Panic porn and 2016 PTSD: Why we don’t talk about the real story of this election
We're five weeks out, and one candidate has a sizable advantage over the other. Which means only one thing: It's time to panic.
Hello, happy debate night, and welcome to Margin of Error, a newsletter from me about the polls and the way they are covered.
If you’re new here, thanks for signing up! If you like what you read (or even if you don’t but found it interesting!), I’d be grateful if you would encourage others to sign up.
If someone sent this your way or you found this post through Twitter or other channels, make sure to subscribe below.
The New York Times has a useful page embedded in its politics section. It breaks down each day’s polls and how to interpret them. There’s a feature on the right side of the page. This neat graphic caused a considerable freakout on left-leaning Twitter recently.
The Times’ snapshot of state polling averages includes a comparison to measure what the final results could look like if polls were as wrong as they were in 2016 (or in 2012). Below that, Electoral College outcomes are explored, using the comparison of the possible 2016, er, margin of error. One of the key takeaways is that if polls are as wrong today as they were four years ago, Joe Biden would narrowly lose the election to Donald Trump.
This is a pretty brilliant feature. When I say brilliant, I mean that it serves as a somewhat valuable, if also somewhat misleading, context for the state of the race. But also brilliant in terms of the way it’s undoubtedly going to siphon clicks from nervous Democrats over the next five weeks. It’s what I’m loosely terming “panic porn” geared at Democrats. It’s like The Needle on steroids, for five weeks instead of a few hours.
Predictably, these are the sorts of reactions it got (and, really, is still getting every day):
Biden is leading in Pennsylvania, Michigan, and Wisconsin by an average of more than 5 points in each state, according to FiveThirtyEight. He is within 2 points of Trump in the reliably red states of Georgia and Texas. Last week, Biden took the lead in the average of Ohio, a state Trump won by 8 points in 2016.
But we’ve decided to talk about this election using a very different narrative, amid a flurry of PTSD from the biggest electoral anomaly of our lifetimes. We're five weeks out, and one candidate has a sizable advantage over the other. Which means only one thing: It’s time for Democrats to panic.
This is not 2016
This time four years ago, Tom Jensen was getting nervous. This was not because of any statewide polls his polling firm, the Democratic-leaning Public Policy Polling, was conducting. Rather, he saw it in state legislative polls PPP was conducting on behalf of clients. He saw some of the movement in places like Pennsylvania and Michigan, where PPP’s final statewide poll showed Hillary Clinton with a 5-point lead.
“I kind of rationalized to myself that it was OK, because it must just be other districts that weren’t the ones we happened to be getting hired to poll where Hillary was doing better than Obama had and those must even things out,” Jensen told me. “Well, that was not the case!”
This year, Jensen told me, PPP is conducting polls in legislative districts across key battleground states. He estimated that the firm has done 66 polls in districts in Arizona, Georgia, Michigan, Minnesota, North Carolina, Ohio, and Pennsylvania over the past three weeks. He says it’s remarkable how consistently better Biden is performing -- all about 6 to 8 points higher than Clinton did four years ago.
There is the shift you hear a lot about: the suburbs. The type of districts that Trump narrowly carried in 2016 are now going for Biden by 4 to 5 points, Jensen said. Then there are the shifts that are not as widely covered: Less-educated, more rural districts that swung wildly -- from Barack Obama by 10 to Trump by 20 or more. Jensen says Biden isn’t winning these districts, but he’s keeping the score close. (For what it’s worth, PPP correctly nailed the 2012 election, which saw a significant amount of error in the other direction.)
“Getting medium-blown-out instead of totally blown out in the rural areas goes a long way toward winning back the Michigans and Pennsylvanias and maybe even the Ohios of the world,” Jensen said.
“Four years ago all the battleground legislative polling was really giving me some pause about what was going on,” he told me. “This time almost every single one of those polls is reinforcing to me the feeling that the state of the presidential race is fine.”
The key differences
Sean McElwee, the cofounder and executive director at Data for Progress, a progressive polling firm, says he sees several distinctions between 2016 and 2020.
“However, no one should take the election for granted,” he told me, though he quipped that “the odds are good enough it's worth putting a few hundred into the betting markets.”
