Advertisement

Subscriber-only Newsletter

By David French

I’m having the strongest sense of déjà vu.

In 2012, I was a Republican partisan. This was when I was a conservative constitutional litigator and occasional Republican Party activist, before my journalism career. I’d helped form a group called Evangelicals for Mitt all the way back in 2005, hoping to persuade evangelical Republicans to support a Mormon for president. We’d done our small bit to help push Romney over the finish line in the primaries, and most conservatives seemed convinced that he should win. Republicans had swept the 2010 midterms, the unemployment rate remained high, and Barack Obama’s approval rating was below 50 percent.

But there was a problem: The polls were bad. Almost all of them showed Romney losing to Obama, and so conservative media started a movement to unskew the polls. There was even a website created, Unskewed Polls, that purported to fix the polling errors, and unskewed polls showed Romney winning.

Conservatives believed that pollsters were deliberately undercounting Republican votes to discourage Republican voters and sway the results of the election. So to unskew the results, they reweighted the samples to include a higher percentage of likely Republican voters. Conservatives thus created a parallel universe where Romney was leading, and many people at most senior levels of the campaign believed that mainstream polls were wrong and Romney would win — including, reportedly, even the candidate himself.

I was in Boston at the Romney party on election night, and when Fox News called Ohio for Obama, there was a palpable sense of shock, followed almost immediately by denial. I remember fielding emails in the days and weeks after the election from angry Republicans who wondered, even then, if Obama had cheated.

I thought of 2012 when I read in an Axios report this week that “President Biden doesn’t believe his bad poll numbers, and neither do many of his closest advisers.” That belief isn’t absurd on its face. After all, polling is difficult, and there have been a number of recent polling misses. As Axios notes, Donald Trump overperformed his polling in 2016 and 2020, and Democrats overperformed in 2022. And the sampling process is tricky as well.

For example, as The Times’s Nate Cohn noted on X, there is a stark difference between high-propensity voters, who are more likely to support Biden, and low-propensity voters, who are more likely to support Trump. Some percentage of those low-propensity voters will turn out. The key question is how many. But it’s one thing to criticize any given poll, and it’s another thing entirely to dismiss aggregated results, taken over months, that show the same thing: a race that is incredibly tight, far too tight for Biden’s comfort.

The purpose of this newsletter isn’t to adjudicate the polling dispute but to show an example of how the partisan mind works and how partisans process negative information. I could use any number of other examples. In a column last week, my colleague Ross Douthat rightly observed that........

© The New York Times