Is the Blue Wall shakier than it looks?

A dissent.

Last night, I noted that the election is almost certain to come down to the Blue Wall states — Michigan, Pennsylvania and Wisconsin.

Win all three and Kamala Harris will be president. Win only two and things get murkier.

My conclusion was two-fold:

1. These three states have voted for the same candidate in every election since 1988

2. The Washington Post polling averages suggest that Harris is ahead in all 3 right now

My friend Damon Linker, who writes the indispensable “Notes from the Middleground” Substack, sent me a note in response — in which he made the case that Harris’ edge in the Blue Wall might be considerably more shaky than I presented it.

Damon said I could share his analysis with my readers. Here’s what he wrote:

Good post, Chris. But it throws me back on the question of how accurate those incredibly close polls are.

In 2020, Real Clear Politics had the following polling averages on Election Day // Final result:

PA: Biden up 1.2 // Biden wins by 1.2 [ON THE NOSE]

MI: Biden up 4.2 // Biden wins by 2.8 [1.4 points short of the polling average]

WI: Biden up 6.7 // Biden wins by 0.7 [ 6 POINTS SHORT OF THE POLLING AVERAGE!!!!]

So, a HUGE discrepancy in accuracy across the Blue Wall.

How does Harris look today using your average? // And using RCP?

PA: Harris up 2 // Harris and Trump TIED

MI: Harris up 2 // Harris up 1.3

WI: Harris up 3 // Harris up 0.8

My takeaway: If the polls are exactly correct or biased against Harris, she wins. But if there's anything close to the anti-Trump bias of 2020, she definitely loses WI and will be extraordinarily close in PA and MI.

This gets at a broader question that eats away at me: Are the polls under-measuring Trump’s support?

As Damon notes, polling — nationally and in swing states — in 2016 and 2020 consistently underestimated Trump’s support. (This is a very good read on how — and where — polls missed in 2020.)

Pollsters now have those two elections to work off of — and maybe they have figured out how to adjust their data for that hidden Trump vote.

“We don’t always see the misses in the same direction,” Chris Jackson, the senior vice president of public affairs for Ipsos, said recently. “I can tell you that the polling industry has done substantial changes to how we do our surveys to try to account for what we think was driving those errors in 2020. So while there undoubtedly will be errors in the future, they’re probably going to be driven by different things and go in different directions.” 

Maybe! But also maybe not!

“Polling has really been seriously damaged since 2016,” Pennsylvania John Fetterman told The Hill recently. “And that’s one of the truths, is that Trump is going to be tough in Pennsylvania, and that’s absolutely the truth.”

I am not sure whether or not polling is “damaged.” But I do think that polling for Trump support is incredibly difficult — because there are plenty of people who won’t tell a random stranger that they are going to vote for the former president but definitely will. Or even people who lie to pollsters and say they are for Harris but really are going to be for Trump.

I think Damon’s point is a very good one. Given the closeness of the race, even a 2-3 point over-performance of polls by Trump likely means he breaks through the Blue Wall and wins the presidency.

Food for thought…