By JACOB ROUSSEAU and SEAN MCGOEY

Capital News Service

Even in the most favorable of polls, the path to reelection for President Donald Trump seems discouraging.

An incumbent who was not far behind in the presidential race in March has seen Democrat Joe Biden’s lead extend to 10.3 points, according to FiveThirtyEight’s national average. The New York Times has had the former vice president ahead by an average of double-digit points in its recent polling averages.

Many observers — not to mention voters — are hesitant to trust those numbers, though, after some polls leading up to the 2016 election contributed to a widely held perception that Trump had almost no chance to defeat Hillary Clinton. That perception was reinforced by the October release of a devastating “Access Hollywood” video that captured Trump boasting of sexual assault.

In addition to general wariness about polls, the 2020 election brings with it its own set of distinct issues as well: a pandemic that has altered the way that millions of people are casting their ballots, concerns about states implementing voter suppression tactics and the possibility that the president’s campaign may attempt to claim victory before all the votes are counted.

But looking at state-level polling offers some important insights that can help people better understand the status of the race for the 270 Electoral College votes that will secure the White House for Biden or Trump.

FiveThirtyEight’s polling averages show Biden holding similar leads to Clinton in key swing states like Pennsylvania and Wisconsin two weeks out from Election Day. Polls also suggest that the former vice president is running well ahead of 2016 outcomes across the board, even competing in longstanding Republican strongholds like Texas and Georgia.

Perhaps most crucial to Biden in 2020 are the fewer poll respondents in nearly every battleground state who are undecided or planning to vote for a third-party candidate, a marked contrast to this point in the 2016 cycle.

Based on the polling averages, Biden would still hold a lead in enough states to secure a slim Electoral College victory even if every currently undecided voter broke for Trump in the election’s final days.

“Before the pandemic, the economy was great and it looked like a victory for Trump in not just Iowa, but nationally as well,” Timothy Hagle, professor of American politics and public opinion at the University of Iowa, told Capital News Service. “Then the economy tanked and that has really affected everyone. I think that Trump may still have a shot of winning Iowa, but it is probably closer than it was in 2016.”

Based on the data, a Trump victory would need a major election-changing event similar to the revival of the Clinton email controversy late in the 2016 election that likely contributed to her defeat.

In that contest with Trump four years ago, Clinton was showing significant leads in polls until then-FBI Director James Comey’s letter to Congress that said the FBI had “learned of the existence of emails that appear to be pertinent to the investigation.”

According to FiveThirtyEight, Clinton was showing leads on the day of the election in some of the major swing states, including Wisconsin, where she led by more than 5 points but lost the state by nearly a full percentage point. That was a swing of more than 6 points.

It was not just in Wisconsin that election day polls had state political snapshots incorrect: swings of more than 4 percentage points in Michigan and Pennsylvania paved the way for the Trump campaign’s eventual upset victory.

Swing-state polls shifting in Trump’s favor after the Comey letter demonstrate that a major news development late in an election cycle could potentially sway enough voters to upend polling forecasts.

But Michael Hanmer, professor of government and politics at the University of Maryland and research director at UMD’s Center for Democracy and Civic Engagement, stresses that differences in polling snapshots from the actual election outcome don’t themselves indicate that previous polls are incorrect.

“I would say that trying to predict the outcomes is probably not the best way to think about polling at all,” Hanmer told CNS. “It’s really just a snapshot in time, and a lot can change from when the poll was done to the election.”

Presidential election forecasting incorporates these polling snapshots with other factors such as state demographics, economic data and previous election results in an attempt to predict the outcome of a future election.

After seeing the contrast between the pre-election polls and actual results in 2016, Sabato’s Crystal Ball — produced by the University of Virginia Center for Politics — changed its strategy for the 2020 election to interpret national polls in a way that is more favorable to Trump.

“Everyone’s a little burned from 2016, so we add maybe a point or two in Trump’s favor for every one of these polls,” said J. Miles Coleman, associate editor of the Crystal Ball, which uses state-level polling and state voting history to determine each state’s likely electoral outcome.

Coleman acknowledged that overcompensating for the so-called “Trump effect” could lead to their forecasts overestimating Trump’s chances.

Even if Trump were to win all 85 of the electoral votes in states that the Crystal Ball currently categorizes as “toss-ups,” he would still fall short of 270 electoral votes.

As the model projects, the states that are at least “leaning Democrat” would be enough to deliver Biden 290 electoral votes. Inside Elections, a nonpartisan analysis of the presidential election, currently has 319 electoral votes in favor of Biden — including 213 votes that they have deemed “solid Democrat.”

Coleman suggested that even if Biden’s position isn’t as strong as some of his biggest polling leads would seem, he may have enough cushion to survive some late tightening.

“I am somewhat skeptical of the national averages,” he told CNS. “But if Biden is up by 10 points right now, he is probably out of that danger zone.”