There’s been a lot of talk about Clinton trailing in Michigan and Wisconsin. Pennsylvania may be a far bigger problem. Clinton is up by just 4 percentage points, and that lead is slipping as more of the vote comes in from outside the Philadelphia media market. Clinton cannot lose Pennsylvania and win the election.

ABC News has called Colorado for Clinton, her third swing-state win, in addition to Virginia and New Mexico. So she’s holding up just fine in diverse, wealthy states. But she’s lost Ohio and is in profound trouble in Michigan and Wisconsin, perhaps along with Pennsylvania. Her weakness in the Midwest could wind up costing her the Electoral College.

As Andrew noted a few minutes ago, Clinton seems to be struggling in Michigan, a state where she was favored but where there are a large number of manufacturing jobs. As of September, about 14 percent of nonagricultural jobs in Michigan were in the manufacturing sector, the third-highest share of any state. The two above it: Indiana, where Trump won comfortably, and Wisconsin, where Clinton is in more trouble than expected.

Nine states are deciding on marijuana legalization ballot initiatives today. We’re spotlighting each during the course of the day. California’s polls close at 11 p.m. Eastern.

If the ballot measure in California passes, adults age 21 and older will be able to possess up to 28.5 grams of marijuana or 8 grams of concentrated marijuana, as well as grow up to six plants and consume marijuana it privately. Medical marijuana is already legal in California.

If the measure passes, it will create two new taxes: one at $9.25 per ounce for flowers and $2.75 per ounce for leaves, with exceptions for certain medical marijuana sales; the second would be a 15 percent tax on the retail price of marijuana. Revenue from these taxes would be spent on drug research, treatment and enforcement; health and safety grants addressing marijuana; youth programs; and preventing environmental damage resulting from illegal marijuana production.

According to the 10 state polls we’ve seen this year, the measure looks likely to pass easily.

Back when he was trying to win the Democratic nomination, Bernie Sanders repeatedly said he was more electable than Clinton, citing hypothetical polls pitting each of them against Trump in head-to-head matchups. There’s no way to know if Sanders’s lead would have held up if he’d won the nomination and faced the full force of opposition from Trump and the Republican Party, but some Sanders supporters must be wondering if their favored candidate would be holding up better today, considering what was perceived to be his appeal to at least some of what has become Trump’s general-election constituency.

Looking at the map of Wisconsin, Clinton is in big trouble. Much of the vote from Milwaukee is in, and Clinton is still down by nearly 3 percentage points statewide. If Trump wins here, Clinton is in big trouble nationally.

NBC News just projected Colorado for Clinton. That’s a necessary win for her, but it isn’t sufficient.

When Florida voters legalized medical marijuana tonight, it became the 26th state (plus D.C.) to legalize or decriminalize the drug. That means over half of all states in the U.S. have made medical marijuana legal. Now activists are looking toward California, where polls have not yet closed, to sway societal attitudes. The state could be massively important for the movement to legalize marijuana. The population of California is larger than the combined population of all seven other states (Maine, Massachusetts, North Dakota, Arkansas, Arizona, Montana and Nevada) whose marijuana ballot measures have yet to be called.

ABC News just called Virginia for Clinton, which improved her odds in our forecast. As a reminder, though, Clinton’s problems aren’t in the called states so far — they’re in the numerous uncalled states where Trump is either favored with most of the vote in (as in Florida) or states like Michigan where pre-election polls favored Clinton but the actual result is now too close to call.

Right now, Trump leads in both Michigan and Wisconsin. If those leads hold, I don’t really see a path for Clinton to win. If, however, Clinton wins those two states, she’ll be on track to win. If she loses one of them, she still has a path that includes Arizona, where the race is tight. Even if she does win either Michigan or Wisconsin, though, Trump is the favorite.

The race in Michigan has tightened and, according to our live forecast, it’s now too close to call. Why? We’ve talked a fair bit about the divisions emerging between college-educated and non-college-educated voters in the country, and in Michigan, voters with a college education have voted for Clinton over Trump 50 percent to 44 percent, while non-college-educated voters are going for Trump, 48 percent to 45 percent.

But the race is incredibly close there because voters without a college education account for nearly six in 10 voters. Preliminary exit polls have shown that while Clinton is winning union households by a 16 point margin, that’s down from previous elections, where Democrats typically win that vote by about 20 points or more.

Maine is shaping up to be an interesting case study in the popularity of the minimum wage. The presidential race there remains too close to call, as does a ballot initiative on legalizing marijuana. An initiative to require background checks to purchase firearms is currently losing in a close vote. But based on preliminary results, it looks like the state will approve an initiative to raise its minimum wage to $12 an hour by 2020. That shouldn’t be too surprising: Polls consistently show support for raising the minimum wage, even among Republicans.

Repeating myself a bit, but Clinton is still the favorite to win the popular vote, according to the Upshot’s model, despite now having become a fairly clear Electoral College underdog. Our pre-election forecast had shown about an 11 percent chance of such a split in Trump’s favor.

The race is very close in Michigan. One motivating factor in the state is the decline of manufacturing jobs, particularly in the automobile sector. Although Clinton was favored to win the state by FiveThirtyEight’s forecast and many others, Trump has touted a message that could appeal to many voters there: International trade has harmed the county. Sure enough, exit polls indicate that 50 percent of Michiganders agreed that trade with other countries would “take jobs away” from the U.S. Only 31 percent thought trade “creates more jobs.” And among Trump supporters, a whopping 65 percent had a negative view on trade.

Could Clinton win Arizona to salvage her Electoral College chances? It’s not impossible — her numbers are holding up reasonably well in the western part of the country so far, and she’s down by only 2 to 3 points in votes counted so far. But the Upshot’s model thinks the remaining vote there is more favorable to Trump.

There’s not enough information to call any of the races in Wisconsin yet, including the competitive Senate race between incumbent Ron Johnson and former incumbent Russ Feingold. Preliminary exit polls, though, reveal the complexity of Wisconsin politics. Trump did poorly in the state’s fairly late primary, after failing to win over important Republicans like Gov. Scott Walker, U.S. House Speaker Paul Ryan (who represents the state’s 1st Congressional District), and influential talk radio hosts like Charlie Sykes. However, Trump’s anti-trade message seems to have resonated with a substantial portion of Wisconsinites: Nearly half of Wisconsin voters (49 percent) reported feeling that trade with other countries takes U.S. jobs, and they preferred Trump by a nearly 2-1 margin. However, Trump’s other signature talking point, immigration, has a tougher audience in America’s Dairyland: According to the early exit polls, 57 percent of voters there think that immigrants do more to help the country than harm it. Those voters supported Clinton, 63 percent to 28 percent.

We’ve designated Wisconsin and Pennsylvania as too close to call also. I’d turn first toward betting markets, though, which see Trump as the Electoral College favorite.

The presidential races in Wisconsin and Pennsylvania are too close to call. Our model now gives Clinton a 51 percent chance of winning overall; Trump now has a 48 percent chance.

The AP has called the senate race in my home state of Iowa. Chuck Grassley, the state’s senior senator, is headed to a seventh term.

We’ve warned for a while that exit polls should not be used to project winners. Tonight is a perfect example of that. The races are far closer than the exit polls suggested.

Two nonbinding votes in the Florida Keys over whether to release genetically modified mosquitoes appear to be split: 58 percent of the county voted to release the mosquitoes somewhere on the islands, while 65 percent of the residents on the island where the release is proposed to take place voted against the experiment, according to a local paper, the FL Keys News. The final decision will be made by the Mosquito Control Board; three of the five members have said they would go with the will of the public. With the vote on the two measures split, it’s not clear whether the release will happen.

I’m going to keep it coming with this rural/urban split that we’re seeing emerge from preliminary exit numbers. National exit poll numbers are showing a demonstrable attitudinal difference between rural and urban/suburban voters. Forty-one percent of rural voters said in exit polls that life for the next generation will be worse than it is today; only 27 percent of urban voters and 35 percent of suburban voters felt the same way. A whopping 72 percent of rural voters think the economy is doing poorly, compared with 57 percent of urban voters. And as far as which candidate is best suited to solve the economy’s problems? Sixty-three percent of rural voters say Trump is the man for the job, and 49 percent of suburban voters agree; city-dwellers, meanwhile, think Clinton is best suited for the task — she garners 57 percent of their support.

That’s quite a stark city mouse/country mouse divide — those sentiments won’t be going away anytime soon. I traveled to Maine’s rural 2nd Congressional District this fall and heard a lot of frustration with the status quo from voters particularly with the idea that national Democrats fundamentally misunderstood rural cultural issues, like Second Amendment rights.

The forecast you see on the right-hand rail is based only on called states. But we did build in a Plan B in the event of an election like tonight: We can override our initial forecast by declaring a state “too close to call” and changing the model’s forecast. We’re doing that in Michigan right now, setting the odds at 50/50 there, and we’ll be monitoring other states for whether they need similar treatment.

The presidential race in Michigan is too close to call. Our model now gives Clinton a 55 percent chance of winning; Trump now has a 44 percent chance.

A snapshot of Clinton’s struggles in Ohio in two counties: Obama won Athens County, home of Ohio University, by 35 points in 2012. She won there tonight by just 17 points. And Obama won by 12 points in struggling Ashtabula County — subject of a feature by my colleague Clare Malone — but Trump’s ahead there by 19 points.

The geographic divide in Pennsylvania is stunning. Outside of the eastern part of the state, Clinton is winning only two counties: Allegheny (Pittsburgh) and Erie. That said, there are a ton of votes in the southeastern part of the state. Still, the current Clinton lead of 12 percentage points will come down considerably.

Markets are taking a beating as Trump is hanging tight with Clinton in swing states. But not only that, market uncertainty is rising, injecting greater swings in the pricing of stocks. The VIX, also known as the “fear index,” is a measure of just this — volatility in the stock market. As the components of the VIX index are no longer trading, we can only look to futures to see what the VIX index is doing. And those futures are rising sharply. Market volatility is expected to increase dramatically — indicating investors did not expect a Trump victory, or don’t view it favorably.

Talking about shocking counties, look to New Hampshire. Trump is up by 3 percentage points in Grafton County. That’s amazing considering Obama won it by a little less than 25 percentage points in 2012.

ABC News has called a win for Colorado Prop. 106, the “End of Life Options Act.” This makes Colorado the fifth state with legislation to allow terminally ill people access to medication to humanely end their lives.

As the presidential race unfolds dramatically, the death penalty is quietly having a successful night. The AP has called the vote on the death penalty ballot initiative in Oklahoma; the voters there have adopted a state constitutional amendment strengthening the punishment there, guaranteeing the state the power to execute and the ability to choose the means of execution. And Nebraska results are coming in. With 12 percent reporting, “repeal” leads in the vote on Referendum 426, 52-48. If that result holds, capital punishment will be reintroduced in the state. The state’s legislature had eliminated it last year.

Clinton and Trump together have 96.0 percent of the vote right now — third-party candidates have 4.0 percent. That’s about in line with expectations, but the third-party vote may nevertheless be enough to potentially have swung several swing states.

Nine states are deciding on marijuana legalization ballot initiatives today. We’re spotlighting each during the course of the day. Here are the states whose polls close at 10 p.m. Eastern:

Montana

If the ballot measure in Montana passes, it will repeal the three-patient limit for medical marijuana providers, giving qualifying patients easier access to the drug. Voters have had a wild ride with marijuana legalization in Montana. Medical marijuana was legalized in 2004, and the rules were amended in 2011 to stop advertisements for it and limit the scope of the business for providers and prescribers. Advocates tried unsuccessfully to repeal it in 2012. Then, after the 2011 bill was tied up in courts for five years, it went into effect in August.

We’ve seen only one poll for this measure, and it shows the measure losing by a 7 percentage point margin. With only one poll to look at, though, nothing is assured.

Nevada

If the ballot measure in Nevada passes, adults age 21 and older will be able to possess up to 1 ounce of marijuana and consume it privately. Adults who do not live within 25 miles of a marijuana retail store would be allowed to grow up to six plants. Medical marijuana is currently legal in the state. A 15 percent tax on the drug would be spent first on enforcing the measure; remaining funds would go to K-12 education. As in many other states voting this into law, current medical marijuana facilities in Nevada would be encouraged to transition into recreational marijuana facilities.

All seven state polls we’ve seen this year suggest that the measure will pass.

As a reminder, the odds you see on the right-hand side of this page are based only on pre-election projections and called states. Clinton isn’t really a 73 percent favorite right now — Trump holds narrow leads in many swing states, some of which are likely to be called for him eventually, so her actual odds are probably lower.

With 17 percent of precincts reporting in Michigan, Trump is ahead by more than 4 percentage points statewide. One county’s early returns look particularly surprising: Trump leads Wayne County, which includes Detroit, by about 3 percentage points with 11 percent of precincts reporting. That result would be wildly different from those in the past two presidential elections, when Barack Obama won the county with more than 70 percent of the vote.

With the rising chance of a Trump presidency, there is also a very outside chance of a three-week Democratic majority. How would that come about? On the off-chance that the Democrats can take 50 Senate seats while Trump wins the presidency, the new Congress will be sworn in on Jan. 3. The new president doesn’t take office until Jan. 20, giving the sitting vice president — Democrat Joe Biden — three weeks to cast the tie-breaking Senate vote.

Prices in betting markets implied that chances of a Brexit were just 15 percent heading into that vote, but then the U.K. voted to leave the European Union.

Heading into today, betting markets gave Trump about a 20 percent chance of winning the presidency. He’s up to 55 percent now.

A lot of the important decisions about elections are made by states, which means there’s a lot of variation. One example of that for those looking at early returns is when polls close. For states in the Eastern time zone, some closed as early as 6 p.m. and a few as late as 9 p.m. Is there a partisan pattern?

It turns out there is — as the chart below shows, when we look at the states that close later, the percentage of votes won by Mitt Romney in 2012 shifts way down. We should be careful about making causal claims — this pattern could be caused by any number of factors. But it’s an interesting pattern nonetheless.

Markets are reacting very negatively as Trump is exceeding expectations in the electoral map tonight. One thing to watch is the Mexican peso. As Trump’s chances of winning improve — and thus the potential for negative trade and immigration policies relating to Mexico rises — markets would adjust by devaluing the peso and showing declines in Mexican stock markets. Sure enough, the Mexican peso has depreciated sharply since 8 p.m. eastern time. (Or, in other words, the Mexican peso-to-U.S. dollar exchange rate has risen.)

Clinton now leads in Virginia. There is still vote out in Democratic areas. It’s close, but Clinton should win there.

Stating the obvious, but it’s very hard for Clinton to win the Electoral College if she loses Michigan along with Ohio, North Carolina and Florida none of which look particularly safe for her right now. Even if she were to hold the rest of her firewall and win Nevada, she’d be stuck at 263 electoral votes and would need to do something unexpected like flip Arizona or Georgia into her column.

Increased turnout of rural voters for Trump could be an interesting subplot to emerge from this election. In addition to places like Michigan, which saw a surge in rural voters, preliminary exit poll numbers in Virginia, which has yet to be called, indicate that Trump is outperforming Mitt Romney in the more rural central and western parts of the state 29 percent to 23 percent. That could be helping Trump keep the race close with Clinton, whose strength lies in the more urban environs of Northern Virginia.

Fox News has declared Clinton the winner in New Mexico. That’s a sigh of relief for her campaign.

In Virginia, Clinton is now down by just 5,000 votes with 14 percent of precincts still remaining. Clinton still has a good chance of winning the state.

Clinton is leading by only a percentage point in Wayne County, Michigan. That’s a county Obama won by 48 percentage points. Either that result is wrong, a lot of the vote in Detroit (which is in Wayne County) is out, or the map is looking very different than it used to.

As a point of context — one reason people find Trump’s competitive margins across a wide range of swing states so surprising is because exit polls showed Clinton beating her pre-election polls in most states, instead of underperforming them. Remember not to pay too much attention to them next time around.

Prediction markets now have the Electoral College as being almost even money.

Back in late September, our editor-in-chief, Nate Silver, wrote that Clinton was leading in exactly the states she needed to win — and that it wasn’t such a good position to be in. Now we can see why, as her firewall might be crumbling: According to the New York Times’s live forecast, Clinton has below a 60 percent chance of winning in Pennsylvania and Michigan, the two most electoral-vote-rich states in that firewall. She could still win them both; if she doesn’t, she has a much tougher path to winning the election.

One of the biggest splits in voting behavior this cycle has been between college-educated and non-college-educated white Americans. This map of American Community Survey data on educational attainment uses light orange to show states with lower rates of college degree holders. All of the states with lower rates of college education that have been called so far have gone to Trump: Wyoming, Alabama, Tennessee, Mississippi, Oklahoma, West Virginia, Indiana and Kentucky. Nevada, when called, may well buck that trend.

Michigan — a state where Trump leads by 3 points based on votes counted so far — is a state that ought to be making Democrats nervous. With lots of white, working-class voters and lots of undecideds, it was underrated as a swing state. Clinton will gain a ton more votes in Detroit, but the Upshot’s projections have the state as very nearly being a tossup now, and it’s probably a bigger risk to Clinton than Pennsylvania.

Trump’s lead is down to a percentage point in Virginia, with 82 percent of the vote reporting. There are still enough votes outstanding for Clinton to make up this margin. But this doesn’t look like the big lead that Clinton would have wanted. It looks like the race will be close.

Some Senate benchmarks:

• In Florida, Republican Marco Rubio is outperforming Trump by 6.4 points. The race was called for him earlier tonight.
• In Indiana, Republican Todd Young is underperforming Trump by 10.9 points, but he nevertheless won the state.
• In Missouri, Democrat Jason Kander is outperforming Clinton by 10.1 points.
• In New Hampshire, Democrat Maggie Hassan is outperforming Clinton by 2.3 points.
• In North Carolina, Democrat Deborah Ross is underperforming Clinton by 2.3 points.
• In Pennsylvania, Democrat Kate McGinty is outperforming Clinton by 1.1 points.

Early returns in Michigan are not great for Clinton. Trump is winning by big margins in rural areas and by 2 percentage points overall. The New York Times’s live forecast, which takes into account where the votes are coming from, is now giving Clinton just a 54 percent chance of winning the state and a 58 percent chance of winning the election overall. Before votes were counted, Clinton had a 94 percent chance of winning in Michigan according to the Times and a 79 percent chance according to our forecast.

Our live forecast currently gives Clinton an 83 percent chance of winning the state of Michigan, but there are some interesting things to pull out of the exit polls in that state that speak to the strength of Trump’s candidacy. In 2008 and 2012, rural voters accounted for only about 19 percent of the vote in the state, but according to preliminary exit poll results, they account for 27 percent of the state’s vote. And Trump does well with those rural voters, winning them by about 15 points.

We’ve been talking all night about the historic gender gap that we are likely going to see come out of this election, with women favoring Clinton over Trump in huge numbers — it’s news to no one that Trump has been a controversial figure in no small part because of his comments about women.

And yet, we’re seeing some interesting figures coming out of the Florida exit polls — Clinton isn’t doing quite as well with women there as she is in other key states, which might be one reason the race in Florida is so close. According to preliminary exit poll results in Florida, Clinton is winning 51 percent of women, compared with 53 percent in Michigan and New Hampshire; 54 percent in Ohio, Wisconsin, Georgia and Colorado; 55 percent in North Carolina; and 58 percent in Pennsylvania

The latest results seem a bit more favorable for Trump — our live model puts Clinton’s chances at 73 percent, down from 78 percent earlier tonight, and other models have likewise moved back toward Trump. Investors don’t seem to be happy about that — Bloomberg’s Joe Weisenthal notes that markets have fallen sharply over the past few minutes. That’s consistent with recent research finding that markets would prefer a Clinton victory.

Betting markets have shifted — quickly — in Trump’s favor. His chances of winning have risen to about 25 percent, according to Betfair.

The Upshot’s calculator has Clinton projected to eventually win the popular vote by 3 to 4 percentage points — right in line with where national polls had the race. And yet, they show her winning only 290 or so electoral votes, which obviously means that Trump has a decent shot to win the Electoral College. That potentially seems to validate our finding that Clinton was in a worse position in the Electoral College than the popular vote, since her coalition is not concentrated in swing states.

Election watchers right now are feeling a sense of déjà vu — in the 2014 Virginia Senate race and again right now, Virginia was expected to go easily Democratic, but the early returns are instead suggesting a tight race. In fact, Trump leads the state by just over 3 percentage points. From our look at the data, Clinton is running even with Obama except in the rural, southwestern part of the state, where Trump is outperforming Romney. Still, while 69 percent of precincts are in, just 54 percent of the 2012 major-party vote is in. We suspect that Clinton has a lot of outstanding votes in Virginia Beach and the Norther Virginia suburbs, but it isn’t crazy to think that Virginia could be the crack in the Clinton firewall.

