (see video). We had to ban a number of these users for trolling and posting over and over on everyone else’s comments.

When we looked into the profiles of these trolls, they were generic and vague: profile photos were of trees, or sports team logos or of Trump on a red background; names were also common or generic Al Jones, Art Clarke (apologies Al; Art). These were open profiles, but with very little information on them. We suspected that these people had been paid to spread (dis)information, and attack other commenters’ views. Apparently, this is a thing.

And it’s not just Facebook: Twitter is Trump’s preferred broadcast medium, and trolls spread disinformation there, too.

These platforms are driving the news agenda, and traditional news media — newspapers and TV news networks — regularly report on social media activity. What has tended to get traditional media attention throughout the campaign — the loudest, most active and most vitriolic — gets broadcast to broader audiences, breaking the echo chambers and shocking ‘the other side’. Because of varying levels of trust in media by various voting groups (see The Economist study above), consumers will tend to believe only those outlets sharing their political views, dismissing others as trumped up or untrue. This in turn creates a ripe environment for misinformation to spread, where “less truth more audience.”

“Donald Trump is not simply a made-for-TV candidate; he is a made-by-TV candidate.”

By most rational accounts, Donald Trump shouldn’t have been in this race. In part, we can blame the collective media for this, which includes the networked public. In their quest for pageviews, clicks, buzz and ratings, news networks needed a race. So, they elevated a demagogue and diminished an accomplished stateswoman. This dynamic has been poignantly parodied by The Beaverton and SNL.

To be fair, that’s the media business. Their customers pay for audiences, so media companies are incented to drive audience ratings. For TV networks, elections are generally big business. Not so, this year. Despite record amounts of money raised by each campaign, TV advertising hasn’t seen the spend of previous campaigns. In their bid to maintain relevance, networks did what they do best — they told stories.

And they had a familiar and willing character to play a starring role.

Donald Trump is not simply a made-for-TV candidate; he is a made-by-TV candidate. This reality TV character has been broadcast and cultivated for decades; mix in a healthy dose of social media and it’s an intoxicating cocktail virtually certain to drive media metrics.

As a supreme media manipulator, it was silly to underestimate Donald Trump — but most people did anyway.

There’s an old saying that a lie spreads halfway around the world before the truth can get its boots on. Donald Trump knew this better than anyone. And when he was called on the many times he lied, he’d shrug it off saying that the mainstream media was rigged against him. But he’d use Twitter to do so, and his millions of followers, who were deeply distrustful of mainstream media, would believe him anyway.

And while many in mainstream media laughed off suggestions that anything could be rigged against a billionaire — a ‘one-percenter’ — Trump persisted with the message.


For three reasons: echo chambers, decoys and siege mentality.

First, the logic of the echo chamber phenomenon described above goes like this: If the mainstream media is just showing ‘bad’ things about Trump supporters, and I’m a Trump supporter, and I’m not bad, the mainstream media must be rigged.

Thus, if the mainstream media is dismissing claims of election rigging, but everyone around me says otherwise, the mainstream media must be lying. It’s rigged.

Rinse and repeat.

Second, specific ‘rigged things’ acted as decoys for the press. Many of Trump’s supporters may genuinely have believed that the election would be rigged, and that he — and by extension they — shouldn’t accept the results should they not be to their benefit. Yes, this is truly terrifying.

But it’s also far too simplistic an explanation and in the end, this was a red herring. Responding to his non-commitment to respect the election results in the final debate, Clinton called out Trump for thinking that every time something didn’t go his way, that thing was rigged.

It was a persuasive moment for her, but unfortunately for her, it was never about Donald Trump. It was about the people to whom he was giving voice. By focusing on the ‘rigged election/not accepting the results’, the press — and Clinton’s campaign — failed to connect the dots that the rigged electoral system represented a symptom of a broader, interconnected system rigged against many of the people Trump galvanized.

Which leads to the third and final reason Trump persisted with the message: it fostered a siege mentality between Trump and his supporters and the opponent. Every war needs an enemy, and every rallying cry needs a cause. Their enemy: a decades-long political insider, a former First Lady, Senator and Secretary of State; their cause, systemic corruption — cultural, social, economic and political; all the things rigged against them. In Hillary, Trump’s supporters saw — and were persuaded to see — both the symbols of this corruption manifest, and thus the causes of their perceived plight.

