The Problem With Polling, Surveys and Opinion Is That People Fib
If this election season has taught us anything this far, it's that polls are a lousy way to get a true sense of someone's opinion.
Think about it: Donald Trump, after crushing the competition in Indiana Tuesday night, forced Sen. Ted Cruz to bow out, meaning Trump is now the presumptive Republican nominee for president.
This comes after nearly every political pundit, analyst and media heavyweight -- you know, the folks paid to predict these kinds of things -- said Trump's entire candidacy was a joke, he was doing everything wrong, and he would drummed out early, clearing the way for that inevitable Jeb Bush-Scott Walker battle.
Part of that misguided thinking was just hubris: The political class, whether they be former politicians, operatives or media "personalities," couldn't fathom a person like Donald Trump scoring points with the public. He was just...well, wrong in so many ways. How many times have you even seen him at the White House Correspondents Dinner?
But that cultural bias was backed up by science...sort of. Poll after poll showed lackluster support for Trump. These weren't informal surveys, either. Resources were devoted to putting real statistical analysis behind the polls, because picking a president is a big deal. As Jason Trennert, managing partner of Strategas Research, said on CNBC this morning, his investment firm "paid a well-known political strategist a lot of money to tell us Trump has a zero chance of winning."
The only poll that counts -- the primary elections themselves -- started coming in and not only did Trump win most of them, he almost always won by a far bigger margin than any of the polls leading up to the elections predicted. For all the money spent on surveys and polling, none of the research was worth the millions spent on it.
There's a pretty simple reason why, and it comes down to human nature. My old boss, Roger Ailes (who knows a thing or two about political pugilism), used to always warn that you had to view any poll number you saw about candidates through a lens of human political bias. For instance, on election days, many exit polls skewed Democratic, because of a willingess on the part of Democratic voters to actually talk about whom they voted for. (Or, as Ailes would put it, "Liberals like to talk more about their feelings. Conservatives want to get back to work.")
There was some of that dynamic at work here, except for the "liberal" part. Trump's dominance over polling came out of an exclusively Republican electorate (or, in some states, those who identified as Republican). Those Republicans voted for Trump when it counted, and no one else came close. When people started talking about cobbling together delegates from the other candidates to challenge Trump, it became clear the math didn't work. People voted for Trump in landslide fashion.
It's just that no one wanted to admit it.
That means a lot of your friends -- you know, your buddies who used the #NeverTrump hashtag on Twitter or pooh-poohed his latest verbal gaffe -- closed a curtain in a voting booth and pulled a lever for Donald J. Trump.
Pick a reason: He's an outsider. The country needs a business-minded president. He speaks plainly. We need a model as a First Lady. The rest of the field was weak. We liked his hair. (Well, not that last one.)
The reasons don't matter. The results do. Trump hasn't built a silent majority. It's a fibbing majority, an electorate that doesn't want a Trump bumper sticker or lawn sign, and may even outright lie about political preference, but will still come out big for Trump.
That's bad news for Democrats, who themselves are struggling with an insurgent Bernie Sanders who is outperforming his own polls against Hillary Clinton. It's tough to face an opponent when you can't determine his source of strength and power, and Trump's appeal is undeniable.
But there are lessons for business leaders, as well. Business thrives on data, and many brands use the same polling and survey methods employed in politics. They spend a ton of money trying to figure out how to message and position their products. Businesses like to assess customer desires and needs. The biggest reason a lot of these surveys and polls are wasteful is that they often start out with a confirmation bias ("Are you satisfied with your current car?") and are really just designed to reinforce a plan, rather than shape one.
But people lie, too. For all the work trying to smoothe polling, correct biases and samples, it's tough to get around the human instinct to hide true feelings. It's why people complain in opinion polls about Wal-Mart and how it pays low wages and closes small businesses, but those same people shop there in droves.
Trump's victory should be a wakeup call for anyone who makes a living peddling research, messaging and polling. Gauging opinion was always an alchemy that involved taking science and applying a certain amount of art. Now, it seems, that art is so abstract, the number-crunchers are having trouble figuring out what it all means.