The betting odds have been a good barometer of misplaced conventional wisdom this cycle. In early September, they narrowed to about even amid another topic I covered recently in this newsletter, “law and order.” The conventional wisdom around “law and order” was that the unrest in Wisconsin was going to help the president. Because the conventional wisdom assumes that, like 2016, there is going to be a comeback and a grand finale. That the reality-television host has one more trick up his sleeve.
But the conventional wisdom belies the story of this horse race — which is that Donald Trump is losing, and right now losing by margins that would be the biggest electoral blowout in at least 12 years.
McElwee noted three key differences between 2016 and 2020, which all caused major problems for pollsters last time. There are fewer undecided voters, there are fewer third-party voters, and there are fewer unknowns when it comes to Trump.
How these blend together is a crucial story of what makes this election different. During the Democratic National Convention in August, I noted on Twitter that in the FiveThirtyEight average of national polls, Biden had been at or above 50% in the average for about two months. That’s still the case today. In 2016, Hillary Clinton’s polling average never -- never! -- climbed above 46%.
Part of the reason more voters are dug in this time is that they have stronger feelings about Trump (whether for or against him), and they generally view Biden favorably. Or if they don’t like either candidate, this time, they’re breaking for Biden.
As McElwee noted, in 2016, Trump was the sideshow. Now he’s just the show.
“This cycle, the negative coverage is much more likely to disproportionately hit Trump, making it more difficult for him to climb out of the hole,” he said.
Or is ‘panic porn’ actually good?
In addition to Jensen, I asked a few other Democratic pollsters about the conundrum they face as data people. As someone with a presumably vested interest in the outcome of the election, how do you square the pretty clear data we’re seeing with any emotions that may emerge from a Pennsylvania poll that shows Biden “only” up 4?
Will Jordan, a Democratic pollster at the firm Global Strategy Group, told me that there’s really no squaring it. But he does think the caution is warranted, and that the “panic porn” may even be necessary, or at least helpful, for Democrats to avoid the same overconfidence.
One of Jordan’s lingering questions is whether pollsters have improved enough on their 2016 flaws. Back after the 2016 election, my former colleague Allan Smith and I explored what went wrong in the polling industry. Most people know the basics. National polling turned out to be pretty good, but state polling missed, especially in the key states of Pennsylvania, Michigan, and Wisconsin.
Of course, lest they want the polling industry and their jobs in it to collapse, those pollsters’ job is to fix what went wrong. The most common fix has been weighting for gaps in education levels, which many pollsters agree was to blame for issues in 2016. People with a college degree are more likely to respond to telephone surveys than people without one, and voters in the latter group tended to break for Trump.
There’s evidence that they’re fixing things. A cursory look at six of the latest, high-quality polls conducted in Pennsylvania, for example, over the past week finds that all six weighted for education. In the current FiveThirtyEight polling average, Joe Biden leads by 5.2 points.
Of course, that’s just a random sample. Jordan says he remains concerned that a significant number of pollsters still do not weight for education. “That’s crazy to me,” he says. Some forecasters, like this new one called Plural Vote, have started incorporating that into their models and, as a result, have forecasts that are slightly more favorable to Trump’s chances.
Jordan also pointed out the significant wild card: A lot of the panic and discussion stems from the odd fundamentals of the race. This is the one that, even amid all the data points, trips me up, too. Some analysts estimate that Biden can win the popular vote by as much as 4 or 5 points while Trump squeaks by again.
When you look at the current national average of about Biden plus-7, it seems less crazy to panic about a 2- or 3-point swing between now and Election Day.
“The fact Biden might have to win by 5 points nationally to win in the Electoral College is genuinely insane and a very heavy finger on the scale of ‘who’s winning’ compared to the norm of past elections,” he told me.
“So really the true problem is sort of the inverse of the appeal of ‘panic porn.’ The ‘panic porn’ is healthy, simply as a corrective, because the deck seems pretty genuinely stacked in a certain way.”
Thanks for reading Margin of Error. If you have any tips, comments, or insights about polling, email me at bplogiurato@gmail.com, or find me on Twitter @BrettLoGiurato.
If you liked what you read, share it with a friend and make sure to subscribe.