That such a narrative would find so eager an audience speaks to the third troubling trend: the underlying conditions that led people to believe the system is rigged against them. The reasons are a mix of economic, political, social and cultural factors, and are represented by key themes consistently surfaced throughout the campaign so as to galvanize Trump’s coalition (see below).

Brainsights’ data reveals that many of Trump’s most persuasive statements — statements that drove subconscious tune in, emotional resonance and memory development across all audience and voter segments — were soaked in contempt for Clinton. But more broadly they invoked the systems and trends that she was increasingly seen to represent: a dishonest and unrepresentative Washington; an unfair and unequal economy; social policies that ‘take care of others, but not me’; and cultural shifts that ‘move America further away from the America I want’.

Trump did this by using Clinton’s strengths against her, finding an opening and exploiting it. This was classic challenger brand strategy. Clinton’s greatest strength — her experience, both personal and professional — was relentlessly portrayed as her greatest weakness. And nothing was out of bounds — her gender, her ideals, her personal life.

The crack — the opening — was Clinton’s emails. Trump could anchor the entire character assassination in Clinton’s emails — the abuse of power, the poor judgment, the opacity, the ‘cover-up’. He could sow enough doubt, suspicion and interest in them, so as to distract her campaign and the press (see SNL link above). In the rational mind, it was a non-issue. In the irrational mind, it was ‘yea, but I don’t quitetrust her. She must be hiding something’.

In order for the strategy to work, Trump had to portray himself as the complete opposite: transparent, outsider, ‘one of us’ (that is, one with his supporters).

And so, in a perverse way, the 3am Twitter tirades, the ‘locker room talk’, the groping, and the rambling incoherence in much of the debates all served to underline Trump’s position. At its basest level: with Trump, what you see is what you get. With Clinton, it’s impossible to know.

Trump became a believable alternative.

He reinforced this with his language, calling Hillary a liar and a cheat frequently, and generally speaking in absolutes (‘nobody’, ‘totally’, etc). It didn’t always persuade voters, but it’s clear from the data below that much of it struck a meaningful subconscious chord.

Calling Trump’s supporters “the deplorables” was only part of the problem for Clinton. More to the point, it was the contrast between Trump’s perceived radical transparency and Clinton’s closed demeanor, which manifested in her language: scripted, verbose and at times wonkish.

And it’s worth reiterating here that this study was not done in the heart of Trump country, but in the cosmopolitan city of Toronto. If this rhetoric connected meaningfully with Americans living in Toronto, how much deeper might this have resonated with those in the Rust Belt, or Deep South?

The Data Behind the Rigged System

To provide more context and data to the connecting power of “rigged” and its derivatives, we pulled some key candidate moments relating roughly to various systems — political, economic, social and cultural. They are categorized for ease of consumption, but are interrelated and as such, should be considered as connected pieces of a whole.

Rigged SystemPolitical
Trump Campaign Keywords: Emails, Pay for play, Clinton Foundation
Why it stuck to Hillary?

  • Former First Lady, Senator and Secretary of State, Hillary spent the best part of 20 years at the very top of American politics.

  • She’s a Clinton — it’s perceived as a political dynasty.

  • The FBI’s inquiry into her use of a private email server as Secretary of State, and questions over her conflict of interest raise suspicions about her integrity.

Key Quotes and Persuasiveness Scores:
DT: “She gets away with it, and she can run for the Presidency? That’s really what you should be talking about. Not fiction.” (+26% persuasiveness)

DT: “She’s lied hundreds of times — to the people, to Congress and to the FBI” (+44% persuasiveness)

DT (on “the fall of Aleppo”): “But it has fallen….from any standpoint. I mean what do you need? A signed document?”
(+39% persuasiveness)

HRC (on accusations of Clinton Foundation ‘pay to play’): “Everything I did as Secretary of State was in furtherance of our country’s interests and values” (-36% persuasiveness)

Rigged SystemEconomic
Trump Campaign Keywords: NAFTA, trade agreements, Open Borders, Wall Street
Why it stuck to Hillary?

