One reason for Mr. Biden’s strength in Arizona and Florida might be older voters, who represent an above-average share of white voters in the two states. On average, Mr. Biden leads among voters over age 65 by a margin of 53 percent to 44 percent nationwide, including a lead in every live-interview national poll reporting a result for the group. It is a substantial improvement over Mrs. Clinton’s six-point deficit among the group in pre-election polls in 2016.
Mr. Biden’s early strength among older voters is not easy to explain. It cannot be fully accounted for by the changing composition of each age group, although the ascent of the baby boomers into the oldest age cohort may be part of the reason, along with the gradual departure of the more conservative Silent Generation from the electorate altogether. Mr. Trump seems to have made gains among voters 45 to 65, or perhaps even younger, canceling out his losses among older voters over all.
Mr. Biden’s relative strength among older voters may also help counteract his expected weakness among likely voters, relative to registered voters, when pollsters apply likely-voter screens later in the cycle. In polling, Republicans usually fare about two points better among likely voters than registered voters, who tend to be relatively young and diverse.
This cycle, the coronavirus pandemic raises additional questions about the eventual turnout, particularly if it leads to widespread voting by mail. But no matter the method, Democrats typically find themselves at a turnout disadvantage, and it is doubtful that Mr. Biden will maintain the whole of his current polling advantage among likely voters. Even when Democrats benefited from a surge in turnout in well-educated suburbs during the 2018 midterm elections, Republicans fared better among likely voters than among registered voters.
Together, Mr. Trump’s relative advantage of one to two points among likely voters compared with registered voters — and his relative advantage of three and even four points in the tipping-point states — means that the typical national poll of registered voters is probably around four or five points worse for Mr. Trump than his standing among likely voters in the most pivotal states. Mr. Biden’s already narrow polling lead in states like Wisconsin, Pennsylvania or Arizona might be vanishingly small after a likely voter screen.
Of course, no one knows what American life will look like by the time of the election. Perhaps the country will still be in lockdown, saddled by 30 percent unemployment, and convinced that the president’s slow response cost lives and damaged the economy. Or maybe the country will be swept by euphoria as lockdowns are lifted a month or two ahead of the election and a liberated population sends its children to school, visits friends, goes to the park and enjoys double-digit G.D.P. growth in the third quarter.
The pandemic also could change how the election is administered, potentially yielding a novel turnout that’s impossible to predict at this stage. It undoubtedly has the potential to reshape the views of the electorate, even if it hasn’t done so yet.