This is WESA Politics, a weekly newsletter by Chris Potter providing analysis about Pittsburgh and state politics. If you want it earlier — we'll deliver it to your inbox on Thursday afternoon — sign up here.
We don’t talk much about political polls here. And while I reserve the right to report on them down the road — especially when I want someone else to do my work for me — we probably won’t discuss them much in the future, either.
That’s because I, and many people I hear from, are neurotic enough as it is.
Case in point: a New York Times poll released this past weekend that showed Donald Trump eking out a narrow lead over Joe Biden in Pennsylvania and four out of five other swing states.
It’s not so much the overall result that shocks people. It’s the findings that, for example, Biden drew support from only 50% of Black voters, a constituency long allied with Democrats. Or the fact that Biden and Trump are tied among Pennsylvania 18-to-29-year olds. How can that be when those voters said they supported Biden by two-to-one margins in 2020? Or when they rank abortion rights — which Trump boasts about eroding — as a top issue?
Berwood Yost, who directs the Franklin & Marshall College Poll, said he’s heard those doubts.
“When you look at some of the crosstabs” — where results are broken down into demographic groups — “I don’t blame people for saying, ‘Are you kidding?’”
Still, Yost pushed back on much of the criticism you hear of polls — for example, that younger voters don’t pick up the phone when pollsters call. Polling outfits connect with voters in multiple ways, and those strategies have adapted with time. (And in case you’re wondering, younger voters polled expressed less concern about the war in Gaza than you might expect from those campus protests: As with other groups, they’re more concerned about economic factors such as inflation.)
Too, it may not be polls that are muddled. Voters don’t always follow the news, and they don’t always have to make sense. Ask voters in your own family and see! Or consider the fact that one in six people served by the Times said they blamed Joe Biden for the overturning of Roe.
But even taking all that into account, Yost acknowledges that while “polls do a lot of things well, the thing they do least well is predict the future.”
Partly that’s because making such predictions is a two-step process: First you have to figure out what people think, and then you have to guess who among your sample will actually do something about it.
Because people often lie about their intention to vote, pollsters look at a number of factors to predict who will show up — and those predictions are more art than science. Years ago, the Times gave the same raw survey data to four different respected pollsters and asked them to compile the results. They came back with four different predictions — again, using the exact same set of responses — with outcomes varying by 5 percentage points.
We’re not seeing that kind of variability right now: Since March, polls of Pennsylvania have almost uniformly shown Trump with a lead in the low single digits.
But another consistent finding is that disaffection with both major party candidates is high, and voters want other options.
“What leads to volatility is that 20% of voters are saying they wish they could choose someone else,” Yost said.
Which means one consistent finding within these polls is that the polls may not be able to tell you who is going to win … even though that’s why you are looking at them in the first place.
That’s one reason Yost laments, “I wish we didn’t have to ask the horse-race question. But if we didn’t ask it, no one would pay attention at all.”
Still, he acknowledges the risk of paying too much attention: When an outlet like the Times devotes multiple stories to a poll over multiple days, “We could have a discussion about whether they are reporting the news or making it.”
Part of the downside is that media outlets devote resources to reflecting public opinion instead of informing it. (Polls aren’t cheap!) But in a close race like this, the fact that polls can affect a race is maybe the best case for why they matter: They offer a data point on what is important to the people whose leaders we are choosing.
“If you truly believe this race couldn’t go either way in Pennsylvania, then God bless you,” said Patrick Murray, who directs polling at New Jersey’s Monmouth University. When it comes to young voters and other groups, he said, “There's no question that Biden is softer than a typical Democrat has been in the past 10 or 15 years. That isn't a prediction for Election Day: It's a statement of where he needs to do some work.”
For example, the fact that abortion rights is a key issue — and that a sizable chunk of the electorate seems confused about who has weakened those rights — is a big reason Democrats and their allies will be talking up the issue. If today’s polls prove inaccurate this fall, in other words, it may be because people did take them seriously.
In any case, Murray said, “Saying a poll is wrong today, six months before the election, is the same as saying the Pirates’ record today is wrong, if they happen to go on and win 90 games this season.”
To be clear: No one is predicting that will happen by this fall, either. But if you want to help shape the direction of the country — which is what all this is supposed to be about — all you need to know is this contest is anyone’s ballgame.
“You shouldn’t be focusing on someone being up 3 points,” Murray said. “With anything less than a 5 point margin, a poll is saying it could go either way.”
If you can keep that in mind, you’ll sleep better. And you’ll have a longer list of things to do when you wake up.
