Among the four presidential candidates, RealClearPolitics puts Clinton ahead of Trump by about two points, within the margin-of-error. But HuffPost Pollster gives Clinton a three-point lead. Both models show Trump closer to Clinton than he was a month ago. But he gets no more than forty percent of the vote in either projection.
Beneath the narrowing of national polls, strange signals have emerged at the state level. Across the country, Clinton is competitive in traditionally Republican strongholds. Meanwhile, Trump could steal electoral votes from Clinton in places Barack Obama won in both his election victories.
Trump barely leads in must-win Ohio, but appears to have a chance at winning Iowa and Nevada. Obama won all three in 2012. But Clinton is competitive in Georgia — a state Mitt Romney won by eight points — and is close to Trump in North Carolina. Losing either state would doom Trump's chance of an electoral college victory.
Since 1984, Georgia has gone red every presidential cycle — except for when Bill Clinton won the state in 1992. A Clinton pickup of the state would be a major shift in the electoral map.
Many state polls also exclude Gary Johnson and Jill Stein. National polling shows the Libertarian and Green party candidates may peel millions of votes away from the two-party candidates. But because pollsters do not consistently include the third party candidates, it is difficult to gauge what impact Johnson or Stein may be having in any particular state.
A recent Fox News poll showed Johnson and Stein underperforming their national averages in swing states. But because these polls are not consistent, the trends in support are less reliable. According to RealClearPolitics, national surveys peg Johnson and Stein at 8.8 and 2.8 percent support.
The takeaway: The race has tightened and, unlike in the past, Trump's support does not seem artificial. But polling still gives Clinton an edge. With a clear majority of voters likely to tune in, all eyes will be on the polls following Clinton and Trump's first debate.