Watch Broward, Palm Beach, Miami-Dade and Hillsborough Counties in Florida for signs of a Clinton comeback. Especially Broward, which has been slow to count today’s votes; nearly 30 percent of people in the county are African-American. Obama won more than two votes for every one won by Romney there in 2012, and Clinton is leading by an even bigger margin in the southeastern county among votes counted so far this year.

Despite a broad decline in its use nationwide, the death penalty may be bolstered in a few states after tonight’s vote. According to pre-election polling, California, Nebraska and Oklahoma looked poised to retain, reinstate and reinforce capital punishment, respectively. And now some real results are starting to roll in. In Oklahoma — the state with the highest per-capita execution rate since 1976 — the “yes” votes on State Question 776 currently have a 66-34 lead, with 11 percent reporting. A “yes” vote would amend the state’s constitution, affirming the state’s right to execute and choose the method of execution.

We’re in something of a state of suspended animation right now. The results so far are pretty well in line with pre-election polls, which showed a close race in many swing states and Clinton more often having the lead. But the problem with having a lot of narrow leads is that you don’t always convert them to wins, and so far no major states have fallen to Clinton.

While the state of Colorado has already legalized recreational marijuana and statewide ballots there don’t feature any questions about marijuana legalization, the ballot in the county of Pueblo sure does. Voters are weighing two marijuana issues there today. If issue 300 passes, voters will ban recreational marijuana sales in the city of Pueblo. If issue 200 passes, all marijuana facilities across the county will be shut down by Oct. 31, 2017.

Here’s a potential sign of a shift: Barack Obama won the township of Bradley, Maine — in the largely rural, 2nd Congressional District — by 12 points in 2012. Trump just won it by 10 points tonight. Maine splits its electoral votes by congressional district, so more results like that would put Trump on his way to winning one electoral vote there.

It’s possible that Clinton will win Ohio while losing Florida, which might seem like a huge surprise given the narrative of the race, but the fact is that the two states weren’t polling all that far apart. Clinton trailed Trump in Ohio by only 1.9 points in our pre-election forecast and led him in Florida by only 0.6 points.

We have arrived at the third and final video in our “New Bellwethers” series. Throughout Election Day, FiveThirtyEight’s senior political writer Clare Malone has been exploring counties in key states where Clinton and Trump are polling very close to how they are nationally. Our last stop: Maricopa County, Arizona.

If you’re just joining us, the second episode of “The New Bellwethers” video series — Berks County, Pennsylvania — can be viewed here.

Trump and Clinton are neck and neck in Florida. The issue of immigration is a crucial — and divisive — issue in the state. No surprise, exit polls indicate that the candidates’ voters have radically different views on the impact that immigrants have on the country today: 50 percent of Trump voters believe immigrants hurt the country, while a whopping 83 percent of Clinton voters say the opposite, that immigrants help.

Nine states are deciding on marijuana legalization ballot initiatives today. We’re spotlighting each during the course of the day. Arizona’s polls close at 9 p.m. Eastern.

If the ballot measure in Arizona passes, adults age 21 and older will be able to grow up to six marijuana plants in their residence, possess up to 1 ounce of marijuana and consume it privately. Medical marijuana is already legal in Arizona, and the existing pot industry could adapt to fill the recreational market. The retail industry would be capped at 10 percent of the number of liquor store licenses in the state. A 15 percent sales tax would cover implementation and regulation costs; any extra would benefit schools and the state health service, and cities would be allowed to pass their own restrictions.

According to the seven state polls we’ve seen in 2016, it looks like a toss-up. Three polls say the measure has a slight chance of winning, and four say the reverse.

Oh-me-oh, oh-my-oh, what’s up with Ohio?

Rob Portman is projected to win the Senate race, which we expected, and right now our live forecast is showing Trump with a 68 percent chance of winning the state. A quick dive into the state’s exit polls tells us that a plurality of voters there, 42 percent, said that being able to bring needed change was the most important attribute they looked for in a candidate. This would seem to favor Trump, not Clinton. Clinton also appears to be struggling with union households in the state; she and Trump are splitting that vote, whereas President Obama won that group by 23 points in 2012.

The question remaining in Florida is how much vote is left in the southeastern part of the state. Trump is ahead right now by a little over 100,000 votes. That is a lot of votes to make up, though it is possible for her to do it.

Betting markets — and The Upshot’s statistical model — now show Trump as slightly more likely than not to win Florida, expecting that he may pick up just enough votes in the Florida Panhandle and southwestern Florida to make up for what will be further Clinton gains in Broward County and Miami-Dade County.

What we’re waiting for in Virginia is results from the northern part of the state to come in. Right now, Trump is leading in the state by about 5 percentage points. There probably are enough votes left in the north for her to make up that deficit, but this race is definitely far closer than Clinton wanted it to be.

Political parties, partisan groups and nonpartisan organizations have sent election observers around the nation. One of them is Richard Burns, a New York nonprofit management consultant who has a law degree. He volunteered as a legal observer today — his fourth election doing so — in Pennsylvania as part of the Clinton Victory Counsel. Compared with the reports of problems in Pennsylvania, including reports from some voters that the machines changed Trump votes to Clinton ones, Burns found the Friends Meeting House in Lancaster calm.

“There were a number of recently naturalized citizens, middle-aged adults, beaming and saying this was their first time voting,” Burns said in an interview earlier today from the polling location. Lancaster is in Amish Country, a region that has seen a massive growth in the Latino population, which outstripped the Amish as of the 2010 Census.

One thing seems fairly clear, as we wait for the first swing state dominoes to fall: If Trump wins the Electoral College tonight, and it remains something of a long shot, it’s going to be with narrow wins in a large number of swing states instead of something more emphatic. And it’s going to be a very long night, possibly including an Electoral College-popular vote split.

As I told Clare in our video chat a few minutes ago, all the exit poll data so far has made me really despondent about the state of our politics. The gulfs are widening along gender, education, class, and urban/rural divides. This is, of course, nothing new — we’ve been talking about it for months — but I suppose Election Day is turning into a big, clarifying, heartbreaking reminder.

In Vermont, Republican Phil Scott is currently beating Sue Minter by 14 percentage points with 15 percent of the vote in. The race hasn’t been called, and with few votes in from heavily Democratic Burlington yet, that makes sense. But should Scott hold on to win, that victory wouldn’t be as surprising. Yes, gubernatorial elections are increasingly tracking trends in presidential elections. But paradoxically, on-cycle elections for governor have actually been less nationalized than gubernatorial elections held in midterm years. Maybe that’s because on-cycle elections attract more voters, including more moderate voters. That could be good news for some gubernatorial candidates today, including Montana Democrat Steve Bullock and Missouri Democrat Chris Koster.

NBC News has projected that Republicans will retain a majority in the House of Representatives.

Clinton trailed Trump by 10 points in our Missouri presidential forecast — so we’ll be looking toward whether Democratic senate candidate Jason Kander can run 10 points better than Clinton tonight. So far, he’s toeing the line, outrunning Clinton by 9 points based on returns so far from Jackson County, which includes Kansas City, Missouri, and its suburbs.

In ABC News’s preliminary exit polls, Clinton has a clear advantage over Trump in favorability ratings. In none of the states that were polled, however, did a majority of voters view either candidate favorably.

Trump’s lead is climbing in Florida as more of the Panhandle is reporting. He’s now up by about 65,000 votes or 0.75 percentage points. This could end up coming down to the wire.

As we all stare intently at results from Florida — it’s essentially tied right now — here’s an interesting tidbit from the exit polls there: Voters who consider the economy the most important issue facing the country favor Clinton over Trump by a 50-43 margin. Voters who are most concerned about immigration, meanwhile, went for Trump by a whopping 38 points, 68-30.

Polling by the firm Latino Decisions on election eve found that Florida’s Latino population had cast an early ballot for or planned to vote for Clinton over Trump 67-31. In 2012, Obama bested Romney among Florida Latinos by 58-40. The state has a large Cuban-American population that has trended Republican but that was dissatisfied with voting for Trump. Still, nationally, the firm shows Clinton winning the Latino vote 79-18, a significantly higher margin than in Florida.

The Florida results also showed Senator Marco Rubio reaping a lower share of the Latino vote than his challenger, Patrick Murphy, in a 40-58 split, although Rubio led in statewide polls and has now won the race according to preliminary exit polls.

Nine states are deciding on marijuana legalization ballot initiatives today. We’re spotlighting each during the course of the day. Arkansas’s polls close at 8:30 p.m. Eastern.

If the ballot measure in Arkansas passes, medical marijuana use for patients with qualifying conditions will become legal. A medical marijuana initiative was defeated in Arkansas in 2012. The marijuana would be taxed, with half the revenue going to vocational training and the other half divided among the general fund and other state programs.

We’ve seen three state polls this year: One showed voters slightly in favor, one showed voters slightly against, and the third showed voters overwhelmingly in favor. So there’s a slight lean toward legalization, but nothing’s certain yet.

This is why it’s important to wait for the Panhandle to report in Florida. Trump has just jumped into the lead in Florida by 12,000 votes. Still a lot of votes to be counted in that state.

Clinton leads in the Electoral College 68-37 based on states called so far, and our live election night forecast is becoming slightly more confident in its predictions as the candidates bring in states from their respective bases. Clinton’s chances of winning are up to 78 percent in the forecast.

The marijuana ballot measure in Florida passed by a landslide, legalizing medical pot for specific debilitating diseases. With 76.9 percent in favor and 29.1 percent opposed, the results lined up with the polls we looked at.

As Nate mentioned, Mark Kirk just became the first incumbent senator to lose his bid for re-election. Kirk has long been a top Democratic target: In 2010, he won the seat formerly held by Barack Obama in deep-blue Illinois. Outside spending to save him was relatively limited. During his time in the Senate, Kirk has cut a relatively moderate profile: He received an “F” rating from the NRA, is pro-choice, and among the Senate’s most centrist Republicans.

In that, Kirk fits a pattern. In recent years, the incumbent senators who have lost have been disproportionately moderates whose partisanship puts them at odds with their state. In 2012, the Democrats took out Republican moderates like Massachusetts’ Sen. Scott Brown — and Brown was the third most liberal Republican in the prior Congress. In 2014, Republicans returned the favor, beating moderate Democratic incumbents in Alaska, Arkansas, Louisiana, and North Carolina. Alaska’s Mark Begich was the 39th most liberal Democratic Senator, and the other three incumbents were still more moderate. This replacement of moderates is one key driver of the polarization we see in the halls of Congress today: moderates whose partisanship doesn’t align with their state find it hard to keep their seats.

(University of Pennsylvania students Jackson Gu, Max Kaufman, Thomas Munson, Owen O’Hare and Liz Sanchez helped with this post.)

We’ll also be updating our snake chart as the night progresses. Here’s each candidate’s remaining path to 270 electoral votes.

Perhaps the most shocking result so far is in Missouri. Democrat Jason Kander looks to be running very strongly in the exit polls. That would be a huge pickup for Democrats.

ABC News has called the Illinois Senate race for Tammy Duckworth, making this the first pickup for either party so far tonight. Republican Mark Kirk previously held the seat. The call isn’t a surprise — Duckworth was up 12 points in our forecast — but it’s something of an indignity for an incumbent senator to have a race called against him immediately after polls close. Kirk’s problems went from bad to worse following a debate where he mocked Duckworth’s heritage.

In preliminary exit polling from ABC News, more voters said they were bothered a lot by Trump’s treatment of women than by Clinton’s emails. However, state by state, the numbers ranged widely, with more voters in Arizona and Texas bothered by the emails.

NBC News says that New Hampshire and Pennsylvania are too early to call. That is excellent news for Clinton. She is leading in both. Trump likely needs to win at least one of those two states if he wants to win.

According to exit polls from ABC, 36 percent of Pennsylvania voters say society gives more advantages to whites than to minorities. Thirty-one percent said society favors minorities more than whites, and the other third said neither group is favored. With Clinton holding her own among nonwhite voters and Trump prevailing with whites, this could add to the societal divide in the state.

Right now, Clinton is up in Florida by more than 2 percentage points. The key counties look good for her. Keep in mind, though, that there are still plenty of votes to be counted in the Republican-leaning Panhandle.

New Jersey, which was just called for Clinton by ABC News, is the closest thing to a swing state that’s been called so far: Clinton had “only” a 96.9 percent chance of winning it, according to our pre-election forecast. Trump was a 97.5 percent favorite in Indiana, which was called earlier in the night. South Carolina — which other networks have called for Trump but ABC has not yet — is a little more competitive, however, as our forecast had put Trump’s odds at 89.7 percent there.

If you’re puzzling at the results in Virginia — more than a fifth of precincts are reporting and Trump leads by more than 10 points — don’t expect an upset there just yet. There’s still a ton of the vote left to report in Northern Virginia, specifically the heavily Democratic Washington D.C. suburbs.

NBC News has called South Carolina for Trump. (Our partners at ABC News have not, yet.)

Indiana and West Virginia have both been called for Trump. This isn’t surprising, but it does warrant a moment of thought. Until 2004, West Virginia was one of the most reliably Democratic states in presidential elections. Indiana has tended to lean Republican in presidential contests, going back to about 1968 (with the notable exception of Obama’s victory there in 2008). But it’s also home to moderate Democrats like Evan Bayh, who is competing for a Senate seat and was once considered a rising star in the Democratic Party. West Virginia is still represented by conservative, pro-gun Democrat Joe Manchin in the Senate — who has run on his opposition to the president whose party label he shares. The easy calls in Indiana and West Virginia for Trump probably don’t signal anything big about tonight’s presidential contest — but they do signal the kinds of changes that have happened in the two parties over the past few decades.

Not a surprise, but Clinton is really overperforming a typical Democrat in wealthy, highly educated areas. She’s winning Loudoun County, Virginia, by 16 points, for example, when Obama won it by just 4 points four years ago. If anything, these demographic shifts look even more profound in the vote so far than they did in the pre-election polls.

I’m a little worried about relying too much on bellwether counties for projecting results in Florida because the vote patterns in the state may have changed a lot since 2012. Still, Clinton is up by 10 percentage points in early returns from Hillsborough County. That’s good for her, but again, we need to wait for more data.

Georgia is still too close to call. Our final forecast gave Trump a roughly 80 percent chance to win the deep-red state. But Georgia has been undergoing a major demographic shift, with a growing share of African-Americans and Hispanics. This change is apparent in the exit polls from today: 39 percent of Georgia’s electorate is nonwhite.

Nine states are deciding on marijuana legalization ballot initiatives today. We’re spotlighting each during the course of the day. Here’s the batch of states whose polls close or start to close at 8 p.m. Eastern:

Maine

If the ballot measure in Maine passes, adults age 21 and older will be allowed to possess up to 2.5 ounces of marijuana and consume it privately. Medical marijuana is already legal in the state.

A 10 percent tax would be placed on marijuana sales.

According to the three state polls we’ve seen, it looks like the vote will be close. All three have the measure favored to pass, but not by much.

Massachusetts

If the ballot measure in Massachusetts passes, adults age 21 and older will be able to possess up to 10 ounces of marijuana, grow up to six plants for personal use and consume marijuana privately. Its use would be regulated similar to how the state handles alcoholic beverages. Medical marijuana is already legal in the state.

If the measure passes, the state will create the Cannabis Control Commission to oversee marijuana legalization. A 3.75 percent tax would be placed on marijuana sales. Revenue would be placed in a Marijuana Regulation Fund to pay for administrative costs. Cities and towns would be allowed to add a local tax of up to 2 percent.

Of the nine state polls we’ve seen in 2016, nine have the measure passing and two have it failing.

North Dakota

A yes vote on the ballot measure in North Dakota is a vote to legalize medical marijuana to treat specific debilitating medical conditions. A similar measure failed to reach the North Dakota ballot in 2012 after thousands of fraudulent signatures were found.

What’s the outlook? It’s not clear at all. No official poll has been done on marijuana legalization in the state in two years, and I haven’t found any polling on it.

One more note on education: Preliminary exit polls in Indiana indicate Clinton lost among voters without a college degree by 23 points. Among college graduates, however, she lost by just 4 points, and she won handily among voters with graduate degrees. Exact vote shares are different in other states that have reported, but the same basic pattern holds.

Democrat Evan Bayh, whose position tumbled in the polls of the U.S. Senate campaign in Indiana, is trailing Republican Todd Young by 15 points based on returns so far. The Democratic-leaning parts of Indiana haven’t reported much vote yet, so his position will improve. Still, Republican chances of winning the Senate will shoot up to 68 percent if Indiana is called for Young in our election night model.

According to ABC News exit polls, the state of West Virginia will lend its five electoral votes to Trump. This was expected, but it’s also an example of a state that was once blue becoming red.

The state was a touchstone of speeches at both the Democratic and Republican national conventions. At the latter, West Virginia Sen. Shelley Moore Capito gave a speech blasting the president and the Democratic candidate. “President Obama has hurt the heart and soul of my state, the proud coal miners and the communities where they live … creating a cycle of pessimism and disgust,” she said. “West Virginia voters are the backbone of this economy. And Hillary Clinton is promising to put them out of work?” The state has lost jobs as the Environmental Protection Agency has regulated mining and fossil fuel prices have stagnated.

We’re seeing some huge gender splits in the exit polls. In Ohio, for example, Trump leads by 15 points among men but is losing by 14 points among women. That 29-point spread is much bigger than the 18-point gender gap in Ohio in 2012.

Preliminary exit poll results suggest that we may see a record gender gap among voters in Virginia — a 19-point advantage for Clinton among women, 57-38 percent. It’s not too much of a leap to suggest that this might be connected to Trump’s treatment of women — 62 percent of women in the state said they were “bothered a lot by Trump’s treatment of women.” In North Carolina, Clinton appears to have won among women by 13 percentage points, which ABC News says appears to be another record.

What’s interesting is that although there was a record gender gap nationally in 2012 — about 20 percent — it didn’t show up so much in these states. Obama won female voters in Virginia by 9 points and won that group by only 2 points in North Carolina. We are seeing something new.

Trump’s campaign slogan, as pretty much everyone knows by now, is “Make America Great Again.” Maybe it isn’t surprising, then, that Trump voters are pretty pessimistic about the direction the country is headed. According to preliminary exit polls, two-thirds of Trump voters (67 percent) think the country is headed in the wrong direction, and nearly as many (60 percent) think the next generation of Americans will be worse off than this one. Clinton’s voters are much more optimistic: 89 percent think the U.S. is “generally going in the right direction,” and 60 think the next generation will be better off than this one.

As dark as this campaign has been, however, voters are generally more optimistic in 2016 than they were in the midterm election two years ago. Overall, 37 percent of voters think the next generation will be better off, up from 22 percent in 2014.

Even if he loses tonight, Trump is likely to have places where he significantly outperforms past Republican nominees. In Vigo County, Indiana, for example — home to Terre Haute — Trump is winning by 13 percentage points, whereas Barack Obama won the county by 1 point in 2012.

Trump’s win in Indiana is no surprise — FiveThirtyEight’s model gave him a 97.5 percent chance of winning there, despite the fact that Obama carried the state in 2008. Trump’s big margin in Indiana signals Clinton’s challenges with less-educated voters. Just 25 percent of Hoosiers ages 25 and up have a bachelor’s degree, one of the lowest shares of any state. Trump also won Kentucky, where even fewer residents have a college degree, while Clinton won Vermont, one of the most educated states in the country. Of course, those are far from the only factors separating those states.

I’m looking at returns from two major counties in Florida: Duval and Palm Beach. In the early vote, Clinton is doing better than the amount that Obama won by overall in those counties in 2012. In Palm Beach, she’s running 7 percentage points ahead. If that holds, it’s going to be a very long night for Trump.

In preliminary exit polls according to our partners at ABC News, voters were more scared by the thought of a Trump presidency than a Clinton one. Women, especially, expressed fear at the thought of President Trump. Overall, though, voters are expressing more fear than excitement over both a Trump and a Clinton win.

If you’re looking for ticket splitting, look no further than Florida. In the initial returns from the Jacksonville area, Marco Rubio in the Senate race is running 8 points ahead of where Trump is running in the presidential race. That’s good news for Republicans looking to hold that Senate seat.

We’re getting reports that voting hours will be extended for anywhere from 20 to 60 minutes in a number of Durham, North Carolina, polling places. There were problems with voting machines in this area earlier today — computers broke down and some voting had to be done on paper — and polling officials are likely trying to account for this delay.