  • Bill Clinton was President when NAFTA was signed and ratified. Free trade and moving manufacturing jobs overseas have decimated the manufacturing heartland of Michigan, Ohio, and Pennsylvania.

  • Hillary’s election would put Bill back in the White House.

  • Hillary is known to have large Wall Street donors, and was a United States Senator from New York over Wall Street’s boom and bust years last decade.

Key Quotes and Persuasiveness Scores:

DT (Referring to the Trans-Pacific Trade Partnership): “She lied when she said she didn’t call it the gold standard in one of the debates. She totally lied.” (+28% persuasiveness)

HRC: “Donald bought Chinese steel and aluminum. In fact, the Trump hotel right here in Las Vegas was made with Chinese steel.” (+57% persuasiveness)

HRC: “There’s only one of us on this stage that’s actually shipped jobs to Mexico and that’s Donald.” (-12% persuasiveness)

HRC: “But what is really troubling, what we learned in the last debate, he has not paid a penny in federal income tax. And we were talking about immigrants a few minutes ago Chris, you know, half of all immigrants, undocumented immigrants in our country actually pay federal income tax, so we have undocumented immigrants in America who are paying more federal income tax than a billionaire. I find that just astonishing.” (-7% persuasiveness)

Rigged SystemSocial
Trump Campaign Keywords: The Wall, Obamacare, Great Migration, Clinton Foundation
Why it stuck to Hillary?

  • Hillary is a Democrat closely associated with the preceding Obama administration, having served as Secretary of State.

  • This close association means (perceived) greater involvement in Obamacare and other hallmark “progressive” social policies of the Obama administration, like same-sex marriage

  • Secretary of State is America’s most senior diplomat, ie, the face of America to the world

  • Hillary is well-travelled, well-educated and well-informed, a so-called “liberal elite”

Key Quotes and Persuasiveness Scores:

DT: “We cannot take 4 more years of Barack Obama and that’s what you get when you get her.” (+86% persuasiveness)

DT: “Now, I wanna build a wall. We need the wall. The border patrol, ICE, they all want the wall. We stop the drugs, we shore up the border. One of my first acts will be to get all of the drug lords — all of the bad ones, we have some bad bad people in this country that have to go out — SNIFF — We’re going to get them out, we’re going to secure the border. And once the border is secure, we’ll make a determination at a later date as to the rest. But we have some bad hombres here and we’re gonna get em out!” (+34% persuasiveness)

Trump On Refugee Migration to America as “The Great Trojan Horse” (+ 10% persuasiveness)

HRC: “You know President Obama said the other day, when you’re whining before the game is even finished, it just shows, you’re not up for doing the job.” (-17% persuasiveness)

HRC: “America is great because America is good” (-19% persuasiveness)

Rigged SystemCultural
Trump Campaign Keywords: The Wall, Obamacare, Great Migration
Why it stuck to Hillary?

  • Hillary has shattered the very highest glass ceilings of American society, having been the first female Presidential candidate of a major political party.

  • Follows — and it closely associated with — the shattering of another major barrier in American society — race — with Obama’s Presidency.

  • Symbol of cultural progress on issues of gender, racial, ethnic, sexual and minority rights and equality

Key Quotes and Persuasiveness Scores:

DT: “I will do more for African Americans and Latinos than she can ever do in 10 lifetimes.” (+88% persuasiveness)

DT: “Nobody has more respect for women than I do. Nobody. Nobody has more respect” (-6% persuasiveness*)
(*Note: We’d have expected a much lower score.)

DT: “And you’ll see, we are going to stop radical Islamic terrorism in this country. She won’t even mention the words and neither will President Obama.” (+13% persuasiveness)

HRC: “The Supreme Court needs to stand on the side of the American people, not on the side of powerful corporations and the wealthy. For me that means that we need a Supreme Court that will stand up on behalf of women’s rights, on behalf of the rights of the LGBT community” (-22% persuasiveness)

HRC: “It is clear when you look at what Donald has been proposing — he started his campaign bashing immigrants, calling Mexican immigrants rapists and criminals and drug dealers — that he has a very different view of what we should do to deal with immigrants” (Average persuasiveness)

DT: “I believe that if my opponent should win this race — which I truly don’t think will happen — we will have a second amendment which will be a very very small replica of what it is right now” (+5% persuasiveness)

HRC: “Donald thinks belittling women makes him bigger” (+34% persuasiveness)

On Recency, and the Importance of a Compelling Call to Action

This chart shows moments of high persuasion (defined as 100% or greater than the average) for the final debate.

