The prospect that Donald Trump will be restored to power by the great and good American people, or even the less great and good Electoral College, is disturbing enough that it has spawned a wave of denial. How could people want go through that again? A species of this belief is a persistent refusal to believe polls showing President Biden is really that unpopular and that Donald Trump is really winning.
The argument has various strands but goes like this. Despite alarming polls, Democrats beat Trump in 2020, defied predictions of a “red wave” by winning the 2022 midterm elections, and have won a series of off-year and special elections since. The polls may show Trump winning, but the election results say otherwise.
That belief is endorsed by a New York Times profile of Simon Rosenberg, which begins like so:
Simon Rosenberg was right about the congressional elections of 2022. All the conventional wisdom — the polls, the punditry, the fretting by fellow Democrats — revolved around the expectation of a big red wave and a Democratic wipeout.
He disagreed. Democrats would surprise everyone, he said again and again: There would be no red wave. He was correct, of course, as he is quick to remind anyone listening.
It is true that the conventional wisdom predicted a Republican blowout. But it is not true that the polls did. The polls were highly accurate, as numerous post-election reviews found — see this, or this, or this. To the extent the conventional wisdom was off, it was because reporters and pundits ignored the polls and relied on history (the president’s party always loses seats in its first midterm election) or vibes.
There is a small exception here. Toward the end of the cycle, Republicans flooded the zone with partisan, low-quality polling that tended to push the average of polls in a Republican direction. Polling analysts noted this at the time it was happening: “This year, a wave of polls from Republican-leaning firms is driving the averages,” wrote Nate Cohn on the cusp of the 2022 elections.
Of course polls aren’t always accurate. They can err in either direction, and both kinds of errors have occurred in recent elections. In 2016 and 2020, polls underestimated Trump. In numerous off-year and special elections since then, they have underestimated Democrats (or Democratic-supported ballot initiatives, such as measures supporting abortion rights).
Do any of these things suggest the polls are underestimating Biden right now? No, not really. You can tell a consistent story about polls and results by observing that Democrats have started performing better with educated voters, who vote reliably even in low-turnout elections, while Donald Trump has attracted more working-class voters who only turn out in presidential elections (or, alternatively, when he is on the ballot.) That narrative explains both Trump’s over-performance in his two races, and the Democrats’ strong performance in the midterms (where they did not suffer a drop-off) and extremely strong performance in special elections and referenda.
The profile of Rosenberg notes that he “even has a Substack newsletter offering insights and daily reassurance to his worried readers — ’Hopium Chronicles,’ the name taken from what the pollster Nate Silver suggested he was ingesting back in 2022.” That links to a post-2022 election piece in the Times headlined, “The ‘Red Wave’ Washout: How Skewed Polls Fed a False Election Narrative.”
Casual readers may get the impression that this shows the 2022 polling was badly wrong. But it focuses instead on the effect of partisan Republican polling on the journalism narrative. It’s important to understand that Biden’s trouble right now is not with Republican polls. It’s with credible polls produced by mainstream-media organs.
There are two credible reasons for Democrats to feel optimism. The first is that the improving economy finally seems to be working its way into the public consciousness. Economic sentiment has started to turn upward, albeit from a dismal place.
The second is that it’s still early. The election doesn’t take place for seven more months, and polling right now doesn’t tell you as much as it does in October (and even then, there’s a margin for error). Indeed, Biden’s polling seems to have gained slightly in the last couple weeks.
But simply counting on the course of events or the innate awfulness of Donald Trump to have a predictable effect on Biden’s polling is to count on good luck. There’s no reason to believe the events of the next seven months will help Biden more than Trump. And if you’re telling yourself a story that the polls have some predictable skew that allows you to mentally add a few points to Biden’s total, you’re choosing hope over data.