We’ll know soon enough how much of the electorate was made up of women. Early-voting numbers, though, show that women’s share of the electorate has surged — by several percentage points compared to 2012, according to Drew Brighton of the voting-data firm TargetSmart. According to the firm’s breakdown of 46.3 million early votes, 56 percent were cast by women — consistent with earlier reports of big turnout among women. And that proportion was about the same across just about every age group, as well as among both women who are newly registered and those who’ve been on the rolls for longer. If the final vote tallies are consistent with the early ones, that could provide a boost to Clinton, who led by big margins among women in polls but trailed among men.

Rachel Maddow earlier said on MSNBC that we’re looking at “class warfare” in American politics and that there’s a possibility that Clinton will win among college-educated white voters (or at least come close). These comments highlight the complex nature of what we mean when we talk about “class.” It’s often used interchangeably with “income” — which is an important factor in determining partisan identification. But in this election, we’ve also heard a lot about education. Although higher incomes tend to correlate with identifying as a Republican, the education variable means something different in 2016. Trump appears to be poised to do much better with voters with less education than those with more years of school under their belt. In other words, education is correlated with income but has different political implications. Being clear about these variables isn’t just an academic exercise; it’s important for politics. When partisanship becomes a function of several correlated, mutually reinforcing social identities, party divisions become cemented — and ripe for resentment.

A local news station is reporting that two polling places are on lockdown near the Azusa shooting. The L.A. County Registrar tweeted out that, because of the lockdown, voters would be able to cast their ballots in other locations — voters should avoid the lockdown area.

In a bad early sign for Trump, exit polls in Georgia imply a close race there, with Trump leading only about 48-47 according to preliminary exit polls put out by CBS News.

We have our first calls of the evening. Trump has won Kentucky and Indiana, while Clinton has won Vermont, according to ABC News. On the Senate side, Republican Tim Scott has won re-election in South Carolina, as did Rand Paul in Kentucky. Democrat Patrick Leahy has won re-election in the state of Vermont. We’ll have new model odds shortly.

The story in Virginia is exactly what we thought it was. Clinton looks to be crushing Trump in the Washington, D.C., suburbs, while Trump is winning in the central and western parts of the state. The potential problem for Trump is that the regions he is winning make up a smaller percentage of the vote than the regions Clinton is winning.

With one of seven precincts reporting, Elliott County, Kentucky, is at 69 percent for Trump. If the result holds, this would break a historic streak: The county was formed in 1869, and the Democratic candidate has won it in every presidential election since then.

This wouldn’t be as big of a shock as it sounds. A narrow victory for Obama in 2012 foretold Elliott going red, like most of the state.

How important is Florida? (Polls in the eastern part of the state close in a few minutes.) If Clinton wins it, her probability of winning the Electoral College would shoot up to 93 percent from 71 percent, according to our election night model. And if Trump were to win it, his chances would increase to 59 percent from 29 percent.

This is a developing story, but the Los Angeles Times is reporting a shooting near a polling place in Azusa, California, a city in the San Gabriel Valley. It’s being reported that one person is dead, multiple people have been shot and the shooter is heavily armed and active. We will, of course, be keeping our eye on this.

Trump has repeatedly claimed the election is “rigged,” possibly positioning himself to not accept the results of tonight’s election. In exit polls, Trump and Clinton supporters, unsurprisingly, disagree on whether the votes in their state will be counted accurately. Among Clinton supporters, 91 percent are “very” or “somewhat” confident, but only 77 percent of Trump supporters are.

Virginia, where polls close at 7 p.m., was a consistently Republican state, at least at the presidential level, until Obama won it in 2008. But now it’s practically a blue state — our model gave Clinton a better than 85 percent chance to win the state.

One factor giving Clinton an edge in Virginia: 37 percent of residents ages 25 and up have a bachelor’s degree, making it one of the most educated states in the country. Clinton, of course, is expected to outperform Trump by a wide margin among college-educated voters.

Virginia is also in strong shape economically, with an unemployment rate of 4 percent (vs. 4.9 percent for the U.S.) and a median household income of more than $66,000 (vs. $56,000 nationally). But Virginia is to some degree a tale of two states: The eastern counties near Washington, D.C., are among the wealthiest and best-educated in the country. But many counties in the western part of the state have median household incomes below $40,000.

Nine states are deciding on marijuana legalization ballot initiatives today. We’re spotlighting each during the course of the day. Florida’s polls start closing at 7 p.m. Eastern.

If the ballot measure in Florida passes, it will legalize medical marijuana for specific debilitating diseases. Florida is particularly interesting because low-THC marijuana is legal in the state when consumed by a method other than smoking. The ballot measure would make it more widely available to patients. In 2014, a similar ballot measure failed. Unlike the 2014 measure, this year’s initiative requires parental consent if the patient is a minor.

According to the 13 polls we’ve seen in 2016, the measure is likely to pass. Every poll has the measure going through, with the smallest margin at about 25 percentage points and the largest margin at 70 percentage points.

We’re going to see our first states being called soon enough, but in the preliminary exit poll results from Pennsylvania, I was struck by respondents’ answers to questions about whether they had been affected by either candidate’s ground game. You’ll recall that Pennsylvania is a state that Trump really needs to win, and the effectiveness of his ground game has been doubted by many (earlier this year, I reported on an internal Trump memo that outlined the campaign’s unorthodox strategy of going after unlikely voters). In Pennsylvania, 23 percent of voters say they were only contacted by the Clinton campaign, 13 percent only by Trump’s, 16 percent by both. That could make the difference tonight.

Early in the night, before we have a critical mass of returns, one thing I’ll be looking for is whether and where Trump is doing better than Romney did four years ago. So I’m looking over the Kentucky counties that have reported votes so far. And it doesn’t appear that Trump is doing significantly better than Romney. Trump is doing better in some counties but worse in plenty of others.

The Philadelphia district attorney’s office is investigating a report that a man wearing a badge labeled “poll police” attempted to interfere outside a polling place near 39th and Haverford in West Philly.

A police report was filed in the 16th District; a spokesperson from the D.A.’s office confirmed that officials “are aware” of the incident and “we are investigating.”

BillyPenn learned the confrontation happened after a question about a voter’s eligibility that was eventually resolved. After the ordeal, a group arrived in a pickup truck and the man with the badge approached the poll watcher, asking whether the poll watcher was making trouble. He took off his jacket and showed the badge. The poll watcher apparently thought the man may have been a Trump supporter.

The report follows an Election Day that had largely passed without major incident and with few complications. At an afternoon press conference, the D.A.’s office said it had received 10 calls about machine malfunctions, a handful of calls about illegal assistance and 13 complaints of electioneering.

This post was produced in conjunction with BillyPenn, a website covering the Philadelphia area.

It’s been a long election. Throughout the day, I’ll be highlighting some of the moments that made me say, “Whoa, can you believe that was this election?!” (Here are the first and second entries.)

Remember when there was a kid with stickers all over his face behind Hillary Clinton?

Remember when Donald Trump gave out Lindsey Graham’s phone number during a speech? (I just tried calling it, by the way. No dice.)

Remember when the political press spent a few days running down the details of whether Ben Carson had stabbed a kid but had his blade deflected by the kid’s belt buckle?

If you have any nominations for “remember when,” get in touch by Twitter.

As you start to see how different demographic groups voted this year, it might be helpful to see how they voted in the past. We’ve got you covered:

One thing we’ll keep an eye on in Indiana is the amount of ticket splitting. In preliminary results, Bayh, in the Senate race, is running 8 points ahead of where Clinton is running in the presidential race. As more Democratic areas come in, we’ll see if that’s enough for Bayh to win the Senate race even as Clinton is losing the presidential race in Indiana.

The first states will be called soon, and the state-by-state results in 2016 will be added to the long history of presidential elections. Here’s that history so far:

We’re still looking at preliminary exit poll data, but how does the 2016 electorate look so far compared to 2012? The two electorates look mostly the same, as you would expect. But there may be some slight differences. (These numbers are likely to change.)

1. College graduates are 50 percent of the electorate in 2016 versus 47 percent in 2012.
2. Liberals are 27 percent of the electorate compared to 25 percent in 2012.
3. Senior citizens (65+ year-olds) are 17 percent of the electorate in 2016 compared to 16 percent in 2012.
4. Evangelical white voters are 27 percent of the electorate in 2016 versus 26 percent in 2012.

Exit polls ask voters how they feel about candidates and the issues. I wish they asked more about how people feel about the process of voting. Exit pollsters are there mainly to serve the media, including our partners at ABC News, and I get why they focus on the most newsworthy tidbits. But they’re also in a unique position to talk to people right after they’ve experienced voting, and it’d be fascinating to know how their experience went: How long did they wait? Was the ballot confusing? Did anyone try to stop them from voting?

My own experience voting this morning in Queens, New York, was both inspiring and troubling. Inspiring, because people lined up in the cold to vote in a state where the race at the top of the ballot is a foregone conclusion, and because poll workers were abundant and kind. Troubling because the process was inefficient and confusing — you had to figure out from watching other people where to take your ballot to fill it out, and then where to take it to scan it; no signs or people pointed the way.

This is far from a scientific exit poll, but I asked people on Twitter how their experience affected their faith in our electoral process. Most respondents so far said nothing changed for them; of the rest, more gained faith than lost it. You can still vote in this exit poll for a few more hours.

It’s not all politics today at FiveThirtyEight. Or maybe it is. … I called Walter Engelund of NASA’s Langley Research Center this afternoon to talk about a piece I’m writing about the technical challenges of sending humans to Mars. He told me that the U.S. space program is a politically charged topic. “Almost without exception, whenever a new administration comes in, we get a change in priorities,” he said. Sending humans to Mars will require sustained support, and that means politicians who are willing to fully fund NASA and prioritize its Mars program.

Both Clinton and Trump have expressed enthusiasm for the space program, but they’re vague on details. Trump has given at least one speech promising not to cut space funding, but he also told a town hall rally in New Hampshire back in August that he wants to “rebuild our infrastructure” before sending people to Mars. Clinton has called herself an “enthusiastic supporter of human space flight” and indicated that going to Mars is a “consensus horizon goal,” whatever that means.

The Paris Climate Agreement went into effect this week, and although the presidential campaign paid scant attention to climate change, there’s a relatively straightforward approach to the issue — one backed by most economists — that’s being offered to voters in Washington state. It’s a carbon tax, and if Initiative 732 passes, it will make Washington the first state to impose a tax on carbon emissions.

The new carbon tax would start at $15 per metric ton of carbon dioxide in 2017, climb to $25 per ton in 2018 and then automatically rise 3.5 percent plus inflation every year after that until it reaches $100 per ton (as measured in 2016 dollars). Proponents estimate that the measure would boost the price of gas by about 25 cents per gallon and raise the price of electricity from coal-fired plants by about 2.5 cents per kilowatt-hour. The tax is intended to be “revenue neutral” by swapping it for reductions in other taxes, such as state sales tax and certain business taxes. Ironically, the the legislation would give an inadvertent tax break to Boeing for its sale of commercial aircraft, an important contributor to climate change.

Recent polls show a tight vote. While economists generally favor a carbon tax, the devil is always in the details, and the details are what are being disputed. The measure is backed by more than 50 University of Washington climate scientists, climatologist and activist James Hansen, former Secretary of Energy Steven Chu and Leonardo DiCaprio, who say the law puts a necessary price on carbon emissions. But some environmental groups, including the Sierra Club, oppose the initiative, saying that doesn’t do enough to help low-income people who might be hurt by the tax, it doesn’t shift enough money to clean energy and that the tax swap could lead to a state budget deficit. Interests funding the opposition campaign include the fossil fuel and manufacturing industries and local utility Puget Sound Energy.

When it comes to funding for science, Clinton’s priorities look a lot like Obama’s. But Trump’s are very different. (SPOILER ALERT for those of you who didn’t see that coming.) In particular, there are a couple of big changes that a Trump presidency would likely bring to the federal science budget.

First is NASA. Trump digs space exploration, and while it would be inaccurate to say Obama has been anti-astronaut or something, he has definitely de-prioritized NASA’s space-travel mission in favor of its earth-science mission (i.e., studying climate change). Over the course of the Obama presidency, the budget for NASA earth science research increased by 70 percent. His disinterest in returning to the moon, in particular, has been critiqued on both sides of the aisle. If a President Trump wanted to build bipartisan scientific goodwill, a moon mission would be a good way to do it.

The second thing is the Department of Energy, where funding has been heavily weighted toward research on and support for renewables. For instance, the DOE’s budget for energy-efficiency and renewable-energy programs grew by 40 percent between Obama’s 2016 and proposed 2017 budget — while funding for fossil fuel programs (think R&D and clean coal tech) fell by 27 percent, and it was lower than the clean-energy budget to begin with. It’s likely that a Trump administration would reverse these trends.

The urban-rural divide appears to be growing ever stronger this election season. But at least one rural part of the country continues to vote solidly Democratic: the so-called Black Belt. Made up of a swath of counties that sweep across the South and have large African-American populations, these counties were solidly blue in a sea of red in recent elections. You can see an example below in the 2012 results from Alabama.

John Zippert, secretary for the Federation of Southern Cooperatives, has worked with African-American farmers in the rural South for nearly 50 years. Even without Medicaid expansion in most of these states, which Zippert thinks is desperately needed, he said President Obama has been good for the region. He is cautiously optimistic about what will come if Clinton is elected. “When you elect Hillary Clinton, you don’t just elect Clinton, you elect the 3,000 people she appoints, including 50 from the Department of Agriculture that we are going to have to work with every day,” he said. He hopes that group will be diverse.

There are longer-than-expected waits to vote in several parts of the country. In Durham, North Carolina, a pending lawsuit asks that poll closing times be extended by 90 minutes, to 9 p.m. Delays or problems with the voting process have been reported in states including Colorado, Florida and Pennsylvania.

I spoke with Pita Juarez of the One Arizona coalition, which is working to turn out Latino voters. She said that while turnout was strong, voting was slow in the morning because of problems with Wi-Fi and malfunctioning ballot machines. With people getting off of work in that state, she worries the lines will get even longer.

The polls have closed in most of Indiana and Kentucky, although the polling stations in the Central time zone in those states are still open. Don’t expect a call from the networks in either state until 7 p.m., but we should start getting results from them sooner than that.

In 2014, the story of a young woman with fatal brain cancer began making the rounds online and on TV news shows. Twenty-nine-year-old Brittany Maynard knew she was dying, and she wanted to do so on her own terms, by taking prescribed medication if her pain became resistant to morphine and unbearable. This desire prompted her to move to Oregon, where, by the time she died on Nov. 1, 2014, she had become perhaps the most widely known user of Oregon’s Measure 16, called the Death with Dignity Act. Maynard’s activism brought new attention to right-to-die legislation, and her widower, Dan Diaz, has lent his support to Colorado Proposition 106, the End of Life Options Act, which would make Colorado the fifth state to permit some form of assisted suicide.

Modeled on Oregon Measure 16, Proposition 106 would allow terminally ill patients (those with less than six months to live, as determined by two doctors) to take a lethal dose of medication. To receive the drugs, the person would be required to voluntarily request it three times — twice orally and once in writing — with witnesses present. The measure would also criminalize the coercion of patients.

A September Rocky Mountain PBS and Colorado Mesa University showed that 70 percent of registered voters favored the measure. Advocates include Gov. John Hickenlooper and the Boulder and Denver Medical Societies. A February 2016 survey of members of the Colorado Medical Society, the largest physicians organization in Colorado, found that 56 percent of respondents supported “physician assisted suicide.” The Catholic Church and disability rights groups are among those rallying against the measure, and the Denver Post also has come out against it, over concerns that it could “entice insurers to drop expensive treatments for terminal patients.”

An update about what’s going on in Nevada right now, since it’s pretty convoluted, open to a certain amount of interpretation, and a lot of you seem to have questions about it. This afternoon, a judge denied the Trump campaign’s request that early-vote ballots be sequestered and that information on poll workers be given out, though the suit might well continue, according to elections experts I’ve talked to. From a macro view, the purpose of the Trump suit seems to be to lodge a complaint on the record about potential improprieties in heavily Latino Clark County in case the race is close in Nevada. To be clear, the facts on the ground do not, at this point in time, seem to suggest that there were any improprieties (see Jon Ralston’s reporting from the state if you’d like to know more.)

One question this suit would have to grapple with if it ever reaches a point of action is: When did voters got in line to vote during early-voting hours? On Election Day itself, Nevada law is clear that you cannot vote if you’ve entered the line past the official closing time of a polling place. But the rules are foggier when it comes to early-voting days. If they got in after the time when polling places officially closed, that could indeed be problematic … or it might not matter. It depends on what legal expert you’re talking to. Over at Slate, Rick Hasen argues that since early voting locations are often at sites used for other purposes — like the Nevada grocery store polling site that’s in question in the Trump suit — the rules are more flexible, citing Nevada code: “The schedules for conducting voting are not required to be uniform among the temporary branch polling places.” This is basically the argument that Clark County lawyers made this afternoon.

In an email, Professor Ned Foley, an elections expert at Ohio State University Law School, said he saw the validity of this argument, that voters should be allowed flexibility on early voting days, but offered the other side as well:

“I also see the counterargument, that once a scheduled closing hour is set for each day of early voting, the government must stick with that closing hour, meaning that a voter who shows up late needs to come back the next day, or on Election Day itself if that’s the next day of voting. Some points in support of this counterargument are: rules are rules and needed to be followed, especially in elections, so that all political parties and voters know the rules in advance; thus once set, shouldn’t be changed. Also, letting late voters cast a ballot isn’t fair to voters who didn’t try to go after the scheduled closing hour; they didn’t know that the rule would be bent for some.  Of course, for any voter who was already in line at the closing hour, there’s no dispute: they get to cast a ballot.”

The law, jealous mistress that she is, offers no clear answers.

Jonathan Bernstein posted an underreported statistic about the election the other day (citing Boris Shor): 73 percent of Democratic state legislators are endorsing Clinton, while 5 percent of Republican state legislators are endorsing Trump.

This strikes me as incredibly important. Yes, a number of prominent Republicans have refused to back Trump and have even urged a vote for Clinton, and a few others like John McCain have gone back and forth in their support. But only in the state legislative endorsements do you get a sense of the dramatic differences between the major-party candidacies this year.

Like most other information about state legislators, this hasn’t been well-covered. Yet it might have given Republican voters greater confidence in casting a vote for someone else or even skipping the presidential race altogether if they knew how many of their party’s elites were opposed to its presidential nominee.

Throughout this evening, we’ll be updating election night forecasts as states are called for presidential and senate candidates. To clear up any misinterpretations, we’re not trying to project states based on partial returns. So if (for example) Trump is leading Missouri by 5 percentage points with 40 percent of precincts reporting, that won’t matter to the model.

Instead, our election night model is much simpler than that. It relies upon only these three things:

1. Our pre-election forecasts.
2. States that are “called” by our partners at ABC News.
3. The amount of time that has passed since the polls closed in a state, if it hasn’t been called yet.

To repeat, these forecasts do not use votes counted so far. They also do not use exit polls. They do not look at margin of victory. The only input is a single designation for every state: “D” (called for the Democrat), “R” (called for the Republican”) or blank (not called yet), based on calls made by the ABC News Decision Desk. We can also call states for independent candidates or project that the Georgia or Louisiana Senate races will go to a runoff.

Having a state called for you helps in two ways in the model.

• It gives you electoral votes.
• It helps you in our forecast for the other states. For example, if Wisconsin has been called for Clinton, the model can infer that she’s more likely to win Minnesota. And it really helps candidates if they win in an upset, since that’s a sign that they may be beating their polls everywhere.

Our election models have been running tens of thousands of simulations each day in order to account for the relationship between different states in the Electoral College. The most important takeaway is that state outcomes are correlated: If Trump (unexpectedly) wins Virginia, for example, he’s also extremely likely to win North Carolina. So each simulation creates a plausible map based on a state’s region and demographics. In one simulation, perhaps, Clinton might outperform among Hispanics, therefore winning Florida and Arizona despite losing Ohio and Iowa.

These simulations are useful for our election night forecast also. Once we know the results in some states, we can make better inferences about the results in the remaining ones. So as states are called, we update the forecasts accordingly based on a series of regression analyses that relate every state to every other one.

The model also considers how long it’s been without a call in a state. If it originally expected Clinton to win New Jersey by 10 percentage points, for instance, but New Jersey still hasn’t been called five hours after polls have closed there, it will discount that lead significantly, assuming the state is closer than pre-election polls had it.

