Great, and failed, expectations

Election forecasting as an exercise in humility.

I’ve played some role in NPR’s presidential election coverage for about a quarter-century now, and in all that time only one election really unfolded as the conventional wisdom expected. This helped me to understand that something is not necessarily true just because “everyone says” it is.

In 2000 I covered George W. Bush’s campaign, catching a plane with the candidate to his final rallies and back to his election day base in Austin, Texas. He was on a roll. Al Gore’s campaign seemed desperate at the end, and one of the last scenarios that Democrats discussed that day was that Gore might lose the popular vote, but still prevail in the electoral vote. The opposite happened.

In the early hours of election night, it seemed that George Bush had won the national popular vote, just as many polls said he would, which gave him moral authority to claim victory in the electoral vote. Only in later days did the full results come in, showing Gore up by half a million votes. This made no legal difference but did make a political difference. Bush ultimately prevailed on a Supreme Court ruling, but was seen from the start as a vulnerable president; protesters in Washington chanted, “The loser is the winner.”

In 2004, exit polls strongly suggested that John Kerry would win. He didn’t. Bush’s victory became clear by late in the evening, though it wasn’t affirmed until Ohio was called and Kerry conceded late the next morning.

In 2008, Obama’s landslide election unfolded as expected. The only real surprise was that Obama won my home state of Indiana, the sole occasion Indiana has voted for a Democrat for president in the past sixty years.

In 2012, Obama had a narrow edge, but many Republicans were utterly convinced the polls were “skewed” against Mitt Romney, and believed they were winning until the very end. Notably, given the developments of later years, Republicans broadly accepted this stunning result. Many honorably acknowledged the error.

Dean Chambers, who ran a popular website called “Unskewed Polling,” told Business Insider in an interview, “Nate Silver was right and I was wrong.” Silver was the analyst, then relatively new to the scene, whose polling averages and other statistics had reassured nervous Democrats that Obama was bound for victory.

In 2016, of course, Silver was one of the analysts who had Hillary Clinton favored to win, though he gave lower odds than others. His final forecast, still posted online, had Clinton winning in 71.4% of scenarios, rather than the 97% chance that others gave her. Silver later emphasized that he said Clinton was favored to win, not that she would. In retrospect (and also, I told people at the time) there was a kind of madness in the desperate certainty with which people said Trump would never win. The nominee of a major party can win. True then, true now.

In 2020, Biden was heading for a ten-point landslide that became a solid but comparatively narrow win in the swing states, not confirmed until the Saturday after Election Day.

In 2024 we have seen some Democrats effectively “unskewing” polls. Some have understandably dismissed the flood of partisan polls that Republicans have dumped into the race to affect the polling averages. Other Democrats have taken more of a leap by parsing the crosstabs of high-quality polls that show a very close race. (What is a crosstab, you ask? That’s the messy data behind a poll’s bottom-line result, and/or a word that some analysts use to suggest they deeply understand polling science. I will admit here that I am a regular consumer and not a practitioner of this science, or maybe art.)

Now here we are, and the conventional wisdom calls for an incredibly close race. It certainly could be.

Is there any way, really, to get an advance glimpse of the outcome? Some analysts are looking to North Carolina, a competitive state that seems likely to report its results early. Trump seems to be ahead slightly, while a Harris win would have big implications for the electoral map. Democrats are also looking at Iowa. A single, but respected, poll over the weekend showed Harris surprisingly in the lead there. Does that hold up? Just as important, Democrats have been competing hard in two Iowa Congressional races, and it’s a state where abortion is a big issue. If their investment pays off it may signal movement elsewhere.

I have left out of this account many other twists and turns of past years—election night news that seemed to point in one direction, only to have the final result turn out otherwise. There was the time when networks called Florida for Al Gore in 2000; or the minute when Clinton seemed to be comfortably winning Florida in 2016, until a new wave of returns changed everything. All I can tell you for tonight is that I’ll be trying my best to wait, and even to get a little sleep, before bringing you the best information we have tomorrow morning.