TORONTO -- Recent polling data suggests that Democratic hopeful Joe Biden has a significant and steady lead over U.S. President Donald Trump in both the popular vote and some key swing-state surveys heading into Tuesday’s election.

But if Hillary Clinton’s 2016 electoral upset is any indication, Biden winning the White House is far from a sure thing based on national poll numbers alone.

According to pollster Nik Nanos, there may still be a path to victory for Trump through the Electoral College. And, like in 2016, the national polls may be doing a poor job reflecting the state-by-state polls the very factor pollsters failed to account for in 2016.

“Most of the major pollsters were within the margin of error, predicting (Hillary) Clinton to win the popular vote” in 2016, Nanos explained on Friday’s episode of CTV’s Trend Line podcast.

“The failure was actually in the most important thing, which was the Electoral College, because it's not a proportional democracy in America. It is the Electoral College that actually picks the president.”

As Nanos explains, because Clinton was consistently ahead in popularity for most of the campaign—just as Biden is now—it was wrongly interpreted that would translate to a win in the Electoral College.

As 2016 demonstrated, national leads aren’t always an accurate snapshot of the state-by-state horse race between candidates.

While most of the pollsters in 2016 were within the margin of error predicting Clinton to win the popular vote, several factors played into Trump’s surprise win.

According to Pew Research, polls in some key Midwestern battleground states overstated support for Clinton and understated support for Trump, after having too many college graduates in their samples. Undecided voters in key states also swung toward Trump in the race’s final days when there were no polls in the field.

In today’s race, Trump trails Biden by an average of eight points nationally, and is behind in every important battleground state—a very similar margin compared to 2016.

Nanos points to sites like 338Canada.com, which creates statistical models of electoral projections based on polling data. As of Oct. 29, Biden has a lead of just over 9.0 percentage points in the popular vote.

“What people don't realize is that for these things, forty thousand, fifty thousand simulations are run. And then in those simulations, Biden, for example, wins 80 per cent of the time. But Trump still wins 17 per cent of the time,” Nanos explained.

“It’s kind of like the weather… we read the weather and when there's zero chance of rain it's usually right. But as soon as there's any chance of rain, five percent, 10 percent, 15 percent, 20 percent, that means it can rain.”

In other words, an 80 per cent chance of Biden winning doesn’t mean it’s a guarantee.

“It means that he's more likely on the balance of probability, but that Trump can still win the Electoral College vote and the election,” said Nanos.

HOW DO YOU POLL FOR AN ELECTORAL COLLEGE WIN?

Although Biden has a narrow advantage in critical battlegrounds that could decide the race, there are several states where Trump is ahead or within the margin of error for an Electoral College win.

In 2016, Trump lost the popular vote but had a comfortable margin in the Electoral College, where each state gets a number of votes based on their population.

But polling for an Electoral College win has its challenges, especially because high-quality state-level polling remains sparse.

“When you look at the in the national polling scene and the mainstream media, that they hire major companies that have great track records and they do a good job at the polling and predicting national party support,” said Nanos.

“But when you get to the state level, there's a diversity of firms and organizations commissioning polls.”

As Nanos explains, accounting for state projections is especially difficult during this year’s race, noting that Republican strongholds, like Florida, are turning into toss up states.

“We even have the Democratic vice presidential candidate going to Texas—the Democrats haven't won Texas in a long time. But the gap is narrowing even in that state. So, there are toss up states where the numbers are within the margin of error,” he explained.

“They're based on state polls that I will say are wobbly. And we should be very careful in some of these Electoral College predictions because they're very susceptible to a quick change if just one or two percentage points swing in one direction or another.”