As a final word of caution, I wouldn’t read too much into the first few called states unless there are substantial upsets (Clinton winning Indiana, for example). But once swing states start to be called and the model has more data to work off, its projections will be more meaningful.

Exit polls asked voters about Obamacare, but instead of the usual dichotomous responses of “in favor” or “not in favor” of the law, voters were given three options: did the law go too far, not far enough, or was it about right? While 45 percent said the law went too far, 31 percent said it didn’t go far enough, suggesting that many of the people frustrated with the law don’t necessarily want it repealed.

Earlier today, we noted that betting markets were increasingly favoring a Clinton win. Her odds have risen further in the last few hours. She now has an 85 percent chance of winning according to an aggregation of two betting markets, Betfair and PredictIt. That is up 2 percentage points over the past four hours.

While exit polls are not the best at measuring the racial composition of the electorate, it does seem that minority turnout is way up in Florida. In the preliminary exit polls, 39 percent of voters were people of color. That compares to just 33 percent in 2012.

International trade has been an enormously contentious issue in this election. Trump has lambasted trade deals like NAFTA — and countries like China — as hurting U.S. workers, particularly in manufacturing. Overall, voters were evenly split on the issue in exit polls: 41 percent responded that trade with other countries “takes away U.S. jobs” while 39 percent said it “creates more U.S. jobs” (another 11 percent were undecided). But Trump and Clinton voters are a mirror image on the issue, with 63 percent of Trump voters saying trade took jobs away while 60 percent of Clinton voters thought it created them.

Among the wide-array of topics covered by California’s 17 propositions on the ballot there today is prescription drug prices. The bill, Prop 61, is based on a wonky law and formula that allows the Veteran’s Administration to receive discounts on drug prices, basically 24 percent off what the rest of the country pays, on average. Californians are voting on whether or not to require that some state health programs not pay more for drugs than the VA does.

Supporters of the bill argue that it passing in bellwether California is an important first step in starting a national conversation about drug pricing. The record-breaking $100 million raised to fight the legislation, money that mostly came from pharmaceutical companies, suggests that the industry also sees the law’s potential ripple effects. Pharma-backed opponents claim the bill will create red tape that will slow down or prevent people from getting necessary drugs. They also note that since the drug industry won’t legally be required to abide by the discounts, the state would have to negotiate on each drug, which could make it impossible for some people to get certain medications. It’s also unclear how the law would be enforced, since the VA doesn’t make public the price that it ultimately pays for drugs.

But there’s another camp of No voters that’s a little more complicated, people who are in favor of government controlled pricing of drugs but are against this particular law. From their view, pharmaceutical companies will be unlikely to take a profit hit, and will find a way to make up the revenue elsewhere. Since it’s unclear who would end up footing the bill, they are wary of the law of unintended consequences. As Health News Review points out, a lot of the media coverage about Prop 61 has fallen short of helping California voters understand the effects of the bill. That’s likely because the tangled web that is our health care system makes it difficult to know what the effects would be.

Preliminary exit polls show that many voters were motivated by dislike of one of the candidates more than by support for the person they voted for. One in five Clinton voters said they chiefly oppose the other candidate, and 27 percent of Trump supporters said the same; those figures were just 8 percent for Obama voters in 2012 and 10 percent for Romney voters. Asked about Trump’s treatment of women, 70 percent of all voters said they were bothered some or a lot; 62 percent said the same about Clinton’s emails while secretary of state.

There has been talk that black turnout may be down in North Carolina. The preliminary exit polls suggest that may be the case. Only 21 percent of voters in North Carolina identified as black. That’s down slightly from 23 percent in 2012.

We’ve spent a lot of time talking this year about the visceral reactions that each candidate elicits from the American public — they were both historically disliked — so what’s struck me in the initial exit polls are what Americans said about their level of fear about the future, no matter who is elected.

Preliminary exit polls suggest that 17 percent of voters are excited by the idea of Clinton as president and 29 percent are scared of it; only 13 percent of voters are excited about Trump as president, while 37 percent said that idea scared them.

Colorado’s statewide voter database apparently broke down for about 20 minutes this afternoon. This rendered voters unable to cast ballots during that time period, although people who’d previously filled out their mail-in ballots could drop them off. Voters could also drop off provisional ballots during that time. According to the secretary of state’s office, the system is working again now.

Here’s another interesting tidbit from the exit polls: Voters overall consider the economy the top issue facing the next president. But Clinton voters are significantly more focused on the economy than Trump voters, who are more concerned with terrorism and immigration. That runs counter to the narrative, which we’ve previously questioned, that Trump supporters are motivated primarily by economic anxiety. (On the other hand, opposition to immigration may be driven at least in part by economic concerns.)

The ideological divide between old and young voters looks like it’s as big this year as it has been in more than four decades.

Since 1972, the American National Election Studies polls have asked voters after elections to place themselves on a scale between 1 (extremely liberal) and 7 (extremely conservative). In a pilot study in January, registered voters 25 and under averaged a score more than a point below voters 65 and up, the biggest gap in any presidential year besides 2000.

Molly Ball profiled “Trump’s Graying Army” in the Atlantic and mentioned this trend, based on analysis of the ANES data by Shawn Patterson, a graduate student at the University of California, Los Angeles, who also helped with our analysis. The ideology of older voters has held fairly steady, but young voters vary more across generations.

There are a couple of caveats to keep in mind. This year’s pilot study happened during the primaries, and it’s possible that greater extremes during the primary contests polarized voters. In particular, Bernie Sanders’s popularity with young people might have made them more liberal. Also since we couldn’t know in January if respondents were going to vote in November, the measure here is among all registered voters. This year’s survey also called the extremes of the scale “very liberal” and “very conservative” instead of “extremely liberal” and “extremely conservative.”

The ideological gap might not translate into a historically large gap at the polls. Obama won 18-to-29-year-olds by 23 points in 2012, while Romney won the 65+ vote by 12 points. (Pollsters and exit polls typically don’t break out voters 25 and under into their own category.) That’s roughly in line with figures for this year’s race from online pollsters Reuters Ipsos and Morning Consult. That could reflect a drift back to the center since January, or the difficulty of placing Trump on an ideological scale.

If all voters who approve of President Obama vote for Clinton, it’s going to be a good night for her. The preliminary exit poll puts Obama’s approval rating at 54 percent. That’s far higher than George W. Bush’s was in 2008, when just 27 percent approved of the job he was doing. He definitely hurt John McCain’s chances of winning.

A lot of Republicans thought the polls were “skewed” because there were more Republicans than Democrats in them. Well, if the preliminary exit poll is correct, more voters today were Democrats than Republicans. Democrats made up 37 percent of all voters, Republicans were 32 percent and independents were 31 percent.

Preliminary exit polls suggest that voters don’t feel great about the economy, but they feel a lot better than they did in either of the past two presidential elections. Some 62 percent of voters said the economy was “not good” or “poor,” down from 76 percent in 2012 and 93 percent in 2008. (Only 3 percent of voters consider the economy “excellent.”) Today’s voters aren’t as bullish as in 2004, when only a bit more than half of voters rated the economy negatively.

Voters also feel a bit better about their personal financial situation. Thirty percent say their family’s finances have gotten better in the past year, while 27 percent say it’s gotten worse. In 2012, those numbers were 25 percent and 33 percent, respectively.

I’m seeing a lot of women posting photos of themselves wearing pantsuits as they go to the polls today to vote for Clinton. The Pantsuit Nation meme is a nod to the fact that as recently as 1993, wearing a pantsuit was an overt political act. That’s the year that Sens. Barbara Mikulski (D-Maryland) and Nancy Kassebaum (R-Kansas) committed the radical act of wearing pants to the Senate. “The Senate parliamentarian had looked at the rules to see if it was OK,” Mikulski told CNN. “You would have thought I was walking on the moon. It caused a big stir.”

Clinton has spent much of her career fending off criticisms of her clothing and appearance. Her embrace of the pantsuit has become a symbol of her cool confidence and desire to get down to work. No wonder then that her pantsuits have inspired an Instagram account, a secret Facebook group for her fans and even a flashmob.

As we look back on the 2016 presidential campaign, one of the things many of us will be thinking about for a long time is the content of this campaign. A few days ago on the podcast, Jody, Farai and Harry noted that they’d like to see presidential campaigns emphasize issues rather than personality. This fascinating piece at Slate argues that the campaign was, in fact, about important issues.

I’d like to make the case for focusing on the candidates — not in a way that’s specific to 2016, but just in terms of looking at what’s unique about picking a president. Party labels will tell us about where the candidates stand on the issues. Clues about how they will make decisions can be relatively rare, but they’re really important. From unexpected crises to the individuals they appoint to key policy positions, presidential decision-making is incredibly influential. But decision-making qualities can be hard for voters to determine, even for experienced candidates. This just might be an argument for why the input of party elites is useful at the nomination stage.

One interesting nugget from the early exit polls is the question of honesty. Trump called Clinton’s honesty into question throughout the campaign. And while 59 percent of voters say Clinton is not honest, 65 percent say Trump isn’t honest.

During the Democratic primary, Bernie Sanders pushed for a universal single-payer health care system he dubbed “Medicare for All.” Sanders’s home state of Vermont tried, and failed, to implement a single-payer system in 2014. Costs and political considerations doomed the Vermont plan, but nevertheless, Sanders has thrown his support behind Colorado Amendment 69, which would create a statewide universal health insurance system funded by a 10 percent tax on payrolls and other income. In return for these payments, all Colorado residents would receive comprehensive medical coverage with no deductibles. The system would replace employer-provided and private insurance but exist alongside programs like Medicare and Medicaid. (Participants would still be free to buy additional insurance if they wanted.)

An independent 21-member elected board would govern the ColoradoCare system, whose budget of about $38 billion would dwarf the size of the current state government. Proponents, who include Sanders and Gloria Steinem, say that the system would help slow down increases in health care costs and eliminate wasteful administrative costs. Opponents argue the amendment would give Colorado the highest income tax rates in the country and that doctors might leave the state if they weren’t satisfied with the payments the system provided. Abortion rights groups also oppose the measure, concerned that the program would reduce access to abortion due to a 1984 amendment to the state constitution that bans government money from going toward the procedure.

The measure appears unlikely to pass. A recent poll showed widespread opposition to it, with 65 percent of respondents saying they’d vote no.

The first exit polls are out. Perhaps the most notable thing about them is the favorable ratings for each candidate. It seems to match the pre-election polls with Clinton and Trump both being seen as unfavorable by a majority of voters, but with Trump being more unlikable than Clinton. Only 44 percent have a favorable rating of Clinton, but an even lower 37 percent have a favorable view of Trump.

The answer, as you might have suspected, lies in human psychology — and is nicely stated in Bethany Albertson and Shana Kushner Gadarian’s recent book, “Anxious Politics.” People don’t like uncertainty and the emotional arousal that comes with it. And one way that we cope with uncertainty is by seeking out new information. Now if our uncertainty is about whether there’s a lion behind a cluster of trees, seeking out new information seems like a winning strategy. But if we’re anxious about an election outcome, the information we’re seeking might not be out there quite yet. As Albertson and Gadarian write, “although individuals seek information in the pursuit of lowering anxiety, the information that they are attracted to may not help them accomplish this goal.” So keep reading …

Trump’s campaign didn’t feature much in the way of specific policy proposals (no, “We’re going to make a great trade deal” doesn’t count). But one area where Trump has offered specifics is tax policy. The only problem: Those specifics kept changing, sometimes in dramatic ways. Alan Cole, an expert at the conservative Tax Foundation, had an entertaining tweetstorm today on the winding path of Trump’s tax plans. The bottom line, though, remained the same throughout: Trump’s plan would cut taxes on the richest Americans while reducing government revenue by trillions of dollars.

Unemployment rate (September): 5.5 percent (U.S.: 5 percent)

Median income (2015): $51,492 (U.S.: $55,775)

Poverty rate (2015): 17.4 percent (U.S.: 14.7 percent)

College share (2015): 27.7 percent (U.S.: 30.6 percent)

The bursting of the housing bubble hit Arizona hard — Phoenix, in particular, experienced a huge run-up in home prices during the boom, and the crash left many homeowners underwater and unable to pay their mortgages. But unlike many states that suffered in the bust, Arizona has seen its housing market rebound strongly during the recovery. Still, the state’s unemployment rate remains above the national mark, and employment has barely returned to its prerecession level.

Today roughly 100 million people are trying to vote across the country, each using a wide range of technologies — many of which fail — facing different ballots and governed by local rules that are enforced by people with varying degrees of experience at the polls. In short, there are bound to be problems. So far, the problems don’t seem unusually severe, though even small problems can prevent people from voting and bigger ones could be on the way as people get off work and crowd polling places closing as soon as a couple of hours from now. Here’s an overview of what we’ve seen so far:

• The biggest story is that this is the first presidential election since a Supreme Court decision removed important protections of the Voting Rights Act, leading to hundreds of poll closures around the country. The full impact of those changes probably won’t be clear until all the votes are counted. It probably exacerbated long lines, already a big problem in 2012. At least the weather isn’t likely to dissuade people from voting. Long lines and polling-place closures have a much bigger effect on vote totals than voter fraud.

• Google is monitoring searches for early signs of trouble spots — the search giant is calling this an experiment. Among swing states, there have been surges in searches related to inactive voters and provisional ballots in Pennsylvania, Michigan, Virginia and North Carolina.

• There have been reports of attempted voter intimidation. In Coral Springs, Florida, police moved Trump voters who were blocking entrances to the polls. In East Lansing, Michigan, a man tried to stop two women wearing hijabs from voting.

• Philadelphia Republicans are objecting to their inspectors being denied access to polling stations, though the city’s DA says all is well.

• Eric Trump is among the people who violated laws in some states against posting photos of your completed ballot.

• There’s lots of confusion about what constitutes illegal voting or polling-place behavior — for example, a Republican operative tailed a bus carrying Philadelphia voters to the polls, which is completely legal.

• Neither of the nation’s two biggest cities are in swing states, but New York and Los Angeles both had widespread voting snafus.

You can read more about voting problems at Electionland and, for coverage of the South, Scalawag.

Assuming a Clinton win tonight, Democrats are likely to pick up several state legislatures across the country. If that happens, one of the more likely to flip control would be Colorado’s Senate, currently controlled by Republicans by an 18-17 majority.

Colorado’s statehouse usually doesn’t attract a great deal of national attention; it functions relatively well and hasn’t suffered a great deal of shutdowns or scandals. But here’s something new: It is one of the most rapidly polarizing legislatures in the nation. According to Boris Shor’s research, Colorado now rivals California for the title of most polarized state legislature.

Should the Democrats achieve unified control of the state government today, that could mean considerable leftward movement by the state in the near future. This would fuel even more anger among the state’s active Republicans.

In this era of partisanship, there’s one issue that unites both Democrats and Republicans representing my community in western Colorado — the exorbitant price of health insurance premiums for plans purchased on the state insurance exchange. Health insurance premiums can cost as much as 50 percent more in Colorado’s resort communities and rural western areas compared to the state’s urban areas. The higher costs are the result of higher fees charged by providers, who have less competition than providers in urban areas. I know firsthand: I live in Delta County, a rural area of western Colorado, and the monthly premiums on my ACA plan are rising 48 percent next year, making them more costly than my mortgage.

About 85 percent of people who buy their own ACA plans receive subsidies, but for many of those who don’t, the cost is quickly climbing out of reach. I’ve talked to a lot of self-insured people in my community who are opting out because they can’t afford the premiums. There are no solutions in sight.

A study released by the Colorado Division of Insurance in August concluded that insurance costs in the state’s most expensive regions could fall by about 20 percent if the state was consolidated into a single insurance region. But if that happened, people in the more populous urban areas of Denver and Boulder would see their rates climb by around 9 percent, making the idea politically unfeasible.

The problem of rising premiums isn’t unique to Colorado. The Kaiser Family Foundation has found that premiums next year are rising as much as 145 percent in some areas.

What makes premiums so expensive in certain areas? A lack of competition for services and insurance providers. That problem is only getting worse.

In our “New Bellwethers” series, we’re examining select counties in key states where Clinton and Trump are polling very close to how they are in the U.S. overall. Our second stop? Berks County, Pennsylvania.

If you’re just joining us, the first episode of “The New Bellwethers” video series — Gaston County, North Carolina — can be viewed here. And Clare’s article on Berks is here.

Unemployment rate (September): 5.8 percent (U.S.: 5 percent)

Median income (2015): $52,431 (U.S.: $55,775)

Poverty rate (2015): 14.7 percent (U.S.: 14.7 percent)

College share (2015): 23.6 percent (U.S.: 30.6 percent)

The real-estate bubble was at its bubbliest in Nevada, which meant the state also experienced one of the country’s worst slumps when the bubble burst. Unemployment neared 14 percent in the state at the peak of the crisis, and it would have been worse if many construction workers hadn’t abandoned the state when jobs dried up. Even today, more than seven years after the recession officially ended, the state still suffers from relatively high unemployment and has a larger share of “underwater” homes (homes that are worth less than their owners owe on their mortgages) than any other state.

Before the first exit polls arrive and Election Day starts to get more tense for each side (as if that were possible), we have one last bit of lighthearted fare from staff at a New York City elementary school today. It’s an election-themed rendition of Billy Joel’s “Movin’ Out.” Sample lyric: “Listening to both of them just yak yak yak yak yak yak yak yak — makes my head hurt and how.”

Listen to the the whole song below:

With recreational marijuana measures on the ballot in five states this election, including California, some people are already voting with their dollars. Indices tracking the value of cannabis-industry company stocks, including this one from Marijuana Index U.S., have seen a sharp run-up in price before the election.

The market could actually see a correction if election-related enthusiasm for the stocks has run them up too far. But in the long run, the sector seems likely to gain more investment. Fifty-seven percent of U.S. adults now believe marijuana should be legalized.

In addition to a measure seeking to legalize recreational marijuana, Massachusetts voters also face another highly controversial ballot initiative: expanding the number of charter schools. If approved, Question 2 would lift the state’s cap to allow up to 12 new charter schools. Polls taken earlier this year found that a majority of the Bay State was backing charter expansion: a February poll by UMass Amherst showed 51 percent were in favor, as did a Western New England University poll in April.

However, by this fall, sentiment has shifted noticeably toward the “No” campaign. Four polls taken since September — two by WBUR, in addition to ones by WNE and Suffolk University — have shown, on average, 46 percent of likely voters are opposed and 42 percent are supportive (with 11 percent undecided).

Charter schools are incredibly contentious in the state. Supporters and opponents of question 2 have raised nearly $27 million for the contest — more than five times the amount spent on the marijuana legalization initiative. The state’s popular Republican governor, Charlie Baker, has been a big backer of the “Yes” side, while the Democratic brass has resolutely for “No” — Sen. Elizabeth Warren, in particular.

Looking for a different perspective on voting patterns? Over at ProPublica, Lena Groeger mapped, for every presidential election since 1828, the counties that voted for a losing candidate.

Unemployment rate (September): 5.7 percent (U.S.: 5 percent)

Median income (2015): $55,702 (U.S.: $55,775)

Poverty rate (2015): 13.2 percent (U.S.: 14.7 percent)

College share (2015): 29.7 percent (U.S.: 30.6 percent)

Pennsylvania wasn’t hit as hard by the recession as states such as Arizona, Florida and Michigan — its unemployment rate never reached 9 percent in the recession — and it saw a relatively strong recovery. More recently, however, that progress has stalled; the state’s unemployment rate has risen by a full percentage point since the end of last year and is now solidly above the national mark. Really, though, it is a mistake to think about Pennsylvania as one state. Philadelphia and its suburbs are solidly part of the Northeastern corridor; Chester County, west of Philadelphia, is one of the richest counties in the country. But western Pennsylvania has never fully recovered from the decline of the steel industry; even Pittsburgh, which has experienced a revival in recent years, is still losing population.

How do you catch a cloud and pin it down? Back in 2008, researchers in Italy and Canada proposed an idea that makes a hell of a lot of sense, at least on the surface. Maybe undecided voters have already made up their minds, but just unconsciously — they haven’t quite made explicit decisions about things they already believe. By that theory, you might be able to predict their votes by looking at the results of the same kind of speed association tests used to measure implicit racial bias. They tried it out and found that, while spoken beliefs were good at predicting the behavior of decided voters, these unspoken implicit beliefs were better at predicting what undecided voters would do. And, lo, the pollsters rejoiced.

