Ann Selzer is Iowa’s most respected pollster, someone whose track record of accurately calling the Iowa caucuses was impeccable … until she, er, predicted an easy win for Donald Trump in the final Des Moines Register poll this year. She had Trump beating Ted Cruz 28/23; the actual result was Cruz 27.6, Trump 24.3. She missed, in other words, by a net of eight points. Her final forecast for the general election, published last night: Trump by … seven. What could go wrong?

I’m just teasing. There’s every reason to believe Trump will win this state, which went twice for Obama, even if the margin isn’t quite the landslide Selzer is seeing. He hasn’t trailed in the RCP average of Iowa since September 1st and he led by seven points or better in more than one poll in mid-September. If Iowa, with its huge white working-class population, can’t deliver for Trump than we won’t need to worry about electoral maps or battleground states. It’ll be a national wipeout for Clinton.

Trump is the top choice for 46 percent of Iowans who have already cast a ballot or plan to do so on Election Day, according to the latest Des Moines Register/Mediacom Iowa Poll, compared to 39 percent who say they’re for Clinton…

Forty-one percent of independents support Trump, compared to 34 percent who back Clinton. Trump likewise leads by 5 percentage points — 43 to 38 — among voters under 35…

Among other groups, meanwhile, Clinton’s lead is significantly lower than the advantage she enjoys nationally. She holds just a 3-point lead among women, for example, a demographic Democrats have been banking on to reject Trump and lift Clinton and one that she leads by a dozen or more points in national polls.

How is Trump succeeding in a state where McCain and Romney failed? Partly it’s the college-educated/working-class divide among whites you’ve heard so much about nationally, but in Iowa it’s also due to younger voters tilting red. Iowans under age 35 split 33/26 Democratic in 2008 but break 32/25 Republican today. Selzer’s read on that is that the Bernie Sanders contingent of young liberals isn’t turning out for Clinton this time, a surprising result given that Clinton had been doing better nationally with younger voters in October. (One possibility: The fact that Iowa has seemed solid for Trump for the past few months might have convinced some progressives who don’t like Clinton to begin with not to bother this year.) That helps explain why the margin in this poll more so than the result had data nerds buzzing last night when it came out. Does a seven-point lead for Trump tell us something about Iowa specifically, or does it tell us to expect stronger working-class white support for Trump across the midwest? Hillary doesn’t care about losing Iowa but she cares a lot about Wisconsin and Michigan.

Here’s another bad, bad data point for Clinton:

YouGov’s model is just that, a model rather an actual poll, but there’s data elsewhere that this is really happening. Fox News’s final survey of the race, which had Clinton up just two points on Trump, found Trump ahead 45/42 among whites with a college degree. That group has voted Republican for decades but had tilted towards Clinton in most of this year’s polls; now, with Republicans coming home to Trump in the final weeks of the election, they may be tilting back towards him again. Trump is still weaker among that group than other GOP nominees have been (Romney won them by 14 points), but he’s also stronger than Romney was among working-class whites, winning them by 33 points in Fox’s data. If whites without a degree turn out in higher numbers this year and white college grads end up leaning Republican after all and the apparent surge in Latino turnout in early voting for Hillary doesn’t carry through to Election Day, he’s got a shot after all. But then, not every poll sees college-educated whites abandoning Clinton now: She still leads them by 10 points in NBC/WSJ’s final pre-election poll.

Speaking of sweeping the midwest, though, don’t get too excited just yet. A new poll of Ohio from the Columbus Dispatch, which nailed the outcome there in 2012, is a shocker given Trump’s durable lead in the state this year: Clinton 48, Trump 47.

The poll results indicate that if the Clinton campaign can get young and minority voters to the polls, she almost certainly will win. And a victory in Ohio would almost certainly end Trump’s hopes for the presidency…

As expected, Clinton is winning among those who said they already have voted, holding a 10 point advantage. Four years ago, President Barack Obama’s lead in early voting was key, because he lost to Mitt Romney among votes cast on Election Day in the state…

Perhaps because of Trump’s appeal to working-class voters, Clinton is winning traditionally Democratic union households by only 5 points.

Apart from a week after the “Access Hollywood” tape dropped, Trump had led steadily in RCP’s average of Ohio since mid-September. Clinton hadn’t led a single poll of the state since mid-October — until now. And the Dispatch poll isn’t the only one suddenly showing the state very tight. A CBS/YouGov poll out today has it 46/45 for Trump. Some politicos were surprised to see that Cleveland was the site of her event with Beyonce and Jay-Z a few days ago, as Ohio seemed like one of the least likely battlegrounds to stay in the Democratic column this year. She’s all-in, though: Not only did she choose Ohio for that rally, she’s campaigning there with LeBron James today. Team Clinton’s obviously seeing something in its private polling that makes them think Ohio’s still winnable. And if that’s true, we end up with a suspicion counter to the one borne of the Iowa poll above: How well could Trump be doing in Wisconsin and especially Michigan if he still hasn’t put Ohio away?

Let’s look at one more poll, this time a national one. The ABC/WaPo tracking poll has been all over the map the last few weeks, at one point putting Clinton ahead by 12 points, then detecting a Trump surge over several days that finally left him ahead by a point. Since then, though, they’ve seen the race return to the status quo that governed for most of the year, with Clinton gradually gaining over the last several days until she leads now 48/43, a margin that actually exceeds Obama’s margin in 2012. Why should you care about that when there are all sorts of national polls flying around, from a 44/40 Clinton lead in NBC/WSJ to a 44/43 lead in McClatchy/Marist to a 44/43 lead for Trump in IBD/TIPP to the freaky deaky LA Times poll that has Trump winning in a landslide by nearly six points? This is why:

The ABC/WaPo tracking poll’s final estimate is, quite simply, crazy good at predicting the actual national results, and it’s been crazy good for a long time. (It’s one reason why they earn an A+ pollster rating from FiveThirtyEight.) The only time they’ve missed in a meaningful way was in 2000, when they underestimated Al Gore’s support, but that might be explained by the last-minute DWI oppo hit on Bush that year not being captured fully in the data. This year’s final estimate will be out tomorrow, I believe, but with the race currently at 48/43, you wouldn’t expect a massive overnight shift to Trump to move that number a lot. Clinton will almost certainly be their favorite to win when polls open on Tuesday morning. Anyone can be wrong — ask Ann Selzer — but that’s what Trump will be up against 48 hours from now, even with his strong showing in Iowa.

One last note: Per Nate Silver, fully 13 percent of the electorate is still either undecided or third-party right now versus just three percent who met that description at this point in 2012. That seems counterintuitive when you have two larger-than-life characters in Trump and Clinton, but it’s pretty intuitive when you remember that each of them is about as popular as cancer. Which form of cancer should America volunteer for? Tough question! Still some hard thinking to do for lots of voters out there.