As we head to the US election- less than a month away at the time of writing- it looks like party time on November 3, 2020 for Joe Biden and the Democrats. Poll after poll shows Biden beating 45, many by significant margins.
There is also a vertical win in the cards down the ticket, with the Democrats tipped to retain the House of Representatives and in with a chance to win the Senate. That would mark only the second time since 2008, when Obama won, that one party had the White House, Congress, and the Senate at the same time.
Halcyon days, right?
Yes, if that actually happens, but let’s step back a bit and look at things through the lens of Pollology. That’s not a real word, it’s one I made up to describe the difference between what the polls say and actual election results.
There’s plenty of evidence to suggest that Pollology is a real thing: 2016, when 45 defeated Hillary Clinton despite what the polls said, right to the day of the elections, is the poster child for it. But there’s a longer history and the implications for Biden and the Democrats are deep and worrying.
It’s useful to think about how polls are conducted, and typically that’s either via phone, snail mail or an online survey. Each of these methods have one thing in common- somebody can track you through some combination of your name, address, telephone number or IP address.
Which means that someone, somewhere, will know how you feel about sensitive things like abortion, gun control, gender equality, BLM, candidates for political office, and who you plan to vote for. For a lot of people that’s a scary thought- not just that someone knows how you feel but how that information might be used. Think Cambridge Analytica and AggregateIQ, with both Brexit and 2016 US election.
That’s where Pollology kicks in. It’s axiomatic that people want to be liked- not everyone, perhaps, but most of us. So, when pressed on these sensitive matters, many of us are biased to giving an answer that is politically correct, one that makes us come across as a ‘nice person’.
These responses skew polls and surveys in ways statistical analysis can’t really handle. That margin of error caveat in polls is about confidence intervals and sampling errors, not whether you answered a question as you actually feel about something or someone. Only you know that and we know that you want to be liked.
Shy Voters, Shy Trumpers
Much has been written about this, and a lot of it has focused on races where there is a black candidate and a white candidate. In fact, there’s something called the Bradley Effect, named after Tom Bradley, a long time Mayor of Los Angeles and the black Democratic candidate for California Governor in 1982, who lost to the white Republican candidate, George Deukmejian, despite polls to the contrary. Apparently, poll respondents didn’t want to be seen as racist.
The Bradley Effect has also been used to explain the results of numerous races where the black candidate lost to the white candidate at the voting booth, bucking polling data- most recently the 2018 Andrew Gillum- Ron DeSantis Governor race in Florida.
Which brings us to the ‘shy voter’ concept, which seeks to explain why candidates and parties perform much better than the surveys and polls suggest.
The term tracks back to the 1992 UK election, when the polling pointed to a Labour win, but John Major’s Tory Party pulled through, giving rise to ‘shy Tory’ voter theories: people who didn’t want the pollster to know their real inclination, for fear of being unliked.
That history repeated in the 2016 Brexit referendum, when polls suggested Remain would be the close winner- we know how that went- and again in 2017 when Theresa May called an election based on polls that showed her Tories with a 20-point lead in the polls, only to wind up with a hung Parliament. Hello, Boris Johnson.
Then there’s the ‘shy Trumper’ theory for the 2016 US election- referring to those who wanted 45 but again weren’t willing to admit it to the pollsters.- they revealed their preference at the polling booth, in private. There is legitimate fear that this is happening again in 2020- especially in light of Trump’s tangible toxicity
Wanting to be liked isn’t the only reason polls and results disconnect- far from it. Sample size, likely voters, and how questions are asked, are just some of the many other factors that play into this misalignment.
That’s why one poll can show Biden up 14 points and another have it at 8. And polls are obsolete as soon as the numbers are announced, because it takes more than a few days to conduct a poll and crunch the numbers. In effect, they tell us how people were thinking a few days ago, not right now. A week is a long time in politics, as the saying goes.
What The Dems Must Do
I am not arguing that polls are “lies, damned lies, and statistics” (Mark Twain, not Benjamin Disraeli). They are useful in that they capture some sentiments at one point in time. But the social acceptance desire that informs people’s responses to polls and surveys must always be remembered.
It’s really more a call to action for the Democrats. That call is pretty simple: for God’s sake, don’t take anything for granted this time.
Get out there, act like you’re behind everywhere and as if it’s Game 7 of a championship series or the World Cup Final, and pull out all the stops in every precinct. Remember that the last previous Vice President to win the Presidency was George H.W. Bush in 1988, and that result came as much from a desire to continue the Reagan years as anything else.
We all know what happens when you assume too much.
Just ask Hillary Clinton.
Writer: Deepak Kamlani