Until, that is, somebody tried to replicate the study in the context of an actual election. A different team of scientists tried to use the implicit bias technique to predict how undecided voters would vote in the 2008 U.S. presidential election and the 2009 German parliamentary elections. In both cases, the theory failed. In fact, their research found that implicit beliefs were a better predictor of decided voters’ choices, rather than undecided ones’. (Newer research has since turned up results that don’t match either study.) Why all the disagreement? The authors of the 2012 study think it might be because different kinds of elections have different psychological implications. In a presidential election, for instance, voters have probably thought about the politics involved, asked and answered questions to themselves, and established some kind of personal rhetoric. In other words, their implicit and explicit beliefs might be more likely to match.

A month ago, we took a look at where the two candidates had the most “upside” in this election, relative to 2012. Polls show that Trump will likely outperform Romney among non-college-educated whites, and Clinton will outperform Obama among college-educated whites and all non-whites. We hypothesized a scenario where:

• One in five non-college-educated whites who went for Obama in 2012 would switch their vote to Trump.
• One in five college-educated whites and non-whites who went for Romney in 2012 would switch their vote to Clinton.

Here’s how 2016 would look in this “Vote Swap” scenario:

In our final general election polls-only forecast, this outcome looks like a very real possibility. We give Trump a 70 percent chance in Iowa and a 65 percent chance in Ohio. Hillary is a narrow favorite in North Carolina at 56 percent. Every other state is projected to vote the same way it did in 2012 and Maine’s second district is a nail-biter — we put Clinton at 51 percent.

For most of this year, we’ve been saying that the economy is a roughly neutral factor in the election, good enough not to hurt the incumbent party, but not good enough to help much, either. Recent economic data has mostly stuck to that narrative: Hiring has been slower but still solid; inflation is rising but still muted; overall economic growth is picking up but remains disappointing.

The last big economic report before the election, however, brought some news that might put a spring in voters’ steps. Friday’s jobs report showed that hourly earnings rose in October at the fastest pace since 2009. The report probably came too late to sway many voters, but it is the latest sign that wage growth, long a weak spot in the recovery, is finally picking up.

The first election results won’t start rolling in for a few hours yet, but Republican control of the House of Representatives looks pretty safe. The GOP holds a 247-to-188 seat majority, and Democrats are likely to gain between five and 20 seats. Here’s a look at the most competitive races and how to interpret results as they come in, as compiled by FiveThirtyEight contributor David Wasserman.

On the FiveThirtyEight Elections podcast Monday, my colleague Clare Malone said that this election has felt like one that has centered on Trump. I tend to agree. He’s certainly camped out in my brain space for the past 18 months. But I wonder if a case could be made that the animating force on the GOP side has been a white-hot hatred of Clinton and the establishment she’s seen to represent. We know, for instance, that more Trump voters are motivated to vote against Clinton than for their own candidate. It has certainly felt like this election is a referendum on Trump, but maybe Trump is simply the most extreme standard bearer for what was going to be a referendum on Clinton all along.

I wonder what my colleagues and you, dear readers, think. In the end, was this election more “about” Clinton or Trump?

By now, there have been numerous reports of long lines and crowded polling places, and the possibility of high Latino turnout in Nevada and Florida. If it turns out to be true that this is a high-turnout election, what will that tell us? Political science has turned up a great deal of evidence to help us understand why some individuals vote and others don’t, but understanding the variation from election to election is more complicated. Polarization seems to be good for turnout, as more people have strong preferences about who wins. And that may be primarily what’s at work this year.

But there may be something different going on, too. If some voters went to the polls because of a sense of existential threat, that might be a qualitatively different thing from party polarization. It might also shed light on the question of whether negative campaigns depress turnout – or perhaps help to raise it? With more data, we may also be able to learn more about the potential for “ground game” to get out new and non-habitual voters.

When urbanites learn that I raise chickens on my farm, they often recount an episode of the comedy sketch Portlandia where two diners ask a series of escalating questions about the chicken they’re about to order: How was the chicken raised? How big is the area where the chickens can roam free? What was the chicken’s name? (If you must know … mine are woodland and orchard-ranged, they roam on about five acres, and I don’t give them names.) These are earnest questions with answers that are hard to verify if you’re buying your food far from the farm, which may explain why measures to ensure that farm animals are raised in a humane way have made their way to the ballot in recent years.

In California, Proposition 2 passed in 2008, set rules prohibiting certain methods of confinement for chickens raised for eggs; a 2006 Arizona proposition bars tethering or confining pregnant pigs or calves raised for veal, and a 2002 Florida measure specifies the way pregnant pigs can be housed.

Now Massachusetts is getting a say on the issue with Question 3, the Massachusetts Minimum Size Requirements for Farm Animal Containment, which would prohibit the sale of eggs, veal and pork produced from animals confined in such a way that they were prevented from lying down, standing up, extending their limbs or turning around. Recent polls suggest that support for the measure outweighs opposition. Supporters include organic farmers and environmental and animal rights groups such as the Sierra Club and the Humane Society of the United States. Opposition comes from industry groups such as the National Pork Producers Council, the Massachusetts Farm Bureau and the National Association of Egg Farmers, who say it would significantly increase the cost of these products.

Bellwether counties show us the election in microcosm. By mirroring at a local level what’s going on nationally, they offer a more human look at the political trends roiling the country.

“The New Bellwethers” is a series by FiveThirtyEight senior political writer Clare Malone that examines a few select counties in key states where Clinton and Trump are polling very close to how they are in the U.S. overall (according to Morning Consult polling data collected from August through early October).

Our first stop? Gaston County, North Carolina.

To read Clare’s article on Gaston, click here.

We’re still several hours away from getting any actual election results, but investors seem to like whatever signs they’re seeing. The major U.S. stock indexes all ended the day higher, though they gave up some of their gains from earlier in the day. The S&P 500 closed up 0.4 percent, while the Dow Jones industrial average rose 73 points. (The Dow had been up more than 140 points earlier in the day.)

As I noted earlier, some recent research has found that investors are rooting for a Clinton win. The Wall Street Journal’s Paul Vigna earlier today suggested that the markets were responding to real-time election estimates based on data from Votecastr. Those estimates, which are being published by Slate and Vice, have been generally favorable for Clinton. (At FiveThirtyEight, we’re watching the Votecastr experiment with interest but aren’t relying on its estimates.) It’s hard to know how much weight investors were giving to Votecastr’s estimates or other election-related tidbits; in fact, there’s no way to know for sure that today’s market moves were driven by the election, period. But at the very least investors don’t seem too concerned by what they’re seeing at the polls.

Trump vs. Clinton isn’t the only divisive question on the ballot this Election Day. Many voters will be considering marijuana legalization as well. Over the last 20 years, 25 states and Washington, D.C. have legalized or decriminalized medical marijuana, 15 have legalized the use of cannabis oil and four (plus D.C.) have approved recreational marijuana entirely. And today, nine more states are voting on marijuana measures — five (Arizona, California, Maine, Massachusetts and Nevada) to legalize recreational marijuana, and four (Arkansas, Florida, Montana and North Dakota) to legalize medical marijuana.

Amidst all the legal changes, the country has become a whole lot more comfortable with the idea of legal pot: a Gallup poll shows support climbing from 25 percent in the mid-90s to 58 percent in 2015. Support has increased in all age brackets, and young adults now overwhelmingly support legalization (71 percent of them). “It’s our generation’s Vietnam,” said Brian Vicente, an activist and partner at the marijuana law firm Vicente Sederberg, of the current push to legalize the drug.

Throughout the night, we’ll be bringing you a look at the ballot initiatives in these nine states, and whether they’re likely to pass. Activists, at least, are convinced tonight will be a banner moment: “The country that started the war on drugs and spread it around the world is now having a rebellion within,” Vicente said. “I think it’s going to be a really big, historical night.”

CORRECTION (Nov. 8, 4:35 p.m.): An earlier version of this post listed Montana as one of the states voting on whether to legalize recreational marijuana. It is voting on whether to allow medical marijuana.

In Colorado, incumbent U.S. Sen. Michael Bennet is running for reelection against El Paso County Commissioner Darryl Glenn. Although Colorado is a competitive state, and it tossed out incumbent Sen. Mark Udall just two years ago, Bennet appears to be cruising towards an easy reelection.

Why?

Given the makeup of Colorado’s electorate in presidential years and the recent economic growth in the state, Bennet was likely to be a favorite for reelection. But Glenn also proved to be a poor challenger. His visibility in the state remains surprisingly low, in part due to very modest campaign funds — he has raised only about a fifth of what Bennet has. Glenn’s disadvantage also stems from some personal communications choices. That is, he has refused to speak with the Denver Post, the newspaper with the largest circulation in the state. Finally, his main advertisement — his one chance to introduce himself to Colorado voters — consisted of videos of him lifting weights. Coloradans value health, but apparently not enough to unseat an incumbent.

As I waited in line this morning to vote in Queens, poll workers kept calling people with last names that start with letters between N and Z to move forward, past me, because their lines were shorter. That left me plenty of time to wonder if our usual way of splitting up last names makes any sense. Turns out, it doesn’t.

Sure, there are 13 letters between A and M in the English alphabet, and 13 between N and Z. But voters’ names aren’t evenly distributed between the two groups. TargetSmart, a company that provides voter data to campaigns, looked at more than 200 million registered U.S. voters and found that 63 percent of voters fall into the A-M group and just 37 percent in the N-Z group. Splitting the A-Ks from L-Zs would work much better, with 49.4 percent of voters falling in the first group. (Larkowski is the median last name if you lined up all voters in alphabetical order.)

I also wondered if the A-Ms were different demographically from the N-Zs. TargetSmart checked and found that they were largely the same by age, gender and race — unlike many other voter-access inequities.

Philadelphia District Attorney Seth Williams said that despite widespread concerns about voter intimidation and voter fraud, today’s election in Philadelphia has been normal and without major incident.

“We have no founded complaints of intimidation, no founded complaints of voter fraud,” he said during an afternoon news conference. “We have no walking apocalypse of zombies voting all over town. We don’t have it.” The zombie quip is a reference to claims of voter fraud involving people who have died still “casting votes.”

Williams said that his office’s Election Fraud Task Force had received 69 calls as of 2 p.m., which is on par with the number of calls the task force has received the last three presidential elections. Among the calls received by the task force, there were:

• 13 complaints of electioneering
• Five calls about illegal assistance
• 15 calls about voter assistance
• 10 calls of machine problems

The task force deployed teams of investigators to look into more than 30 complaints today, but the dispositions of those complaints haven’t yet been released. Most of the calls received by the task force were associated with people going to the wrong polling place.

Williams’ assertions though that today is a normal Election Day doesn’t jibe with claims from Philadelphia GOP executive director Joe DeFelice that “voter suppression, disenfranchisement and intimidation” is happening at polling places across Philadelphia. DeFelice told Billy Penn that Republican poll inspectors were denied entry or thrown out of polling stations across the city. However, Assistant District Attorney Andrew Wellbrook said no one had called the task force’s hotline to complain about the issue since 8 a.m. this morning.

The district attorney also said he was aware that Twitter was abuzz about the activities of James O’Keefe, a conservative activist who tweeted that he was “in Philadelphia tailing a pastor’s bus that’s bussing people to the polls.” Williams emphasized that taking people to the polls isn’t illegal.

This post was produced in conjunction with BillyPenn, a website covering the Philadelphia area.

FiveThirtyEight updated its presidential and Senate models for the last time this morning. But a few polls conducted recently came to our attention after we shut the models down. We won’t add them to our forecasts, but they will go into our polling archive. What did they show?

In Georgia, 20-20 Insight put Trump ahead of Clinton, 48 percent to 45 percent. That generally matches most polls out of the state. The poll also showed Republican Sen. Johnny Isakson up on Democrat Jim Barksdale, 51 percent to 44 percent. That’s somewhat worse for Isakson than our polls-plus forecast.

In Florida, St. Pete Polls found Clinton ahead 48 percent to 45 percent. Clinton is forecast to win by a slightly smaller margin in our polls-only model. The same survey has Democrat Patrick Murphy roughly even with Republican Sen. Marco Rubio at 46 percent. Most surveys have given Rubio an edge.

Finally, a national poll from Democratic pollster Mark Mellman puts Clinton in the lead with 47 percent to Trump’s 43 percent. Our national polls-only forecast has Clinton winning by about 4 percentage points too.

My home state of Michigan is one of only 10 states to allow straight-ticket voting, where voters can choose the same party for all offices by making just one selection (though nonpartisan offices must still be individually selected). The practice is meant to expedite the voting process and reduce long lines like those being reported today. A state law passed earlier this year attempted to ban straight-ticket voting but was blocked in a federal district court; the judge said that removing the option would increase wait times and create a “disproportionate burden” on minority voters, especially African-Americans. The 6th U.S. Circuit Court of Appeals refused to reinstate the law, and in September, the Supreme Court issued an order refusing to revive the ban.

Although Michigan attempted to get rid of straight-ticket voting, other states might see the appeal. Long early-voting lines in Nevada resulted in one polling station staying open nearly three hours late; the incident is now the subject of a lawsuit from Trump’s campaign. Voters using their straight-ticket option in Michigan today can at least know that they are doing their part to get out of the polling booth as quickly as possible.

’Tis the season for Californians to alternately curse the state’s long proposition rap sheet, which includes 17 measures this year (there were 18, but one was pulled), and praise direct democracy. As a recent transplant, I wondered whether I needed to get used to the 223-page book describing all the ballot measures or whether this year was something of an anomaly.

It turns out that although this year does have more propositions than average, 17 isn’t that weird, depending on how far back you look. Over the last 40 years, California has had a median of 11.5 statewide ballot proposals each election (that includes all of the midyear and November elections held every two years), though there have been fewer on average in the last decade. According to Ballotpedia, there have been about 120 every decade since 1910. There aren’t even half as many statewide measures up for public vote this year as there were in 1914, when 48 propositions were on the ballot, the most in any one election.

San Franciscans, on the other hand, have reason to gripe. Only twice since 1961 have there been more local propositions than there are this year, according to the city’s Department of Elections. Between the city and state, voters have to decide on 42 measures.

The final tally is in. Of the 100 biggest newspapers by circulation, 57 endorsed Clinton and just two endorsed Trump — fewer than the number that endorsed Gary Johnson (four), “not Trump” (three) or none of the above (five). (The rest didn’t endorse.) Before the final numbers were in, Milo Beckman wrote on FiveThirtyEight that Clinton’s domination of the vote by newspaper editorial boards was unprecedented in modern times.

One of the two newspapers that endorsed Trump is the Las Vegas Review-Journal, owned by conservative Sheldon Adelson, who has spent millions on the effort to defeat Clinton. The Review-Journal erroneously tweeted Friday that Nevada early-voting polling places close at 7, when some closed at 8 or 9 p.m. That’s the night of voting now being contested by Trump lawyers, who contend that people were allowed to join lines after polls closed.

Floridians are voting today on whether to amend their constitution to include the right to generate solar electricity. The catch is that there is no reason to believe they don’t already have that right. It’s the second half of the initiative that has the potential to change things.

Florida’s Amendment 1 would also “ensure that consumers who do not choose to install solar are not required to subsidize the costs of backup power and electric grid access to those who do.”

In plain speak, that gets at the question of whether utility companies can charge solar users extra fees. Utility companies argue that when solar users do tap into the grid — in the evening or when it’s cloudy — they take advantage of the infrastructure without paying the full fee for its upkeep. In turn, utility companies say, if they don’t charge fees to solar users, they would have to raise everyone else’s bill. Solar advocates argue that solar users provide a net benefit to the grid and that fees would disincentivize solar use at a time when renewable energy should be promoted.

Environmentalists and utility companies will disagree, but most of the debate surrounding this initiative has been over the way it’s presented on the ballot. Floridians might think they are voting on a pro-solar amendment, because of the initial part about the right to generate solar electricity. The amendment is actually backed by utility companies, which contributed most of the $26 million raised in support of it.

In March the Florida Supreme Court ruled that the measure was not misleading and approved it for the ballot.

Unemployment rate (September): 4.7 percent (U.S.: 5 percent)

Median income (2015): $49,426 (U.S.: $55,775)

Poverty rate (2015): 15.7 percent (U.S.: 14.7 percent)

College share (2015): 28.4 percent (U.S.: 30.6 percent)

Few states were hit harder by the housing bust than Florida, which lost close to a million jobs between 2007 and 2009. But the state has seen a relatively strong rebound since then; unemployment has fallen below the national mark, and job growth has been steady. Still, scars remain: Home prices are nowhere close to their bubble-era peak (not that that’s necessarily a bad thing), and many families haven’t recovered the savings they lost in the crash.

As of this morning, more than 2.2 million Coloradans had already cast their votes — 86 percent of the 2.6 million who cast votes in 2012. Normally, I’m skeptical of what early votes can tell us about final vote counts, but given how much of Colorado’s electorate has already shown up, we can probably make some meaningful inferences.

The chart below shows the daily early vote counts by party since last week. Unlike in previous elections, where Republicans had a large early vote advantage that dissipated toward Election Day, the parties have been pretty close to parity this year. Democrats had slightly more voters than Republicans early on, but Republican voters took a small lead this week.

But that’s really not the whole story, as roughly a third of Colorado’s registered voters are unaffiliated with a party. And unaffiliated voters have been relatively slow to vote this year. They usually end up comprising about 37 percent of the electorate, but have only made up 30 percent of the vote so far this year. Unaffiliated voters may be facing a more difficult choice this year than usual given the historic unpopularity of the two major party candidates. Many are likely having a hard time deciding for whom to vote, or even whether to vote at all.

What’s also making this cycle a bit odd for Colorado is the fact that it’s only been treated as a swing state for the past two weeks. For a while, it looked to be safely in the Clinton column, and the U.S. Senate race hasn’t been very competitive. But Trump became more popular by the end of October, Clinton only maintains a small edge in polls right now, and the candidates have been spending and visiting frequently lately. This will likely boost turnout, but many Coloradans have only just started paying attention to the race.

A Nevada judge has just denied the Trump campaign’s request that ballots from early voting sites be isolated, but here is just a little more on the details of the lawsuit. What seems to be at the crux of Trump’s argument are allegations that Clark County election officials were letting voters join an early voter line after the time that the polling places had officially closed.

As Ned Foley, director of election law at the Ohio State University’s law school, wrote a couple of days ago, “If it were indeed true that the polls were staying open ‘to bus’ in extra voters who had not been waiting in line at closing time, that would be a violation of the same state law. But I’ve searched for news reports of any such busing in of extra voters, or indeed any casting of ballots by voters other than those already waiting in line, and I haven’t been able to find a single such report.”

Anna, speaking of the youth vote: One ballot measure I’m watching is Prop F in San Francisco, which would lower the voting age to 16. You mentioned that two Maryland towns already allow 16- and 17-year-olds to vote; so do countries including Argentina and Brazil. Polling shows that the San Francisco vote could be very close.

We’ve gotten a lot of questions about Slate’s Votecastr. Slate hopes to tell people who is winning at any time “by combining proprietary, large-sample polls taken prior to Election Day with targeted, real-time tracking of voter turnout on Tuesday.” We here at FiveThirtyEight don’t dismiss the method out of hand, but it is untested. Therefore, we’re not relying on Votecastr in any capacity, although we are interested in seeing how predictive it ends up being.

When I cast my ballot this morning, I was pleasantly surprised to hear the election worker ask the child behind me if she would like a ballot. No, this wasn’t a rare act of voter fraud; it was a kids ballot on offer. It won’t count in this election, of course, but it could have an effect on future votes.

There’s evidence that children who get civics education are more likely to vote when they become adults. This may be particularly true for young people who won’t go on to college (education is one of the factors most strongly correlated with voter turnout). There’s also some evidence that pre-registering 16- and 17-year olds, which is allowed in at least Florida and Hawaii, increases voter turnout for young adults, as do same-day registration laws. That’s important because voter turnout for 18- to 24-year-olds has been consistently lower than for other age groups since the voting age was lowered to 18 in 1972.

There are a couple of places where kids can vote, however. In 2013, Takoma Park, Maryland, became the first city in the U.S. to allow anyone 16 and older to cast ballots in local elections.

Conservative rabble rouser James O’Keefe of Project Veritas released a little video today saying that he was tailing what he called a “pastor bus” in Philadelphia, “bussing [sic] people around, maybe they shouldn’t be doing it.”

To be clear, providing transportation for people going to the polls is not illegal, and many black churches around the country have pursued a “souls to the polls” program in election years to encourage their church members to vote. That O’Keefe is in Philadelphia is no coincidence, though; Pennsylvania is a state that’s key to Trump’s path to victory, and in order to win he’ll need more white voters in the western part of the state to turn out. If Philadelphia’s strong minority community turns out in high numbers, that will be good news for Clinton.