As you can see, Donald Trump’s key moments outnumbered Hillary’s overall, and in particular, were weighted towards the final portion of the debate. In other words, there was a gradual build of his most effective moments, and a tailing off of Clinton’s.

And if you think of the debate as one long ad, the closing arguments would be when the call to action and branding get revealed. It’s the final impression that gets left with the consumer, the voter.

In the closing arguments, Trump outperformed Hillary by 69%. This was driven by an impassioned and compelling call to action, with two statements in particular standing out:

“I will do more for African Americans and Latinos than she can ever do in 10 lifetimes.” (+88% persuasiveness)

“We cannot take 4 more years of Barack Obama and that’s what you get when you get her.” (+86% persuasiveness)

If you were an undecided voter, turned off by the xenophobic rhetoric of Trump’s campaign, but unconvinced by Clinton’s suitability, hearing Trump deliver these two statements with such conviction might sway you. Indeed, that’s precisely what this data suggests. Furthermore, it was delivered at the finale of the last Presidential debate, which capitalizes on the principle of recency for undecided/late deciding voters. This states that what is most likely to be remembered and learned is that which is last consumed. In this respect, Trump was the runaway winner.

So where can we go from here?

Closing thoughts on bias, media and implications for democracies and social institutions.

In peering into the human mind, we’re offered glimpses of what people feel and value. For political content, that means gaining a deeper understanding of their mood, their views on issues, and what they value in a leader. The granularity of the data allows us to examine the micro-moments, but we can also aggregate this data to determine responses to broader macro trends.

Here are five thoughts on what we can do to improve both our collective decision-making, and the systems that presently feed its flaws:

  • Political polling needs to evolve. Cutting through to non-conscious responses (System One) is essential, in an environment rich with emotions, and with a candidate all-too-willing to challenge taboo political topics. Trump isn’t the first politician to do this (see: Brexit), and he will not be the last. If pollsters and those who follow them do not wish to get caught out again, they’d be wise to improve the data that goes into their models.

  • The electorate must be empowered to understand their unconscious biases. Whether it’s the appeal of the strongman, the allure of the simple message, or the power of recency — not to mention racial, gender, ethnic and sexual biases — an opportunity exists for the public to be better informed of the biases that shape their world views. Only by illuminating these biases can we work towards mitigating and removing them.

  • Media and information education have never been more critical.How can we teach people to interrogate information and media to get at a clearer understanding of the issues? We wrongly equivocate the volume of information we’ve consumed with how well informed we are; quality of consumption must also be considered. Is there a way to score this quality of information beyond ‘liking’ and ‘sharing’?

  • Self-selecting media maybe isn’t all it’s cracked up to be. There’s value in hearing and understanding the other side. The more we insulate ourselves from competing perspectives, the more hardened our informational biases will become, and the more savvy marketers and politicians can take advantage.

  • Whereas digital platforms provide ever-greater customization and targeting opportunities, live broadcast events — traditionally the realm of TV, but increasingly available on digital platforms like Facebook, Youtube and Twitter — provide opportunities to address and connect with new audiences. Communicators that do not leverage this important dynamic risk speaking only to existing customers or tightly targeted prospects, thus potentially stagnating growth.

  • Leaders must do a better job managing change and distributing gains. The ‘rigged system’ message won Trump the Presidency. That this was possible speaks to too few people sharing in the benefits of progress, whether economic, political, social or cultural. If the underlying conditions of inequality in social, economic, cultural and political gain are not managed, and if lack of opportunity, education and social cohesion are not addressed, we can expect both more institutional instability, and anger from the people.

What do you think? Is it possible to improve the decision-making of the electorate? Is it possible to improve information quality? Share your thoughts in the comments section.

Feature image source: AP Photo/Pablo Martinez Monsivais via CBC.CA

Learn more about how persuasive each candidate was on the issues by playing with our interactive tool here.