Why the polls were wrong

Nate Cohn of The New York Times Upshot column has an eye-opening analysis of why so many state-level polls were wrong about Trump and Clinton last year. More on that later.

But bad polls aren’t to blame for Hillary Clinton’s loss. Because her campaign didn’t do polls.

That’s the most shattering revelation in a new book, “Shattered: Inside Hillary Clinton’s Doomed Campaign.”

Her campaign, led by manager Robbie Mook, believed in data and analytics. They knew their base voters. Tracked them relentlessly. Modeled their predicted performance obsessively.

And didn’t worry about persuading undecided voters. Or polling them to find out what they cared about, what they yearned for in a President and what they saw – or didn’t see – in Hillary.

That’s why her campaign didn’t see white working class voters turning to Bernie Sanders in the primaries. And didn’t go after them in the general election campaign.

That’s why Hillary, to her great frustration, couldn’t figure out why she wasn’t connecting. And why Bill Clinton, to his great frustration, couldn’t persuade Mook to go after the voters Bubba used to get.

Polls and focus groups get a bad rap. Mainly from people who have never used them and learned from them.

A good poll is the only objective way to tune out all the noise, advice, anecdotes and theorizing from well-meaning people who don’t know what they’re talking about – and find out what real voters really think.

Yes, as Cohn recounts, a lot of state-level polls were wrong in 2016. (Most national polls predicted Clinton would narrowly win the popular vote, which she did.) There are a lot of theories why polls misfired: Maybe undecided voters broke heavily for Trump at the end. Maybe Trump voters were less likely to admit they were Trump voters. Maybe the polls’ voter screens tilted toward Clinton. Maybe – and most intriguingly – polls didn’t weight properly for education levels, which turned out to be a good predictor of voting behavior.

All of those errors can be fixed. A good pollster will produce a good poll.

Yes, data and analytics have their place. Polls have their place. And experienced political guts like Bill Clinton’s have their place.

It takes a village.

 

Avatar photo

Gary Pearce

Categories

Archives

Recent Posts

Why the polls were wrong

Nate Cohn of The New York Times Upshot column has an eye-opening analysis of why so many state-level polls were wrong about Trump and Clinton last year. More on that later.

But bad polls aren’t to blame for Hillary Clinton’s loss. Because her campaign didn’t do polls.

That’s the most shattering revelation in a new book, “Shattered: Inside Hillary Clinton’s Doomed Campaign.”

Her campaign, led by manager Robbie Mook, believed in data and analytics. They knew their base voters. Tracked them relentlessly. Modeled their predicted performance obsessively.

And didn’t worry about persuading undecided voters. Or polling them to find out what they cared about, what they yearned for in a President and what they saw – or didn’t see – in Hillary.

That’s why her campaign didn’t see white working class voters turning to Bernie Sanders in the primaries. And didn’t go after them in the general election campaign.

That’s why Hillary, to her great frustration, couldn’t figure out why she wasn’t connecting. And why Bill Clinton, to his great frustration, couldn’t persuade Mook to go after the voters Bubba used to get.

Polls and focus groups get a bad rap. Mainly from people who have never used them and learned from them.

A good poll is the only objective way to tune out all the noise, advice, anecdotes and theorizing from well-meaning people who don’t know what they’re talking about – and find out what real voters really think.

Yes, as Cohn recounts, a lot of state-level polls were wrong in 2016. (Most national polls predicted Clinton would narrowly win the popular vote, which she did.) There are a lot of theories why polls misfired: Maybe undecided voters broke heavily for Trump at the end. Maybe Trump voters were less likely to admit they were Trump voters. Maybe the polls’ voter screens tilted toward Clinton. Maybe – and most intriguingly – polls didn’t weight properly for education levels, which turned out to be a good predictor of voting behavior.

All of those errors can be fixed. A good pollster will produce a good poll.

Yes, data and analytics have their place. Polls have their place. And experienced political guts like Bill Clinton’s have their place.

It takes a village.

 

Avatar photo

Gary Pearce

Categories

Archives