Democrats are certainly worried about the intimidation of minority voters in the Philadelphia area, but a federal judge just ruled against a court order that would have banned what Bloomberg called “aggressive polling-place activity such as invasions of physical space, aggressive questioning and veiled or actual threats of physical violence that they claim could chill the turnout for Clinton,” by Trump supporters, most notably, Roger Stone. Courts ruled similarly in Ohio, New Jersey, Nevada and Arizona.

The Supreme Court unanimously refused to overturn the Ohio ruling, with Justice Ruth Bader Ginsburg noting, “Mindful that Ohio law proscribes voter intimidation … I vote to deny the application.”

We’ve heard a lot about democratic norms in this election, and people like to gasp about how “un-American” it is to threaten others’ right to vote. Unfortunately, though, as University of Massachusetts-Amherst political scientist Jesse Rhodes describes, violence and intimidation have been fairly common in U.S. history. But this history is largely missing from the text of the Shelby County v. Holder decision, which struck down large parts of the Voting Rights Act in 2013.

The way the Voting Rights Act worked is that a number of areas — not all of them in the South — had to submit any changes to local voting rules to the Department of Justice for “preclearance.” In the 2013 decision, the court held that the standards for which areas had to comply with this were outdated and called on Congress to update the standards. Yes, that U.S. Congress. They haven’t passed legislation to update the act, and a number of questionable measures have gone into effect around the country, though some have been struck down in court. What’s more, these kinds of restrictions work in concert with threats of violence, by calling into question the legitimacy of minority votes and spreading the false idea that fraudulent voting is common. The Shelby decision reads, “History did not end in 1965.” This year has shown us that this is true, but maybe not in the way Chief Justice Roberts meant.

With all the attention being paid to the presidential and Senate races, you shouldn’t forget that 12 governorships are also up for grabs today. Democrats currently hold eight of those; Republicans hold the other four. It’s possible that by the end of the evening, those numbers will flip and Republicans will hold a majority of those statehouses.

Democrats will probably hold on to Delaware, Oregon and Washington. In Delaware, Democrat John Carney looks likely to succeed Democratic Gov. Jack Markell. In Oregon and Washington, Govs. Kate Brown and Jay Inslee should be re-elected with ease.

Republicans are favorites to hold on to North Dakota and Utah. North Dakota Republican Gov. Jack Dalrymple will probably be replaced by Doug Burgum. Meanwhile, Utah Gov. Gary Herbert could be re-elected by over 30 percentage points.

That leaves seven races that could conceivably go either way. Republicans are hoping to replace term-limited or retiring Democratic governors in Missouri, New Hampshire, Vermont and West Virginia. Democrats are hoping to pick off Mike Pence’s gubernatorial perch in Indiana. Two incumbents may also go down in Montana (Democrat Gov. Steve Bullock) and North Carolina (Republican Gov. Pat McCrory).

We’ll be keeping an eye on all these races as the evening goes along.

North Carolina is one state that will be watched closely tonight as the results roll in — Democrats are worried about turnout, and many are blaming the closing of polling places and the restriction of early voting hours on what might be decreased turnout from last election cycle.

But Republicans in the state are heartened by the lower turnout thus far. The state party sent out a release trumpeting what it called the “crumbling” Obama coalition, specifically pointing out the decline in the black vote:

African American Early Voting is down 8.5 percent from this time in 2012.

Caucasian voters early voting is up 22.5 percent from this time in 2012.

As a share of Early Voters, African Americans are down 6.0 percent, (2012: 28.9 percent, 2016: 22.9 percent) and Caucasians are up 4.2 percent, (2012: 65.8 percent, 2016: 70.0 percent).

Nevada — don’t worry, I said it right — is a key state to watch tonight: It has closely contested races up and down the ballot. In recent years, though, the state has been home to some well-known polling errors. In 2010, polls showed Sharron Angle unseating Democratic Senate leader Harry Reid, but they were off by more than 8 percentage points. In the 2012 presidential race, the polls were off by almost 4 percentage points. Even though some leading pollsters put Trump ahead in Nevada, he isn’t the favorite there.

One potential reason for the polling errors? Not all polls interview Spanish-speaking voters, and their attitudes differ from English-speaking Latinos. With the support of the Russell Sage Foundation, Cheryl Kaiser, Efrén Pérez and I just completed a survey conducted through GfK of 452 Latino adults. Thirty-five percent of our respondents took the survey in Spanish, while the remainder took it in English. And attitudes differ markedly between the two groups. Among those taking the survey in English, 51 percent backed Clinton to 18 percent for Trump, giving Clinton a 33 percentage point margin. But among Spanish speakers, the gulf was much more pronounced, with 74 percent supporting Clinton versus 7 percent for Trump. The gap diminishes slightly when we restrict the sample to respondents who were listed in voter registration files as of spring 2016. Still, among that group of Spanish-speaking registered voters, Clinton bests Trump 69 percent to 11 percent.

So don’t be surprised if Clinton outperforms some polls conducted only in English, especially in states with sizable numbers of Spanish-speaking voters.

Many polling problems are the result not of malign intent but of basic human error. After all, voting is decentralized, often complicated and confusing — and the people in charge at the polls are either new to the process or likely rusty since not every day is Election Day. Honest error seems like the cause of a bad screw-up by poll workers in two Greenville County, South Carolina, precincts: They asked voters about their party preference, an invasive question that isn’t supposed to be asked. The Greenville News reported the story as part of the Electionland partnership.

At a polling place at 100 S. Broad Street in Center City, Philadelphia, this morning, there was a long and slow-moving wait to get in, and inside, lines that snaked through the room to get to a voting machine.

Aaand if you want to look for yourself, here is the Trump Nevada lawsuit in full. Most notably, it starts off, “This action is to preserve the status quo so that, if the election of presidential electors from the State of Nevada is contested [emphasis mine], the candidates and courts must redress egregious violations of Nevada election law perpetrated by the Registrar on November 4, 2016.”

Why is it illegal to post ballot selfies like Eric Trump did today on Twitter (before deleting it)? One big argument for banning the practice is that it could be used as a way for the selfie taker to prove that she or he voted for a certain candidate — whether to satisfy a coercive employer’s demand, to earn a bribe for the vote or to convert on a vote swap. Opponents of the bans say there’s no evidence any of that really happens.

In the final installment of “Harry’s History,” Harry Enten examines the impact that the networks’ early call may have had on voter turnout back in 1980. And he explains how the events of that year influence when the call might be made as the ballots are counted tonight.

If you’re just joining us, find the previous episodes of “Harry’s History” here.

A little more detail on the Trump campaign lawsuit in Nevada: According to CNN, Trump’s lawyers filed briefs Monday night asking that early vote ballots cast past the poll’s official polling time not be “co-mingled or interspersed” with other ballots.

To reiterate the law, voters in line at the time a polling place officially closes are still entitled to cast their ballots, so it’s not entirely clear what the Trump team hopes to accomplish.

Per Nevada reporter Jon Ralston, Clark County issued a statement in response to the Trump suit: “The petition from the Trump campaign is a request to have us preserve the records from early voting. This is required by state law, and so it is something we are already doing.”

It’s been a long election. Throughout the day, I’ll be highlighting some of the moments that made me say, “Whoa, can you believe that was this election?!” (Here’s an earlier entry.)

Remember when Ben Carson couldn’t hear his name during a debate entrance and caused a bit of a pileup backstage?

Remember when Jim Webb cracked a smile when describing how he’d killed a man in Vietnam?

Remember when Jeb Bush had a gun named “America” (or at least implied that the definition of America is owning an engraved gun)?

Remember when the bird landed on Bernie Sanders’s podium?

If you have any nominations for “remember when,” get in touch by Twitter.

Joe DeFelice, the chairman of the Philadelphia Republican City Committee, said numerous Republican poll inspectors have been denied entry or thrown out of polling stations across the city and that in at least one case a poll worker was spotted entering a machine and pushing buttons for a voter.

“It’s voter suppression, disenfranchisement and intimidation,” DeFelice said. “Everything they claimed we were going to do.”

The Department of Justice has contacted DeFelice about the issue with inspectors, he said. DeFelice also filed a complaint with the Philadelphia District Attorney’s Office and plans to send more this afternoon. Cameron Kline, a spokesperson for the District Attorney, said the Office is aware of three incidents regarding Republican poll inspectors being turned away. With one Republican inspector per polling place, there would be about 1,600 in the city.

The DA’s office has scheduled a press conference for 2:30 p.m.

DeFelice said the problems with Republican inspectors have been happening throughout the city but the biggest issues have been in West Philly. At Lamberton Elementary School in the Overbook section, he said, the Republican inspector was forced to sit outside in a hallway and wasn’t allowed back into the polling room until a GOP attorney assisted her.

Inspectors are the main workers at polling stations. They can help sign in voters and monitor the polls. In Philadelphia, there are supposed to be majority (Democratic) and minority (Republican) inspectors at each station. Given Philadelphia’s hefty Democratic majority — a 7-to-1 registration advantage — the minority inspectors can sometimes be difficult to find. Greg Paulmier, a Democratic ward leader in particularly blue Northwest Philly, said there were frequently no Republican minority inspectors in the area in the past. That was expected to change this year, with more Republicans stepping up for the job and even more appointed by court last month.

DeFelice claimed the same things were happening in 2012 and that he warned the District Attorney’s Office about it. This time, DeFelice said, the treatment of Republican inspectors is worse.

“They knew this was going to happen and they still ignored it,” he said. “The DA was notified and the Court of Common Pleas was notified and here we are.”

This post was produced in conjunction with BillyPenn, a website covering the Philadelphia area.

FiveThirtyEight favorite Jon Ralston just made a very good point on Twitter.

Ralston is talking about how Trump will win a larger geographic area than Clinton in Nevada and still probably lose the state. That’s because most of the vote in Nevada is contained in just one county (Clark). Ralston’s point can be expanded to the rest of the nation as well. Trump will win more land area than Clinton, but Clinton will win the small areas where most of the people live. We’ll end up with a map relatively close to the 2012 results.

All kinds of problems can prevent people from voting. They can include outright intimidation to routine voting-machine problems to incorrect voting guides that tell people, for instance, that they all need to bring IDs to the polls. (Some states don’t require IDs.) A rather humble one is cropping up at polling places around the country, according to ProPublica’s Electionland project: not enough pens. As someone with dozens of pens, most of them out of ink but not yet discarded, I can identify.

This isn’t exactly shocking news, but voters are not psyched about this election. An online exit poll of early and Election Day voters shows that more than eight in 10 Americans just want this election to be over. More than half of voters said they were angry and half said they were sad about the election, and 39 percent said they were depressed. Hopefully, we’ll see the results by the end of the night and make that 85 percent a little less unhappy.

Unemployment rate (September): 4.8 percent (U.S.: 5 percent)

Median income (2015): $51,075 (U.S.: $55,775)

Poverty rate (2015): 14.8 percent (U.S.: 14.7 percent)

College share (2015): 26.8 percent (U.S.: 30.6 percent)

Ohio has long been among the swingiest of the swing states (though perhaps less so this year), and its economy shows why: It’s pretty much the median U.S. state by a wide range of measures. Its unemployment rate is a hair better than the national mark, its median household income a hair worse. But overall, Ohio is pretty much smack in the middle. And like the U.S. as a whole, Ohio is doing better in some regions than others. Unemployment is low in the central and western parts of the state, and Cleveland has experienced something of an economic revival in recent years. But as my colleague Clare Malone wrote last week, parts of state are struggling — and how they could help swing the state for Trump.

It’s now three and a half hours until we get our first look at exit poll data. That data is currently under quarantine. I remember a time, however, when that data could leak it out. Back in 2004, the very first wave of exit polls suggested by mid-afternoon that Democrat John Kerry was well on his way to defeating Republican President George W. Bush. Many people believed that the exits were right, and it even caused massive movement in the stock market. It turned out that the data was wrong. The exit polls overestimated Kerry’s support, and Bush went on to to win a small but comfortable victory.

That should serve as a lesson both at 5 p.m. and now. We won’t know who is going to win until the votes are actually counted. Don’t believe in reports you read online that supposedly give you a keen insight of who is going to win. No one knows.

I just voted in New York, where long lines have plagued many polling stations. That includes the one across the street from my apartment. (Note: I’m not registered at this polling station.)

Politico New York reporter Jimmy Vielkind responded to my picture asking why New York doesn’t have early voting. Indeed, New York is one of just 13 states with no form of non-excuse early voting. The unwillingness of New York to adopt voting reform is nothing new to political junkies. You might remember that before the Democratic primary, many asked why it was so difficult to change party registration in New York.

I wonder if after this election cycle, New York, one of the most liberal states in the union, might decide to look at amending its system to make it easier to vote.

A friend in Chicago reports some pretty disturbing events from the polls this morning. She’s African-American and her husband is white, and they had pretty different experiences at their voting location. She was asked for a photo ID; he wasn’t. She only received part of her ballot at first, and they gave her the other half after she’d already started filling out the first part. Her husband also reported that the only other white voter in line, a man wearing a suit, was invited to cut in line because he was clearly “in a hurry.” (The man declined.) I have no idea if this is representative of what’s happening elsewhere, but it’s not encouraging.

Here’s a story that’s just breaking: It appears that the Trump campaign is filing a lawsuit against the Clark County Registrar’s Office for keeping polls open yesterday for two additional hours beyond their close time (this is being reported by CNN’s Jim Sciutto). It’s slightly confounding that they are doing so, since voters are allowed to cast ballots past the official poll closing time, as long as they were in line before the end of official voting hours. As a Clark County spokesman said a couple of days ago, “If there’s a line when closing time comes, we just keep processing voters until there’s no more line.”

Unemployment rate (September): 4.6 percent (U.S.: 5 percent)

Median income (2015): $51,084 (U.S.: $55,775)

Poverty rate (2015): 15.8 percent (U.S.: 14.7 percent)

College share (2015): 27.8 percent (U.S.: 30.6 percent)

The recession hit Michigan harder than nearly any state. The unemployment rate nearly hit 15 percent and the state lost more than 150,000 manufacturing jobs. But thanks in part to the federal bailout of the auto industry, Michigan has since experienced a solid rebound: The manufacturing sector is growing again, and the state’s unemployment rate is below that of the country as a whole. Still, the recovery hasn’t resolved Michigan’s longer-run challenges — a declining industrial base, a large population of workers without a college degree, and cities with some of the highest poverty rates in the country.

Chances are that if you’re reading this blog, you’ve already got a pretty good idea who you’re voting for today (or have already voted). But for those who haven’t, check out the policy “briefing book” we put together for the presidential debates. Our reporters looked at some of the biggest policy differences between Trump and Clinton, and analyzed how their claims align with the facts.

How has technology changed the way we follow election night coverage? In Part 2 of “Harry’s History,” Harry Enten revisits technological firsts that debuted on election night in 1948 and 1952.

If you’re just joining us, Part 1 of the “Harry’s History” series can be viewed here.

Oklahomans are voting on a ballot initiative to enshrine the right to farm in their state constitution — meaning that courts would have to find a “compelling state interest,” of the same level applied to free speech questions, in order to maintain or create any agricultural regulations. It’s a tight election that hinges on whether voters think the amendment would protect farmers from frivolous government interference, or whether they think it will protect big agribusiness at the expense of small farmers and natural resources. And sitting squarely in the latter camp are the state’s largest Native American tribes.

The Cherokee Nation, along with an intertribal council that includes the Chickasaw, Seminole, Choctaw and Muscogee Creek nations, oppose the Oklahoma Right To Farm Amendment. Sarah Hill, the Cherokee Nation’s secretary of natural resources, told me that there’s a philosophical and ideological connection between this opposition and the Standing Rock protests against construction of an oil pipeline under the Missouri River in North Dakota.

The Cherokee Nation isn’t governed by Oklahoma laws. But it shares the same land and water — in particular the Illinois River, which has long been the subject of environmental regulatory fights over chicken farming runoff. If this amendment passes, Hill said, the Cherokee Nation wouldn’t be able to trust Oklahoma to regulate industries that impact them both. “I think what you’re seeing is tribes really stepping into that role [of self-governance],” she said. “You’re going to see more tribes inspired to step up and protect our natural resources because of the work of the water protectors,” which is what the North Dakota protesters prefer to be called.

A political race that is arguably the most televised ever has not provided the expected financial gains for those selling television ad space. In June, Trump bragged that he didn’t need to pay for exposure: “I make speeches, I talk to reporters,” he said. “I don’t even need commercials.” His strategy of relying on news coverage rather than ads for much of the race also meant the Clinton campaign could do less reciprocal spending.

Tim Wesolowski, the chief financial officer of the broadcast group E.W. Scripps, said in a recent company earnings call that in the third quarter, the two presidential candidates spent about a third of what their predecessors spent on the company’s stations in the third quarter of 2012. “Donald Trump in particular spent a fraction of what past Republican candidates have spent on television ads,” he said. The company took in $27 million from political ad buys in that period, half of what it anticipated.

Some critics who rated the ads for their effectiveness and aesthetic qualities also found them lackluster. Doyle McManus of the Los Angeles Times spoke to the creators of Spotcheck, a system UCLA scientists are using to let visitors rate campaign ads in real time, and found that negative ads have the most impact and well-made ads are most memorable. Clinton’s ads have been competent but uninspiring, McManus wrote, while Trump’s are fiery and often untruthful.

Unemployment rate (September): 4.7 percent (U.S.: 5 percent)

Median income (2015): $47,830 (U.S.: $55,775)

Poverty rate (2015): 16.4 percent (U.S.: 14.7 percent)

College share (2015): 29.5 percent (U.S.: 30.6 percent)

In term of its economy, North Carolina is really (at least) two states: The “research triangle” around Raleigh-Durham is well-educated and prosperous. Most of the rest of the state is much less so, with poverty rates over 20 percent in some counties. Overall, the state has been solid but uneven: Wages are rising and the unemployment rate is below 5 percent, but the manufacturing sector has been slow to rebound.

Financial markets were up modestly as of midday as investors mostly played “wait and see” with the results of the election.

There is some evidence that investors prefer Clinton to Trump — stock prices rallied during Clinton’s strong performance in the first debate and dropped sharply when FBI Director James Comey said his agency was looking into a new batch of Clinton’s emails. (They rose again after Comey released another letter saying the emails wouldn’t lead to an indictment.) That has led some economists to suggest that markets will tumble in the event of a surprise Trump victory. But even if that’s right, don’t assume that short-term market fluctuations (in either direction) reveal much about how either markets or the economy will behave in the longer term.

If you need another political distraction (besides this live blog), consider making friends with the ghosts of elections past. We’ve created a few short election docs that re-examine seminal moments from past presidential elections. Take a look.

Media coverage of presidential elections, including ours, understandably fixates on swing states — the ones that could go either way on Election Day. But in a recent paper, two game theorists studied the influence of the rest of the states, the ones we’re almost sure will go to Clinton or Trump. Just how many states — and electoral votes — Clinton is almost assured of winning influences how many of the swing states she’d need to win the needed 270 electoral votes and the election. And the same goes for Trump.

The authors of the study created three measures to get at how a candidate’s baked-in lead in the Electoral College sets up the race: winningness, or the percentage of plausible scenarios in which the candidate wins; vulnerability, or the proportion of those plausible winning scenarios that could be disrupted by the underdog flipping just one state; and fragility, or how many different states in a typical winning scenario would flip the result if they flipped. The bigger a candidate’s lead in non-swing states, in general, the higher the candidate’s winningness and lower the vulnerability and fragility.

At my request, co-author Marc Kilgour, a mathematics professor at Canada’s Wilfrid Laurier University, ran the numbers for the state of the race late Monday afternoon, treating the 16 closest races in our polls-only forecast as swing “states” (we’re really looking at 56 presidential races, not all of them states: the 50 states plus the five races for individual electoral votes in Maine and Nebraska and the contest for Washington, D.C.). Clinton wins in 80 percent of scenarios, and her vulnerability and fragility are about half of Trump’s. That’s similar to Obama’s standing heading into the 2012 election, though his fragility was much lower relative to Romney’s.

A caveat: These measures are more theoretical illustrations of the state of the race than they are forecasts. They treat all swing races as tossups, and independent ones at that, and treat all other races as foregone conclusions. In reality, Clinton leads in most of the close races — but if the forecast is wrong in one race, it’s likely to be wrong in the same direction in other races. That’s likely one reason why we’re giving Clinton a probability closer to 70 percent than 80 percent.

