Poll Position: Where Clinton, Trump Stand on Election Eve 2017

The question of who will win the White House will finally be answered on Election Day, but it's good to understand where public opinion is at this moment and what it means for Tuesday night.
To help with that, here are the answers to four questions about the polls.
What do the polls say now?
As of Monday morning, polls show a modest lead for Hillary Clinton. She's ahead of Donald Trump in the RealClearPolitics four-way national average (which also includes Libertarian Gary Johnson and Green Party candidate Jill Stein) by 2.2 percentage points and by 1.8 points in the two-way average.
Clinton also leads in state polls. If every state voted according to its RCP average, she would win with 297 electoral votes to Trump's 241, easily surpassing the needed 270.
Does either candidate have momentum?
A few weeks ago, Clinton led in national polls by as much as seven points. And while her lead has declined, she has nonetheless led in the RCP average for the past 98 days - since the nominating conventions came to a close. She has benefited from three strong presidential debate performances and Trump has had to deal with the release of an audio recording of him making lewd comments about women.
So it's worth asking -- what has changed of late?
It appears the effects of the debates and that recording (if it had any influence on voter preference) have worn off and Trump has gained support among reliably Republican voters. This shows up in the RCP average -- Clinton had about 45 percent of the vote two weeks ago and her standing now is virtually unchanged. Trump, on the other hand, was at 39.9 percent in the RCP average two weeks ago, but he now sits at 42.7 percent in the four-way race. No major party presidential candidate in 2012, 2008 or 2004 improved their national standing by more than two points in the two weeks prior to the Monday before Election Day, yet Trump has improved his polling by about three points in that interval.
Nonetheless, his short-term trend may not be significant enough.
Does momentum or current polling signal definitively that either candidate will win?
No.
The candidate who leads in pre-election polls typically has the advantage heading into Election Day -- which makes Hillary Clinton the favorite (but not a lock).
For example, Obama led in the RealClearPolitics average by 0.9 points heading into Election Day 2012 and went on to win the White House by 3.9 points. The RCP national average also put Obama ahead by 7.3 points in 2008, and he won by 7.6 points. George W. Bush led by 1.5 points in the 2004 RCP national average of polls, and he won by 2.4 points. I could go into elections that predate RCP, but the point is that a polling lead typically indicates an Election Day advantage. If Clinton and Trump perform at their current polling levels (or if Clinton over-performs) she will win.
There are still possible paths to a Trump victory, however.
It's possible that there is some late movement toward him that the polls haven't fully picked up on. Good pollsters often survey voters over multiple days, so if movement occurred very late in that time frame it's possible that public pollsters won't fully detect it before Nov. 8.
The polls could also systematically underrate Trump by misestimating the composition of the electorate. In the 2014 Senate elections, polls in the final week underestimated Republican candidates by about three points on average. And in the 2012 presidential election, national polls showed Obama leading by one point, and he won the popular vote by four points (although state polls were more accurate than national polls in that race). So it's possible that the polls could be off and Trump could win the presidency. But it's impossible to know ahead of time if the polls will err -- and if they do err, it's impossible to forecast whether they might be biased toward Trump or Clinton.
In other words, Clinton has kept Trump from catching up to her in the polls so she is still the favorite to win -- but there is some uncertainty. That's why predictive models at sites like FiveThirtyEight, the Upshot and the Huffington Post (despite differences in assumptions and how they're built) all give Clinton a win probability over 50 percent (64 percent, 84 percent and 98.1 percent, respectively) but don't say that a Democratic victory is a 100 percent certainty.
Which states should I watch on Election Day?
It's important to watch the swing states. RCP rates close states as tossups, but a few of those stand out as especially important.
It's important to watch states that are both reasonably close and have many electoral votes -- such as Ohio, Pennsylvania, Florida, North Carolina and Michigan. North Carolina and Florida are very close -- Clinton leads by one point in Florida and Trump leads by 1.5 in North Carolina. If Trump were to win both of those states as well as Ohio, Iowa, Georgia and Arizona, he'll be at 259 Electoral Votes. If Trump managed to add Pennsylvania or Michigan (he currently trails in both), he would win. But if Clinton wins Michigan, Florida and Pennsylvania along with every state we rate as "Likely Clinton" and "Leans Clinton," she will become the next president.
From a more academic perspective, it's interesting to watch states like Utah, Arizona, Iowa and Maine. Trump has the lead in both Utah and Arizona, but his expected win margin in both states may be less than Mitt Romney's in 2012 or John McCain's in 2008. Yet Trump has pulled Iowa and Maine to the right of where they were in previous cycles.
This illustrates one of the central tradeoffs of Trump's candidacy. His policy and persona appeal to non-college-educated whites (who make up a significant proportion of the population in Maine and Iowa) while turning off Hispanics (in states like Arizona) and some traditional Republican groups like Mormons (in states like Utah) and college-educated whites.