🗳️ Why We Can't Predict Elections Anymore
How all the polling in the world was still wrong about the midterms.
No red wave, just a “pink trickle” (gross), and democracy still seems intact. This year’s midterms were the best kind of election: a mostly boring one. But just like 2016 and 2020, the predictions and hype were wrong. Why?
Coverage of local politics is half-baked
Local news coverage of politics is a dead-end career.
In fact, non-profits do almost as much reporting on state politics today as traditional journalists do. As local news coverage vanishes, so too does the local beat reporter who understands the nuanced political views of residents - which often don’t fall on one side of a two-party divide. As Congresswoman Elissa Slotkin puts it:
“What we have instead is a lot of really wonderful journalists who parachute in for 36 hours, talk to a few people … you know, in a coffee shop … and then make their assessment, as opposed to knowing what is really going on …”
And those half-baked stories are amplified as gospel
The problem then is that these half-baked analyses of local voter sentiment are amplified, shared and re-shared ad nauseam on the media-political industrial complex.
Siloed media is a threat to the First Amendment rights of many Americans, but the segmented media atmosphere we live in today also amplifies the non-partisan, yet half-baked predictive reporting that led many astray this election season.
And because these stories rely on polling and stats that seem to predict specific outcomes with coherence, they are hard not to believe, especially by an audience of the fervently committed. Democrats were primed to believe that this was an election on the fate of democracy itself, while Republicans knew the country was ready to hand them a “red wave” of victories. Both were wrong.
Why? Because the same superficiality that plagues local reporting, also infects the polling practices and election prediction modeling that these stories rely upon to prove their point.
Local stories are based on faulty polling
Information gathering mechanisms, like polling and surveys, that focus on the cooperation of the willing have built-in biases that must be accounted for, especially in the aggregate. For instance, only people with extra time complete polls and surveys. The identity of the questioner determines the willingness of the respondent to participate, as do the means of communication (e.g. in-person, telephone).
And importantly, as we saw in 2016, sometimes people are scared and embarrassed to admit their political viewpoints to a stranger … and even anonymously to a paper form!
Survey response rates also suck and even polling experts this year admit that “We’re down to 1 percent of people on a good day who are willing to talk to a pollster for free. Most of those who respond are a vocal minority who care about crime, while we ignore the silent majority who care about issues like abortion.”
In short, surveys and polling are increasingly imperfect ways to gather information. But beyond low response rates, polls also don’t take significant portions of the electorate into account.
Polling ignores younger voters
Young people, and even many middle-age folks … don’t really take polls. What king of person enthusiastically takes phone calls from unknown numbers? You probably have an answer in mind … so imagine how off-kilter polling that relies on that population would be.
It’s no surprise then that the polls largely ignored Gen Z and Millennial voters. Beating the “odds” young people (especially young women) showed up to vote in huge numbers this week, and that showing shifted the tide in many close to call elections. I saw many current and former students of mine register to vote for the first time. And the evidence isn’t just anecdotal: midterms are usually a low-turnout election cycle, but this year saw the highest youth turnout in recent memory - higher even than 2018, which was already a record year.
And yet …
Knowing all of this, pollsters still made the same fundamental errors in judgment as their predecessors in 2016. In fact, they admitted to it:
“We are, in many respects, stumbling through the dark with headlamps and flashlights.” —Dave Wasserman, U.S. House editor at the Cook Political Report (source)
These admissions and glaring weaknesses in the data, were not vetted because there are few local reporters to cross-check data against local sentiment. And so media outlets integrated these “local” stories with bad data into national reporting, and then a siloed media system, now so practiced at encouraging surprise and fear, amplified these stories for consumption by a mass audience. Thus, why we were so wrong about these midterms.
So what can we do about it?
Support good local reporting on politics. Many foundations and high net-worth individuals are already doing so, you can join them by:
Giving to NewsMatch - which funds non-profit local newsrooms and then finds donors to match each dollar given by individual donors, effectively doubling your contribution.
Looking for a career change? Become a local state politics reporter. There’s a lot of funding for it sitting on the sidelines …
Use, fund or socialize startups that are trying to fill the gap in bad stat work.
🎧 Now Playing on the UnfairNation Podcast
✈️ Catch Me If You Can
Las Vegas | November 11 - 12
Colleagues at the Pacific Council of International Policy (based on the west coast) are accepting applications for the AMPLIFY Fellowship - for those young and mid-career foreign policy professionals who are often underrepresented in the profession. Apply here.
📅 Reading & Watching
The Gatekeepers: How the White House Chiefs of Staff Define Every Presidency
He’s 13 years old and makes sure folks are fed every Thanksgiving. 200 of them.
The problem with doing nothing is that you never know when you’re finished.