That question could be asked about supporters of any candidate, but it’s been asked most often during this campaign about the people supporting the guy with no political experience who has said and done many things that fall far outside the norms of U.S. politics. Some journalists and researchers have attempted to answer that question with largely qualitative, on-the-ground reporting, sometimes with fascinating results, such as in my colleague Clare Malone’s exploration of why Trump won the nomination and George Packer’s May article for The New Yorker about Trump’s white working-class voters. Others, including some of my colleagues, have used data. Here’s a brief overview of what some of these studies have found:

Economics. Despite Trump’s reputed popularity among working-class voters, his primary supporters were better off than the average American, according to my boss, Nate Silver. A study based on Gallup surveys of 125,000 American adults largely agreed: It found that Trump supporters are relatively high earners without unusually high rates of unemployment nor negative exposure to trade or immigration, though they’re more likely to hold blue-collar jobs.

Race, nationalism and identity. Several studies show that Trump voters have high levels of racial resentment and implicit bias against black people. Trump voters on Twitter are much more likely than Clinton voters to have the phrase “all lives matter” or “blue lives matter” in their bio text. Places where a high concentration of white people tell the Census Bureau their ancestry is “American” have high levels of Trump support. Trump has done very well in counties where the nonwhite population grew fastest, mostly by winning white voters in those counties.

Gender. Sexism is a strong predictor of support for Trump, who is running against the first woman to win a major party’s presidential nomination.

Religion. Writing at FiveThirtyEight, Milo Beckman pointed out that Trump’s white voters are more likely to be religious than Clinton’s.

Authoritarianism and anti-elitism. During the primaries, a poll found that support for authoritarianism predicted Trump support. A follow-up study, though, refuted that finding and pointed to populism — or specifically, anti-elitism — as the main driver.

Statistics for the sake of statistics are fun, right? Sure they are! I Am Greater Than Zero is a website that will automatically generate what your share of the vote is in the country, your state and your county. I think the theory behind this site is that you’re supposed to realize you matter as a voter. But all I can think is, if I’m .00004 percent of the vote in Minnesota, well, then, maybe that time Al Franken made my husband laugh in an airport was about the correct amount of time for my senator to spend on me.

Mainers are voting on how to vote in future elections today. If the state’s Question 5 ballot initiative passes, Maine could become the only state in the country to implement ranked-choice voting (RCV) for statewide elections.

Under the system, instead of voting for a single candidate, voters rank candidates in order of preference. If no candidate exceeds 50 percent of first-choice votes, the last place candidate is eliminated and his or her votes are distributed according to second choices. The process is repeated until one candidate gets more than 50 percent of the vote.

Sound complicated? There’s some evidence that voters think so. RCV is already used in various municipal elections around the country, including in San Francisco. One analysis of San Francisco mayoral elections concluded that there is a relationship between RCV and decreased turnout among less-experienced voters — voters who are younger or less educated. That could be the result of more confusion and error caused by a more complicated ballot.

But many Mainers see good reason for implementing the system, in large part because a Maine governor has not received a majority of the vote since 1998. Proponents of the initiative say the system would help elect more broadly popular politicians, decrease the effect of spoilers, and maybe even decrease negative campaigning.

The latest polling on the initiative shows a plurality of voters in favor, but with about 20 percent still undecided. If the measure does pass, there are still some wrinkles Maine would have to iron out. For one, Maine’s constitution explicitly says that governors can be elected by a plurality, and ranked-choice may violate that. That means a constitutional amendment would likely be necessary, a process that would send the issue back to the ballot in a future election.

Later this month, the Supreme Court will hear a major Eighth Amendment case on the death penalty and intellectual disability. But today, voters will decide the future of capital punishment in three states. I’ll be keeping my eye on these ballot measures today.

In California, voters are faced with two competing referenda. The passage of Proposition 62 would eliminate the death penalty in the state, replacing it with the maximum penalty of life without the possibility of parole. The passage of Proposition 66 would retain the death penalty, and make it harder for inmates on death row to prolong their appeals. Its supporters argue that the current system delays justice, and that Prop 66 would work to bring closure to victims’ families. They also argue it would save taxpayers’ money. If both pass, whichever receives more “yes” votes will take precedence. A September USC Dornsife/Los Angeles Times poll found that California voters opposed eliminating the death penalty, 40 percent to 51 percent.

California has over a quarter of the country’s death-row population, but the state hasn’t executed anyone since 2006.

In Nebraska, voters are considering Referendum 426, the language of which is a bit byzantine. A “repeal” vote would reinstate the death penalty there. Capital punishment was eliminated by the state’s legislature in 2015, that bill was vetoed by Gov. Pete Ricketts, and that veto was then narrowly overturned by the legislature. A “retain” vote would retain the state’s ban on the death penalty. An August poll by Global Marketing Research Services found that Nebraskans supported reinstating the death penalty, 58-30.

There were 10 people on Nebraska’s death row as of July, and the state has executed three people since 1976.

And in Oklahoma, voters will decide State Question 776. A “yes” vote would amend the state’s constitution, guaranteeing the state the power to execute and choose the method of execution, and declare that the death penalty “shall not be deemed to be or constitute the infliction of cruel or unusual punishment”. A “no” vote would defeat this amendment.

There are 47 people on death row in Oklahoma, and state has had the highest per-capita execution rate since 1976. A July SoonerPoll, according to Ballotpedia, found overwhelming popular support for the amendment.

Though every vote counts, they don’t all count the same, as Julia just noted. In this map, each state is sized by its number of electoral votes and colored by the number of citizens over 18 (roughly the number of eligible voters) that each elector represents. Wyoming, for example, has three votes in the Electoral College and almost 435,000 citizens over 18 years old, so one vote there is worth 0.0007 percent of an electoral vote, the highest share in the country. Meanwhile, a single person’s vote is worth the least in Florida, where 14 million adult citizens are represented by 29 electoral votes. A vote there only counts for 0.0002 percent of an elector, nearly four times less than in Wyoming.

In addition to being the most vote-rich battleground state in America, Florida also is home to the Senate race of a former presidential contender, Marco Rubio. Rubio said he would not run for re-election to the Senate while campaigning for president, but he changed his mind in June. Fellow Cuban-American Floridians, who have traditionally voted Republican but may not do so this season, say they’re voting for Clinton for president and Rubio for Senate, splitting their ticket. Rubio currently leads in the polls. He hasn’t ruled out running for president again in 2020.

The common assumption about why we have an Electoral College is that the Founders didn’t trust the people to make the most important national decision. This is mostly wrong on several levels. The people who wrote and ratified the Constitution had lots of different opinions and visions, just as individuals do now. Some were suspicious of popular rule, some championed it, and others were concerned about the practical implications.

But the main reason we have an Electoral College is the same reason we have a Congress with two chambers, one that represents states by population and one that represents all states equally. One of the driving concerns was that a direct election process would allow the states with the largest voting populations to dominate presidential elections. This would be bad for small states, and also for slave states, and the Electoral College was created as a concession to those states. Smaller states still have fewer EC votes than larger ones, but they have far more representation than they would if those were allocated by population alone. The idea was to ensure that these states would be relevant when electing a president. Now the imbalance between large and small states is quite stark. As political scientist Andrew Gelman put it, Wyoming has nearly six Electoral College votes per million people, while California has a little less than 1.5 votes per million people. (The post is from 2004, but the numbers still hold up.)

In that sense, the Electoral College is worse than we thought. But the role it plays in making sure presidential elections are about putting together broad geographic coalitions is more complex than we tend to acknowledge. It has reversed the result of the popular vote a handful of times – 1876, 1888, and 2000, never in the case of a clear victory, and perhaps for that reason alone it should go. But its impact on the political system goes far beyond those cases.

In 2012, FiveThirtyEight got every state and Washington, D.C., right in its final election forecast. Other forecasters also nailed the result — or came very close.

Don’t expect the same perfect score this time. Far more states’ presidential races are very, very close. It’s possible our forecast will get every state “right,” but it’s unlikely. On average, across all our simulations in our final polls-only forecast — our default for the presidential race — we “miss” about five or six states; based on our final forecast in 2012, we’d have expected to miss between one and two. (This counts Washington, D.C., but doesn’t include the five congressional-district-level races in Maine and Nebraska.) The Senate also has lots of very close races; on average, we’d expect to “miss” three.

I’m using quotation marks around “right” and “miss” because the forecast is probabilistic in nature. If you make 10 predictions that each give the favorite a 60 percent chance, you should get about six of them “right” and “miss” four. If all 10 of your favorites win, your forecast probably is miscalibrated.

The number of expected misses above are means, not medians, by the way. They’re somewhat skewed by the possibility of a really major miss across the board, which also shows up in some spikes at the extreme edges in our probability distributions for the Electoral College. The more typical result is three or four misses — and they could cancel each other out. But if they don’t, the misses might really add up because some states with lots of electoral votes — including Florida, North Carolina and Ohio — are close to toss-ups.

Piggybacking on Clare’s earlier post about rigged elections, I had a couple of readers message me during the last debate live blog about a 2014 paper in the journal Electoral Studies that purported to show that non-citizen immigrants vote in large enough numbers to affect election outcomes. It’s based on responses to the Cooperative Congressional Election Studies surveys — massive online polls of tens of thousands of people. Respondents were asked about citizenship status and voting. Some of the respondents in each state had their voter registration status and election participation status verified by the survey takers — including some of the people who marked themselves as “non-citizens.”

The analysis is kind of like an LSAT problem: Jim says he is a non-citizen and voted. Jim wasn’t verified to have voted, but Timmy, who also said he was a non-citizen voter and shares a lot of characteristics with Jim, was. How many non-voters voted?

The researchers used this data to determine that non-citizen voting happens and, while the rates of it are low, it could have been enough to swing a Senate seat in Minnesota for Al Franken in 2008 and give North Carolina’s 2008 electoral votes to Barack Obama.

But there’s some weirdness to this. The researchers, themselves, say voter ID laws are unlikely to stop illegal voting because more than two-thirds of the non-citizen voters they found in the survey had been asked for photo ID before they were allowed to vote. So what gives? According to a different paper, published a year later in the same journal, the best way to explain this apparent discrepancy is that most, if not all, of those non-citizen voters were actually citizens, after all.

Titled, scathingly enough, “The Perils of Cherry Picking Low Frequency Events in Large Sample Surveys,” the paper points out that the citizen/non-citizen question has a known error rate of 0.1 percent. That’s low enough that the survey, as a whole, remains reliable. But even that tiny error rate can create big problems when you’re trying to focus in on a small sub-population directly affected by it. In fact, the authors found that the rate of “non-citizen voting” uncovered in the 2014 paper can be completely explained by very low frequency measurement error.

To add to what Clare said about the rarity of voter fraud in U.S. elections, I spoke recently with  Joseph Kanefield; he’s a partner at the law firm Ballard Spahr’s Phoenix office and served as state elections director in Arizona, as well as counsel to former Gov. Jan Brewer, a Republican. He believes allegations of widespread voter fraud are “very detrimental” to the political process. “Generally speaking, the American voting system is a very secure system” and “these laws that have been put in place have been designed to prevent rampant election fraud that took place in the late 19th century,” Kanefield said, citing the secret ballot as an improvement since American democracy’s early days. “The possibility of a deceased person voting is pretty minimal,” he said. A Slate analysis of the 130 million votes cast in 2012 did not show evidence of widespread fraud, whether through the use of deceased voters’ identities or other means.

Tonight, Clinton could make history. If she wins, she will be America’s very first Madam President. Yet even if Clinton’s groundbreaking presidency becomes reality, it won’t change a hard truth: There’s a serious dearth of women in elected office. Created in collaboration with Rutgers University’s Center for American Women and Politics, this video explores the gender disparity across political offices in the U.S. and discusses the difference that more female leaders, regardless of party, can really make.

Betting markets have long pegged Clinton as the favorite to win today’s election. But this morning her odds of winning have risen further. Betfair and PredictIt, two major betting markets, have shown Clinton’s odds rise over the last 24 to 48 hours. An aggregation site puts her odds at 83 percent — up nearly four percentage points in the last four hours (through 11:45 a.m. eastern time). Betting markets, though, don’t have a sterling record, especially recently. Odds makers infamously miscalled the “Brexit” vote in June; betting markets favored “Remain” solidly, but “Leave” won instead. So be appropriately skeptical.

If reports are to be believed, Tom Brady and Gisele Bundchen might not be supporting the same candidate —  on Monday, Trump said that Brady had called to endorse him, but an enterprising Instagram sleuth uncovered a comment from Bundchen disputing the notion that her husband supports the Republican nominee.

If the Brady-Bundchen household is split, it would not be uncommon in 2016; as we wrote last month, men are treating 2016 like a normal election, but women are not. We could be looking at the largest gender gap in a presidential election since 1952. Clinton is earning women’s support by high double-digit margins across demographic lines, while Trump is favored by men with margins that are typical for a Republican candidate.

That said, married couples do tend to share the same political beliefs, as political scientist Eitan Hersh found. But 30 percent of married households did contain what he called “a mismatched partisan pair”:

During the Iowa caucuses earlier this year, I wrote about one such partisan mismatched pair, Betsy and Nick Sarcone. She’s a Republican and he’s a Democrat, turning our gender/party stereotypes on their head.

Today is the last day this year when voters will be hounded by campaigns for their vote. Do those efforts work? Yes, according to a Yale study, if they’re done correctly. Mass emails are chronically ineffective, but personalized messaging, like door-to-door canvassing, is both effective and efficient for voter mobilization. And it’s particularly useful when focused on Latino voters, a factor that might help Clinton.

Latino voters have reported that Democrats have reached out to them with Get Out The Vote efforts three times more often than Republicans have. These grassroots, face-to-face campaigns resonate with Latino voters. In some swing states with large Hispanic populations, campaigns are asking each Hispanic woman who was dedicated to the cause to recruit five friends, who would recruit five of their friends.

The social factors that have in the past kept some Latino voters away from the polls, like rates of education, income and age, can be offset if they feel personally mobilized to vote. That is exactly what these campaign efforts strive to do.

Oh, sure, it’s Election Day for all of you. But for a few of us (OK, mostly me and Nick Bunker of the Washington Center for Equitable Growth), it’s JOLTS Day. JOLTS — the Job Opening and Labor Turnover Survey — measures how many people start and leave their jobs each month. It’s separate from the better-known monthly jobs report, which measures net job creation but doesn’t give as clear a window into hiring and firing.

Tuesday’s report, which contained data from September, was mostly good news: Employers posted slightly more available jobs, and more people quit their jobs voluntarily, which is a sign of confidence in the job market. Layoffs stayed low. Hiring, however, slowed slightly. Over the longer term, the JOLTS report shows how far the job market has come since the Great Recession. In 2009, there were close to seven unemployed workers for every available job opening. In September, there were 1.5, a better ratio than when the recession began.

Trump voters have a higher level of implicit bias against black people than Clinton voters do, according to a recent study by a market-research firm. Both groups, though, had implicit bias against black people, and Clinton voters were closer to the Trump voters on the bias scale than they were to the absence of bias.

This should come as no surprise to Clinton. Asked about implicit bias among police officers at the first presidential debate, she said, “Implicit bias is a problem for everyone, not just police.”

“Hillary Clinton is right,” said Michelle Niedziela, scientific director of HCD Research, the Flemington, New Jersey, market-research firm that conducted the study. “That is what our tests showed.”

Many previous studies of the 2016 electorate have shown that Trump voters have higher levels of resentment toward black people (and higher levels of negative views of black people) than Clinton voters do. Tests of implicit bias are different. They try to measure people’s unconscious views about groups of people. Test takers are shown names or images associated with, say, women or black people, followed by words tied to certain attributes. They’re then timed on how long it takes them to select words with positive associations (“cheerful,” “joyous”) with the same key they used when they saw a photo of a black person’s face. The test is then repeated with various other combinations of photos (or names) and words. Differences in how long it takes to make a selection can indicate implicit bias.

Previous studies, including ones during earlier elections, found similar results: liberals and conservatives both have implicit bias against black people (and many other groups, including, most consistently and negatively, older people), conservatives have higher levels of implicit bias, and the two groups are, on average, closer to each other than to an absence of bias.

Charges of potential voter fraud had already made their way into the 2016 campaign, even before early voting started, thanks to Trump’s claim that the election would be “rigged.” Even before Trump, though, Republicans and right-leaning media sites tended to play up the notion that elections were significantly affected by voter fraud.

But there has been no significant evidence of widespread voter fraud in modern America. As of 2014, only 31 credible instances of in-person voter fraud had been discovered since the year 2000, according to a political scientist who tracks the claims.

The year 2012 saw numerous claims of voter fraud, but very few were deemed malicious or conspiratorial acts — most were due to human error. In a post-election analysis, one Cleveland area news group found that of the 270 allegations of vote irregularity lodged in Ohio, most were not pursued by prosecutors because they were the result of voter confusion. As the study found, the flagged instances accounted for “less than five one-thousandths of 1 percent of the 5.6 million ballots cast in Ohio in the 2012 election.” Ohio’s Republican secretary of state, Jon Husted, said of voter fraud, “It’s rare.”

Throughout the campaign, it’s been noted that Clinton is more popular when she’s not a candidate. She was, generally speaking, more popular as secretary of state, as first lady and as New York senator than as a presidential candidate (although some of the differences are not that large). And this discrepancy, such as it is, is often attributed to gender factors. The underlying assumption is that people like women just fine as public servants but dislike them as ambitious candidates trying to move up the political ladder. This story isn’t as clear or neat as that — but it’s a central way that people have understood Clinton’s candidacy.

The reverse was true of Obama. People loved candidate Obama more than President Obama. It’s somewhat tough to compare because his and Clinton’s career trajectories have been so different. But in 2008, candidate Obama was quite popular. His political troubles started once he got into office and started promoting policies like the economic stimulus and health care reform. His candidacy was heralded as racially unifying, while his presidency has had the opposite effect. And there’s considerable evidence that race is part of the story of public responses to policy during the Obama years. His approval ratings have rebounded in his final year in the White House, so he’s not totally immune to the benefits of being off the ballot. Still, it’s worth thinking more about how our first potential female president seems to lose people when she’s campaigning, while our first African-American president faced similar consequences for governing.

Today is Election Day — but most people in the U.S., and almost half of voting-age Americans, probably won’t vote.

Just who isn’t voting? Tens of millions aren’t eligible — most because they’re under 18 years old, many others because they have been convicted of a felony or for other reasons.

Back in September, The New York Times profiled the people who can vote but don’t. The younger, poorer and less educated an eligible voter, the less likely she is to vote. Black adults also vote at higher rates than white voters at the same educational level.

In recent presidential elections, turnout has been lower in states where the presidential race isn’t competitive. That’s probably driven by more advertising and get-out-the-vote efforts, plus media coverage (and election forecasts) making clear that only certain states are likely to influence the outcome in the biggest race. But remember, there are plenty of other important contests today.

We ran a final run of the FiveThirtyEight Senate models this morning, and the fight for control couldn’t be closer. The polls-plus forecast gives Democrats a 50.7 percent of having a majority in the next Senate, while the polls-only forecast has Republicans with a 50.3 percent of having a majority. In other words, it’s a true tossup.

Democrats need a net gain of four seats to win control, if Hillary Clinton wins the presidency. That would get them a 50-50 split overall in the Senate with Vice President Tim Kaine breaking the tie. How do Democrats get those four seats?

Right now, both models have Democrats as at least a 60 percent favorite in three seats currently held by Republicans: Illinois, Pennsylvania and Wisconsin. Democrats are also slightly favored to hold their seat in Nevada. That means they need one more seat. Currently, the polls-plus and polls-only model disagree on who’s favored in New Hampshire, whose Senate seat is held by a Republican. The polls-plus model has Democrat Maggie Hassan barely winning, but the polls-only model barely gives it to Republican Sen. Kelly Ayotte. If Democrats fail to pick-up New Hampshire, then they need to win in Indiana, Missouri or North Carolina. Missouri is their best shot in that group, though a Democratic candidate has at least a 25 percent chance of winning in each state.

The bottom line is the battle for the Senate is very close. The most likely single outcome is that Democrats end up with a net gain of exactly four seats, and whoever the vice-president is will have to break the tie.

In the video below, I go through our latest (and last) Senate forecast: Will Republicans hold on to their majority? Or will Democrats manage to overtake them for the first time since 2009?

If you’re just joining us, click here for our final presidential election update video. My final Election Update can also be viewed here.

Trump has been saying for weeks that the election will be rigged — without giving any evidence as to why this might be so (there is none). In the final debate with Clinton, he wouldn’t say outright that he would accept the results of the election. Rather, he said, he would “look at it at the time. I will keep you in suspense.”

On a Florida radio program this morning, Trump reasserted that he would not accept the election results “if I think there’s something that was wrong.”

