The final result is in and already the vultures are circling over the record of the pollsters in the US election. Is this another nail in the coffin of opinion polls, another Brexit moment or something more profound? I believe there are three important lessons for market researchers and one more profound lesson for everyone.
#1 Probability Trumps Certainty
Already many are declaring that the opinion polls were “wrong” and that there must be some deep-seated flaw in the methodologies used to measure public opinion and voting intentions. But did the polls get the result wrong and if they did, how wrong were they?
Although Donald Trump won the electoral college vote (by how much is still to be determined) Hillary Clinton won the popular vote. Given that most of the polls throughout the election campaign had Hillary leading by 2-3%, with a margin of error of 2-3%, this is not quite as wrong as it looks.
More importantly, while some individual polls and pollsters were giving Clinton a 99% chance of victory, others were much more cautious. Nate Silver consistently gave Trump a good chance of winning in his “poll of polls” and in his final summary before election day put the chances of a Donald Trump victory at almost 30% (you can see the forecast here). This poll of polls even put Donald Trump’s chances above 50% for a short period in August. Even at 30%, Trump had a 1 in 3 chance of winning – good odds by most standards.
The polls were sometimes a little misleading but not completely wrong. The political pundits and their interpretation of data called it wrong rather than the polls themselves. There was always a significant and solid chance that the polls were pointing to a Trump victory. The lesson is that market researchers should always think probabilistically and avoid the attractions of certainty.
#2 (Sometimes) Detail Trumps Big Picture
The polls were right in that Hillary Clinton won a majority of the popular vote. So why is Donald Trump going to be President in January? The devil is in the detail.
The last time that the President elect lost the popular vote was famously in 2000, when Florida was finally called for George Bush after legal challenges and Al Gore lost as a result, despite having received nearly half a million more votes than George Bush (48.4% vs 47.9%). You have to go back to the nineteenth century for previous occasions when this has happened.
Although Hillary Clinton was (mostly) ahead overall throughout the campaign, there were always a number of states that were too close to call, including Florida which accounts for 29 electoral votes. In the last state-by-state forecast on Nate Silver’s website (www.fivethirtyeight.com) the predicted outcome is for Hillary Clinton to win the state by 0.7% of the vote (a virtual toss up). In four other states the candidates were predicted to be within two percentage points of each other.
If there were polling “errors” initial analysis shows that they were focused on white voters without college degrees, as the difference between the final polls and the results are correlated with the numbers of voters who fall into this group. These differences could be systematic polling errors or late shifts in opinion. There are even theories about “shy” Trump supporters, especially women, but couldn’t it be true that there might be “shy” Clinton supporters too?
Whatever the verdict of the post-mortem, market researchers should always dig into the details as well as looking at the big picture. The “average” is very often misleading, whether that is the average voter or the average consumer. There are always insights at the margins and in the differences that hide under the surface.
#3 Emotion Trumps Reason
Regular readers will not be surprised to learn that a third lesson is that it is almost always more important to understand how people feel before asking them what they think. Emotional frames trump rational arguments in politics (read more here) and anyone who has seen anything of the Presidential campaign will know that Donald Trump’s appeal was very much to the heart while Hillary Clinton’s was very much to the head.
While both heart and head are important to decision making, it is almost always the heart that is in the driving seat. More specifically, and just as in the UK Brexit vote (read more here), Donald Trump was appealing to some very specific emotions.
Recent work by Lerner and co-workers shows that while Fear is associated with feelings of uncertainty and risk, making people feel the world is unpredictable and outside their control, anger is associated with feelings of certainty and predictability, making people feel that they have more control over events that are easier to predict.
Donald Trump understood that making voters angry gave them greater motivation to vote for a change that they believed would give them a greater degree of certainty about their future, whatever reservations they might have about him. Conversely, messages focusing on fear made people more unsure about what to do. Anger trumps fear.
As a side note, it seems that the big data of all the polls combined couldn’t capture a true sense of what voters would finally do. Equally, Hillary Clinton’s huge emphasis on voter data, analytics and ground operation didn’t overcome the more visceral and less cerebral approach of Donald Trump.
Market researchers (and businesses) need to better understand human emotions in order to understand human behavior. You have a much better chance to predict future behavior by understanding how people feel rather than knowing what people think. Perhaps small data trumps big data?
#4 (Sometimes) Big Picture Trumps Detail
Apologies for getting serious and political, but I think there is one much bigger lesson to learn from what has happened this year in the UK, USA and elsewhere.
There are striking similarities between our recent history and the 1930s that are more troubling than any of the concerns with polls and surveys. In the first half of the twentieth century the world had gone through a period of globalization, followed by a global financial crash which crippled the lives of ordinary people. The political movements that followed, along with more than one populist demagogue and eventually dictator, led to policies that institutionalized racism and mass deportations and ultimately led to a world war.
That’s one lesson that we should all reflect on.
[This is an edited version of an article originally published by Asia Research Magazine.]
FiveThirtyEight.com (Nate Silver website)
“Emotion and Decision Making” by Lerner, Yi, Valdesolo & Kassam from Annual Review of Psychology, 2015, 66:33.1-33.25
The Rational Animal: How evolution made us smarter than we think by Kenrick & Griskevicius