In a horse race, it’s (somebody) by a nose

Polling guru Nate Silver says this is the closest presidential election in his 16 years of analysis, and a spate of recent polls bear him out, with Vice President Kamala Harris clinging to a tiny but stable lead in most of them but also well within the margin of error in those same polls.

Silver, the founder of the data site FiveThirtyEight (he no longer works there) developed a system that successfully picked the winner in 49 states in the 2008 presidential election, and went on to predict the 2012 and 2020 elections with more accuracy than most analysts.

As with other analysts, though, Silver missed in 2016.

Currently, Silver gives Harris a 56-percent probability of prevailing in the electoral college on election day, though several clarifications should be made. First, Silver’s number is not a horse-race number — it does not mean she will get 56 percent of the electoral college vote — it’s a betting number, meaning in his models she prevails 5.6 times out of every 10 times the numbers are run through the model. It’s also important to note that that is actually a toss-up in Silver’s world because, he says, a one-percent swing in overall public opinion polls can translate into an 8-percent swing in the probability models.

Still, he says, with the race settling down into a constant rhythm, Harris is well positioned headed into election day, barring some unforeseen event or October surprise.

“Current models say it’s basically a toss-up but, if offered the choice, you’d slightly rather have Harris’s hand to play,” Silver posted on X this past week. “Anybody projecting more certainty than that in either direction is getting way ahead of the data and selling you snake oil.”

Certainly the race is as close as can be, especially in the battleground states. Harris’s most likely path to victory is to hold the fabled “blue wall” states of Wisconsin, Michigan, and Pennsylvania, which, if Trump prevails in North Carolina, Arizona, and Georgia, yields a Harris win of 270-268.

Each side faces potential obstacles. For Trump the matter might be that he simply is not running as well as the polls indicate. For years the conservative media has made much of the fact that Trump scores better on election day than the polls predict. That is true but there’s a counter-proposition—that since 2018 Republicans in statewide elections across the country have underperformed their polling. 

The supposed “red wave” predicted by conservatives in the 2022 mid-terms is the prime example. Not only was there no red wave, the Democrats won the elections overall. In this scenario, Trump’s outperformance of his polling in 2020 is an outlier — a race he nonetheless still lost — and if there is polling error, it probably favors Harris.

The problem for Harris is if the polls turn out to be accurate. That’s because in battleground states the race is razor-tight and the most recent polls show them tightening. 

The polls

A new Marquette Law School poll this week gives Harris a little breathing room in Wisconsin, but not much. In the poll, among both registered voters and among likely voters, Harris notched the support of 52 percent of respondents to Trump’s 48 percent. The results include initially undecided voters who were then asked whom they would vote for if they had to choose. 

“In a multi-candidate race, Harris also leads by 4 percent, 48 percent to Trump’s 44 percent, while Robert F. Kennedy Jr. receives 3 percent, the Party for Socialism and Liberation’s Claudia De la Cruz 0 percent, Libertarian Party’s Chase Oliver 1 percent, Green Party candidate Jill Stein 1 percent, Constitution Party candidate Randall Terry 1 percent, and independent Cornel West 0 percent, among registered voters,” the poll stated. “Among likely voters it is Harris 49 percent, Trump 44 percent, Kennedy 3 percent, De la Cruz 0 percent, Oliver 1 percent, Stein 1 percent, Terry 1 percent, and West 0 percent.”

Kennedy has dropped out of the race and endorsed Trump, but his bid to get his name removed from the ballot failed in the state’s liberal Supreme Court.

That survey was conducted Sept. 18-26,, interviewing 882 Wisconsin registered voters, with a margin of error of +/-4.4 percentage points, and 798 likely voters, with the same margin of error of +/-4.4 percentage points as for registered voters. 

When asked whom they thought is likely to win the presidential election in November, 50 percent predicted Harris would definitely or probably win, while 39 percent said the same of Trump, the poll reported: “There has been an increase in the perceived chance of a Harris win since she entered the race in July, along with a decline in the perception of Trump’s chances.”

But a New York Times/Sienna College poll has the race in the Badger state and in neighboring Michigan closer than that. The latest survey puts Harris up by only 1 percent in Michigan, 48 percent to 47 percent, among the state’s likely voters and by only 2 percent in Wisconsin. Both of those numbers are within the margin of error.

Then, too, a flurry of polls from southern and southwestern states has Trump on top, wiping out Harris’s previous advantage.

According to the Swing State Project, a collaboration between The Cook Political Report and two pollsters, BSG and GS Strategy Group, voters across seven swing states (Arizona, Georgia, Michigan, North Carolina, Nevada, Pennsylvania, and Wisconsin) gave Harris only a 1-point 49-48 lead. Harris led by between 1 and 3 points in five of those states, within the margin of error while in North Carolina the two were tied.

Trump was leading by 2 percent head-to-head in Georgia.

Polls from Quinnipiac and Emerson were not so sanguine for Harris, who is performing better in national polls than in state-by-state polls.

“Five weeks before Election Day, the two key Southeast swing states of Georgia and North Carolina show a mixed picture as former President Donald Trump is ahead of Vice President Kamala Harris in Georgia and the race in North Carolina is too close to call,” the Quinnipiac University poll stated about its survey of likely voters.

“All eyes are on the South as Georgia and North Carolina, turbocharged by 32 electoral votes between them, can make or break the two candidates in a race that looks to be leaning Trump’s way in Georgia at the moment and churning toward a tight finish in North Carolina,” said Quinnipiac University polling analyst Tim Malloy. “On issues, as the presidential horse race thunders toward November 5th, Harris confronts a troubling Trump trifecta: he leads her on the economy and immigration and has the edge when it comes to who would best handle a national crisis.”

A new Emerson poll shows Trump and Harris tied 48-48 in Pennsylvania, while in Arizona, 50 percent support former Trump while 47 percent support Harris. In North Carolina, 49 percent support Trump, while 48 percent support Harris.

In another troubling sign for Harris, statewide Democratic candidates in North Carolin and Georgia lead Republicans, outperforming Harris on the ballot.

Unless something shakes up the race, the nail-biting is likely to continue,

“The forecast has entered a really steady phase,” Silver said. “Why? Lots of polling, not that much happening. So you can have a good Harris day or a good Trump day and it doesn’t move the toplines much.”