It’s easy to forget how new our current configuration of voting rights is. Here’s a look at how enfranchisement grew to encompass the modern electorate.

When our nation was founded, only white landowning men could vote. In 1789, when George Washington was elected, this meant effectively 6 percent of the U.S. population was enfranchised.

By 1856, all white men had the right to vote, after North Carolina ended its property ownership requirement.

In 1868 and 1870, with the 14th and 15th Amendments, black men gained (implicitly, then explicitly) the right to vote, although many were prohibited from doing so by systems including poll taxes, intimidation, and violence.

In 1876, a Supreme Court ruling barred many Native Americans from voting, and in 1882 the Chinese Exclusion Act blocked Chinese immigrants from naturalizing. Several other legal cases barred full enfranchisement of Native Americans and Asians for years to come.

In 1920, with the 19th Amendment, women gained the right to vote after fighting for it for decades.

In 1965, the Voting Rights Act made illegal many forms of voter suppression. Voter registration among blacks in Mississippi rose from 6 percent in 1964 to 59 percent in 1969.

In 1971, against the backdrop of the Vietnam War, the voting age was lowered to 18.

The Pew Research Center estimates our current electorate is 69 percent white, 12 percent black, 12 percent Latino and 4 percent Asian-American. In 2012, women were 53 percent of eligible voters.

Back in 2008, my colleague Nate Silver and Columbia University professor Andrew Gelman co-wrote an academic study estimating the probability any one voter could cast the decisive ballot in the presidential election. The average voter’s chance to make a difference was about 1 in 10 million in the most influential swing states — and much, much lower in states that were unlikely to be competitive. This year, the race looks closer, so votes look more important: Voters in New Hampshire, Colorado, Nevada, Wisconsin and Pennsylvania all have a 1 in 2 million chance or better of swinging the election, according to an updated study co-written by Gelman. But voters in just over half of states — 26 — have a 1 in 1 billion or smaller chance.

That doesn’t mean Gelman says people shouldn’t vote. He thinks it can be rational to vote. He and Nate said the same thing in 2008. Just keep in mind another consideration: There are lots of drivers on the road, including some who don’t drive often or are rushing to beat poll deadlines. So please, drive carefully!

Women were granted the right to vote in 1920 by an amendment to the Constitution (though black women’s suffrage was not truly enshrined until the Voting Rights Act of 1965). Ninety-six years later, Hillary Clinton is the first woman to be a major party candidate for president — and she’s a favorite to win the election according to our model. While it’s taken a long time to get to the point where a woman is a serious contender for the highest office in the land, Gallup has been asking the polling question “Would you vote for a woman for president if she was qualified in every other respect?” since 1937. That year, 64 percent of Americans said no, 33 percent said yes, and 3 percent had no opinion on the matter.

Back in June, I wrote about Americans’ responses to that question over the decades — we’ve mostly become steadily more amenable to the idea, though there was a slight decline in favorability for a woman candidate right around the time Clinton began to be mentioned as a strong contender.

Throughout the day, our live blog will include posts from BillyPenn, a website that covers the Philadelphia area. The city is key to Pennsylvania’s 20 electoral votes, and is being closely watched for election issues.

Social media in Philadelphia are full of reports of breakdowns among the city’s old voting machines. We’ve heard of voting machine issues in the neighborhoods of Fishtown, Mt. Airy and Graduate Hospital. But despite the reports, the long lines and the Donald Trump-led speculation over election fraud, Republican City Commissioner Al Schmidt said Philadelphia’s 14-year-old voting machines for the most part are holding up to the early demand.

“It hasn’t been anything out of the ordinary,” said Schmidt. “It’s been a handful. If there were an issue like something extraordinary we would know it right out of the gate.”

Philadelphia has about 4,000 voting machines in use on Election Days. They were tested in advance and delivered to polling locations the last few days. If they break down on their big day, a team of 70 technicians — about a dozen more than usual — will be ready to fix the problems. Schmidt said the technicians are spread throughout the city, making it easy for them to accommodate requests in Philadelphia’s 66 wards.

He said technicians will first try to restart the machine. Sometimes that works, and the problem can be fixed in seconds. If the issue cannot be determined quickly, the machine will be swapped out with a backup.

For those who have been closely following the election, any talk of voting machine failure could raise some eyebrows. Trump, of course, has said the only way he’d lose Pennsylvania is if the election is rigged. And the cyber security firm Carbon Black rated Pennsylvania as the No. 1 place most vulnerable for a hack. In reality, Philadelphia and Pennsylvania likely have little to fear regarding a cyber attack. The machines are too old. They’re not connected to the internet and run on systems like Windows XP.

But the age of Philadelphia’s and Pennsylvania’s machines does matter somewhat. Most voting machines don’t produce a backup paper copy, as is typical throughout the rest of the country. And 14-year-old machines are 14-year-old machines. Last year, City Council declined to approve the $22 million purchase of new voting machines even though it was part of the city budget.

Schmidt said voting machine issues here generally have to do with hardware. One frequent cause is repeated write-in votes. He said sometimes the mechanism that allows people to write in votes can disrupt the machine.

But so far today the number of machines breaking down has not been substantial.

“If you compare presidential to presidential,” Schmidt said, “we’re actually experiencing fewer problems relative to recent presidential elections.”

Earlier today I wrote about the four states where voters will decide whether to raise their minimum wages. Voters in South Dakota get to decide whether to lower the minimum wage for some workers.

The initiative is what’s known as a “veto referendum.” Last year, South Dakota’s Legislature voted to let employers pay workers under age 18 $7.50 an hour instead of the state minimum wage, which is now $8.55. But opponents of the law managed to collect enough signatures to put the issue directly to voters — the ballot initiative, if it passes, would block the law from taking effect.

Supporters of the lower youth minimum wage argue that it helps young people get a foot in the door — companies might be reluctant to hire teenagers if they have to pay them as much as a more experienced adult. But opponents of the law — and thus supporters of the referendum — argue there is no need to make exceptions to the minimum wage at a time when unemployment in South Dakota is under 3 percent.

We’re looking at a lot of ballot measures from all over the country today, and the process of reporting on this is reminding me of how frustrating I frequently find ballot measures when I’m presented with them as a voter.

Yesterday, I spoke with Eric Feaver, president of a Montana teacher’s union that’s opposing a ballot initiative in that state. He had a quote that summed up my feelings on the whole thing pretty well: “Direct democracy is just wonderful. I’m really glad to have the people engaged in it,” he said, “But you know in a ballot question environment, you just don’t get to ask all the questions and answer them. You don’t have an opportunity to amend the proposal. It’s just there. You vote it yes or no. Then you clean up the mess afterwards.” Yuuuuuup.

And research suggests that Americans, in general, have complicated feels about ballot initiatives. In the abstract, we support ballot initiatives by a wide margin — in a 2000 poll, for instance, more than 50 percent of respondents in every state said it was a good idea to have them. But when you get into specifics, we hedge. Surveys of California voters in 2006 found that more than 60 percent thought ballot initiatives were worded in confusing ways, more than 70 percent thought too much money was being spent by campaigns, and more than 65 percent wanted to see some kind of changes made to the process.

The final polls are in and the model has been updated for the last time! In the video below, I discuss our final forecast for the 2016 presidential election.

Every voter gets one vote, but not all votes count equally — and that leaves the average voter who is black, Latino or Asian with less power to swing the presidential election than the average white voter. BuzzFeed’s John Templon figured that out by combining Census Bureau voting and registration data (broken down by demographic groups) with our own Voter Power Index, a measure of how much power the average voter in each state yields. We determined a voter’s “power” by looking at the likelihood a state will tip the election and the number of electoral votes per voter in the state. Here’s a chart from Templon’s work:

One remarkable feature of this campaign has been that no news has been good news for the major presidential candidates. For both candidates, dominating the news cycle usually has led to weaker polling — a reflection of Clinton and Trump’s status as two of the most disliked presidential candidates on record.

For all the talk by Trump and his surrogates that the media is rigged against him — and complaints by others that there has been too much focus on Clinton’s private email server — media coverage over the last two months has been almost exactly a wash, according to monthly data provided by MediaQuant, a company that quantifies media mentions. Trump got $1.01 billion worth of coverage in September and October, compared with $940 million for Clinton. That’s consistent with an earlier report from the firm, which showed that once Trump was sharing the spotlight with just one rival instead of his many Republican primary challengers, he stopped getting the lion’s share. You can read more about the data and why to take it with a grain of salt — including that it mostly doesn’t reflect whether the coverage is positive or negative — in my August report.

It’s been a long election. Throughout the day, I’ll be highlighting some of the moments that made me say, “Whoa, can you believe that was this election?!”

Remember when Ted Cruz “formally” named Carly Fiorina as his Vice Presidential pick, then dropped out six days later?

Remember when Bill Clinton went on Attorney General Loretta Lynch’s plane for a 20-minute “social” conversation while her department was involved in an investigation of Hillary Clinton’s email server?

Remember when John Kasich visited Arthur Avenue in the Bronx and ate “two plates of spaghetti bolognese, a sandwich with mozzarella, pickles, salami, provolone, and hot peppers?”

If you have any nominations for “remember when,” get in touch by Twitter.

Early voting patterns don’t tell us who will win the election, but they certainly reveal that Americans in battleground states want to have their voices heard. We now have tallies from several that show record-breaking early voting turnout, particularly among Latinos in some states.

In Florida, 2,636,783 people voted by mail before Election Day, while 3,874,929 voted in person during early voting. That’s more than the total number of Floridians who voted in the 2000 election, period. The number of Latino early voters in the state doubled in number since the last election.

In Nevada, 41 percent of in-person early voters were registered Democrats and 35 percent were Republicans, which does not mean everyone voted for their party of registration, of course. FiveThirtyEight’s Harry Enten wrote about the possibility that heavy early voting may have swung the state for Clinton.

In North Carolina, the 3.1 million early absentee votes include 42 percent registered Democrats and 32 percent Republicans, which is a narrower advantage for Democrats than they had in 2012.

Election laws vary by state, of course, but early voting is proving increasingly popular, with a record 46 million Americans voting in advance of Election Day this year.

Politics junkies live for close elections, and Harry Enten is no exception. In the video below, he recaps election night 1960 — one of the biggest nail-biters of the television era.

We’ll be releasing more “Harry’s History” videos on this live blog throughout the day, so there’s plenty more to come!

Obamacare came up relatively little this election compared with 2012 … until the last two weeks of the campaign, that is. As the double-digit premium increases affecting health insurance plans in some states became public at the end of October, Trump has brought up the “catastrophe” of Obamacare with increasing fervor. Clinton has continued to make mostly understated mention of plans to build on the law.

Whatever the fate of Obama’s signature health care bill, whoever wins today will be forced to reckon with serious issues plaguing health insurance in the United States. Some are a result of the Affordable Care Act, but others began long before the law passed. Deductibles were on the rise for employer-sponsored insurance, for example, well before the ACA was signed.

Premium increases on the Obamacare marketplace aren’t necessarily the beginning of the end. They won’t mean much, at least not financially, for the majority of people — most get insurance through their employers or receive federal subsidies that will offset the increases. But the rising prices are a huge deal for the estimated 9 million to 13 million people who either aren’t eligible for subsidies or who buy private insurance off the marketplace. Then there’s the problem of major insurance companies dropping out of the market altogether: 17 percent of the people eligible to buy Obamacare insurance may have only one option for insurance in 2017, according to a New York Times analysis. As another large insurer threatens to leave the marketplaces after next year, it’s clear that no matter who wins today, some major changes are in the law’s near future.

Google is monitoring certain search terms to spot five categories of potential voting trouble: provisional ballots, long lines, inactive voter status, voting-machine problems and voter intimidation. Dots on a U.S. map show places where searches for terms related to any of those categories are spiking. The Electionland project that’s coordinating nationwide journalistic coverage of voting problems is tapping into the Google data for tips on where to investigate.

It’s a very cool experiment — with an emphasis on experiment. There’s no way to know if the map really did pick up problems — and didn’t pick up false positives — until voting is done.

We’ll hear the term “exit poll” on Election Day as often as we’ll be told “please don’t talk about politics” on Thanksgiving Day. So who’s in charge of ’em? Edison Research is the sole provider of the exit polls that you’ll be seeing cited throughout the night. They’re a firm contracted by the major television networks and outlets like the Associated Press to appear at polling places throughout the country today, and they’ve already been on the job for days, surveying early voters throughout the country. As of last week, Edison expected to survey 16,000 early and absentee voters (something they do by phone) and 85,000 voters in person.

Edison picks about 1,000 polling places throughout the country (they don’t tell reporters where, by the by, something yours truly knows from experience) and respondents are given written surveys to fill out. The operating theory is that filling out an exit poll survey should mimic a secret ballot. You can read more about the exit polling process here, in a piece from Pew Research.

One sure thing we’ll see throughout the day are reports of long lines. They can be one of the first indications of a problem at a polling station, and if we’re operating on the premise that voting should be easy for citizens in a democracy, lines are in and of themselves problematic.

A 2014 bipartisan commission determined that on Election Day in 2012, over 5 million voters had to wait over an hour to vote, while another 5 million waited between a half hour and an hour. During his victory speech in 2012, President Obama acknowledged the long wait times and said “we have to fix that.”

Florida, a perennial swing state, was among the states with the worst lines and wait times. A January 2013 analysis by the Orlando Sentinel found that around 201,000 voters likely “gave up in frustration.” Maryland, Virginia and South Carolina were also determined to have problematic wait times, along with a small number of complaints in Ohio.

So why do long lines happen?

They can be the result of unexpectedly high turnout, but in a 2014 report, the Brennan Center for Justice found that long lines were associated with poor resource allocation that disproportionately affected minority communities. In South Carolina, “the 10 precincts with the longest waits had, on average, more than twice the percentage of black registered voters (64 percent) than the statewide average (27 percent),” the study found. In Maryland, the 10 precincts with the lowest number of machines per voter had double the state average of Latino voters.

So what about this year? We’ll have to watch as the day unfolds, but during early voting, long lines in heavily Latino areas were reported in Florida, Texas and North Carolina.

My Election Day started off cute. On a perfect fall day in Brooklyn, a passer-by shouted, “I’m excited to vote. Yeah!!!!” Not only was she full of enthusiasm, she was also about six years old, walking with her father. Yes, parental voting is a gateway drug.

While many children today are getting a chance to watch their parents vote, some adults are watching each other for signs of trouble in the election. Yesterday, the Trump campaign sent out an email stating: “Just a reminder that if you encounter any problems casting a ballot or if you witness any disruptions at a polling location please call our Voter Assistance Hotline….”  There was also a web link. The Clinton campaign also has its own system for taking voter questions by text, email, or Facebook.

These are both in addition to nonpartisan organizations’ voter hotlines, and to the Department of Justice itself. Anyone with questions about voter intimidation or fraud can contact the DOJ Voting Rights Hotline at 800-253-3931 or email voting.section@usdoj.gov.

The talk of a rigged election and concerns about voter suppression make it likely those phone lines will get a lot of use.

On the list of nonpresidential matters up for vote today is a particularly unusual and complicated issue in the Florida Keys: Voters are being asked whether they are in favor of releasing genetically modified mosquitoes on the islands. The release would be part of an experiment by a private company that wants to send male mosquitoes into the wild whose offspring would die before adulthood. The hope is that the males will mate with females and produce the nonviable offspring, reducing the mosquito population and, by extension, the spread of mosquito-borne diseases such as dengue and Zika.

Many Keys residents oppose the research, especially on the small island where the mosquitoes would be released. They feel there are too many risks associated with the new technology, especially since it’s not clear whether the genetically modified mosquitoes would reduce disease, even if the number of mosquitoes goes down. People in favor of the research cite the large quantity of insecticide now being used to keep mosquitoes at bay and the growing problem of insecticide-resistant mosquitoes as some of the reasons that new technology is necessary. The vote is about a lot more than just this experiment: It raises questions about how a society should deal with risk, particularly when it comes to new technology and public health. I dug deep into the issue last month.

Keep your eyes on Latino turnout throughout the day. According to data by the Pew Research Center, Latino vote has been increasing vastly every presidential year — but fewer Latino registered voters say they’re sure they’ll vote this year in comparison to four years ago. Hispanic support of Hillary Clinton is strong, and while that might be good for Democrats, it’s too early to tell. We’ll be taking a closer look as the day progresses.

One possible cause of long lines at the polls today, should they happen: This is the first presidential election since a 2013 Supreme Court decision overturned a key portion of the Voting Rights Act of 1965. There are about 16 percent fewer polling places in counties affected by the decision, according to a report released last week by the Leadership Conference Education Fund, the education and research arm of a group that advocates on behalf of protections for minority voters.

Long lines, ballot confusion, voter intimidation, hoaxes: All of these could crop up today. More than 1,000 journalists and students nationwide are collaborating to monitor, confirm, write about and compile reports of voting problems. They’re being brought together by ProPublica, the nonprofit journalism organization, through its Electionland coalition. You can follow their reports; we will be doing so all day (and reporting on trends here). They’ve been up and running for a few weeks, with posts including a summary of the report on mass closures of polling places in counties affected by the Supreme Court decision.

The minimum wage has been a bit of a confusing issue in the presidential campaign this year. Clinton initially supported a $12-an-hour federal minimum, but she now pledges to “work to get to a $15 minimum wage over time.” Trump has been all over the map on the issue. But voters in four states don’t have to wait for the next president to work out his or her position — they will be voting directly on initiatives to raise their states’ minimum wages.

Supporters of the initiatives in Arizona, Colorado, Maine and Washington state hope to continue recent momentum for efforts to raise minimum wages at the state and local level. (All four states already require employers to pay more than the federal minimum of $7.25 an hour.) In 2014, voters in several states approved minimum-wage initiatives, often by large margins. Other states, including New York and California, have raised their minimums through the legislative process.

Raising the minimum wage is broadly popular, and polls suggest that all four initiatives are likely to pass. Here’s a quick look at today’s ballot initiatives. (Note: “Number of workers directly affected” is the number of people in each state who currently earn less than the minimum wage that would be in effect on Jan. 1, 2017, if the initiative passes. Calculations are based on Current Population Survey microdata for for the six months ending in September.)

Arizona

Current minimum wage: $8.05 an hour

Proposed minimum wage: $12 an hour by 2020

Number of workers directly affected: 405,000 workers (15 percent of total)

Other notes: The initiative would also guarantee paid sick time to workers. After 2020, the minimum wage would rise with the cost of living.

Colorado

Current minimum wage: $8.31 an hour

Proposed minimum wage: $12 an hour by 2020

Number of workers directly affected: 313,000 workers (13 percent of total)

Other notes: After 2020, the minimum wage would rise with the cost of living.

Maine

Current minimum wage: $7.50

Proposed minimum wage: $12 an hour by 2020

Number of workers directly affected: 57,000 workers (10 percent of total)

Other notes: The initiative would also gradually raise the minimum wage for workers who receive tips until it is equal to the overall minimum wage. In addition, it would index the minimum wage to the cost of living.

Washington

Current minimum wage: $9.47

Proposed minimum wage: $13.50 an hour by 2020

Number of workers directly affected: 545,000 (18 percent of total)

Other notes: Washington already has one of the highest state minimum wages in the country. The initiative would also guarantee paid sick time to workers.

Asking about the weather today isn’t just a way to avoid talking politics. People waiting in long lines outside polling places are more likely to keep waiting if the weather is dry. A 2007 study found that rain can suppress turnout by almost 1 percentage point per inch. Fortunately, wet weather looks like it won’t be a big factor today: There are just a few sprinkles of precipitation on the national maps. Rain was also light on Election Day in 2012 and 2008; more fell in 2004.

Welcome to the big show. For more than a year, we’ve lived through primaries and debates, polls and predictions, invective and revelations, but they’ve all led to today, when millions of Americans will cast hard-earned ballots that are the most particulate building blocks of our democracy. At FiveThirtyEight we are going to be here all day and through the night trying to make sense of those millions of ballots being cast.

During the daytime hours, we’ll keep an eye on election administration issues as they come up — long lines, voter intimidation and the like — peppering in some of our favorite journalism of the election cycle along the way. We’ll also recall some of this campaign’s most memorable moments, of which there were quite a few.

When the sun sets, things will really get going around here. We’ll parse and analyze exit poll results as they roll in, dive into state-by-state returns, and we’ll have a new forecast for you all to refresh — a live one! — that will launch at 6 p.m. The forecast will update each candidate’s chances of victory — in each state and overall — as winners are declared in each contest. We’ll have a presidential forecast and a Senate forecast. We’re here through the bitter end, the early morning hours, the last hurrah, so stay with us, too. Election results love company.

Tweet us a question or comment