In the 2016 election between Hillary Clinton and Donald Trump, the polls got a lot wrong. They were particularly wrong when it came to states where white voters without a college education were a dominant electoral force. In Michigan, for example, the polls said Clinton would win by four points. Instead, Trump won by two points. In Iowa the polls suggested Trump was ahead by one point, but he ended up winning it by eight. In Wisconsin, Clinton was projected to clear the state by five points. She ended up losing by eight points.
By 2020 pollsters were publishing assessments of their own 2016 failings, assuring readers they had made amendments to their models. But the effects of this soul-searching were negligible. They had Biden ahead in Wisconsin by seven points in 2020, for example. He ended up scraping through by less than one point.
Below I have considered what might happen if the polls get it as wrong as they did in 2016. If we factor for previous errors then Trump could be four points ahead in Wisconsin, four in North Carolina, two in Pennsylvania, two in Michigan and one in Arizona. Only in Georgia and Nevada would the race still be neck-and-neck. In this scenario, Harris has not done enough to win.
But I suggest that in 2024 the polls have sufficiently self-corrected. In 2016 and 2020 enthusiasm for Trump was underestimated. In the rust-belt, for example, Trump's expected vote was five to ten per cent lower than what he ended up winning in 2016. The mistake here is that the polls routinely scored Republican voter enthusiasm as far lower than it was.
Estimated Democrat turn out would have to drop precipitously this year to see mistakes on the same proportion as 2016 and 2020. Again, this is implausible but not impossible. But if the polls have accurately gauged enthusiasm for Trump this time around, it's very good news for Kamala Harris.
Content from our partners
Related
