Navigation

After off the mark polls, some question value of ‘big data’

The Courier via AP

Supporters watch as results come in at the a GOP election party in Cedar Falls, Iowa, on Tuesday.

The Associated Press For an American public that relies on data for everything from where to find the best taco to the likely victor in a baseball game, Election Day offered a jarring wake-up: The data was wrong.

Donald Trump’s electoral win came despite prognosticators’ overwhelming insistence he would lose. And it has forced many to question not just political polling, but other facets of life that are being informed and directed by data.

“If ‘big data’ is not that useful for predicting an election then how much should we be relying on it for predicting civil uprisings in countries where we have an interest or predicting future terror attacks?” asked Patrick Tucker, the author of “The Naked Future: What Happens in a World That Anticipates Your Every Move?”

Technology has filled people’s lives with crowdsourced, data-driven or otherwise instructive metrics and left many convinced of their validity. We look to Yelp rankings to find a good meal and TripAdvisor to gauge a city’s finest hotel. Netflix tells us which shows are best to watch and Zillow tells us the worth of the home we might buy. Amazon, Google, Facebook — all are ubiquitous presences in everyday life with data at their core.

So many took the predictions of polling aggregators as gospel — and their forecasts of Hillary Clinton’s chances of winning the presidency went as high as 99 percent.

Tammy Palazzo, a 49-year-old corporate trainer in Maplewood, N.J., was among them. She is a political junkie who all day long refreshed FiveThirtyEight, the site from the heralded election-predictor Nate Silver. She purposely sought to look beyond her biases, living in a neighborhood dotted with signs for Clinton, a candidate she admired, and took in a variety of news sources. Then came Tuesday night.

“This is not going the way it’s supposed to go,” she thought, watching returns come in but not in Clinton’s favor. “There was just so much reinforcement from the media that this was pretty locked down.”

Allan Lichtman, a history professor at American University, was among the few who defied the confluence of polls and projected Trump’s win. His model, developed in 1981, uses history as a guide to who will win the presidency through 13 true-or-false questions looking at economic indicators, military failure and success, social unrest and third-party candidacies. He has been consistently right.

“Polls are not predictions. They are snapshots and they are abused and misused as though they are predictions,” he said.

Speech

Click to play

0:00/-:--

+ -

Generating speech. Please wait...

Become a Premium Member to use this service.

Become a Premium Member to use this service.

Offline error: please try again.