Nate Silver's Blog, page 124

Election Update: Democrats Should Panic … If The Polls Still Look Like This In A Week

Want these election updates emailed to you right when they’re published? Sign up here .

Hillary Clinton’s lead in the polls has been declining for several weeks, and now we’re at the point where it’s not much of a lead at all. National polls show Clinton only 1 or 2 percentage points ahead of Donald Trump, on average. And the state polling situation isn’t really any better for her. On Thursday alone, polls were released showing Clinton behind in Ohio, Iowa and Colorado — and with narrow, 3-point leads in Michigan and Virginia, two states once thought to be relatively safe for her.

It’s also become clearer that Clinton’s “bad weekend” — which included describing half of Trump supporters as a “basket of deplorables” on Friday, and a health scare (followed by news that she had been diagnosed with pneumonia) on Sunday — has affected the polls. Prior to the weekend, Clinton’s decline had appeared to be leveling off, with the race settling into a Clinton lead of 3 or 4 percentage points. But over the past seven days, Clinton’s win probability has declined from 70 percent to 60 percent in our polls-only forecast and by a similar amount, from 68 percent to 59 percent, in our polls-plus forecast.

That’s not to imply the events of the weekend were necessarily catastrophic for Clinton: In the grand scheme of things, they might not matter all that much (although polling from YouGov suggests that Clinton’s health is in fact a concern to voters). But when you’re only ahead by 3 or 4 points, and when some sequence of events causes you to lose another 1 or 2 points, the Electoral College probabilities can shift pretty rapidly. A lot of light blue states on our map have turned pink, meaning that Trump is now a narrow favorite there instead of Clinton:

When a candidate has a rough stretch like this in the polls, you’ll sometimes see his or her supporters pass through the various stages of grief before accepting the results, beginning with a heavy dose of “unskewing” or cherry-picking of various polls. In this case, however, the shift in the race is apparent in a large number of high-quality surveys, and doesn’t depend much on the methodology one chooses. FiveThirtyEight, Real Clear Politics and Huffington Post Pollster all show similar results in their national polling averages, for example, with Clinton leading by only 1 to 3 percentage points over Trump.

This potentially ignores a more important question, however. Sure, Clinton might lead by only a percentage point or two right now — with a similarly perilous advantage in the Electoral College. But is that necessarily the best prediction for how things will turn out in November?

Our various models differ on this question. Polls-only assumes that there’s still a lot of uncertainty about the outcome. But it also mostly assumes that the current condition of the race — Clinton ahead by around 2 points — is a statistically unbiased prediction of the Nov. 8 outcome. In other words, it assumes that Clinton is as likely to continue losing ground as opposed to regaining ground from this point forward.

Polls-plus, by contrast, discounts short-term shifts in the polls by hedging them with an index based on economic conditions. Even at this fairly late stage of the campaign, about 30 percent of the polls-plus forecast is based on the economic index rather than the polls. But this year, the economic index assumes the election will be very close. Thus, the recent tightening of the race is not much of a surprise to polls-plus, and it doesn’t differ much from the polls-only forecast at this point.

Another way to hedge against short-term swings in the polls would be to assume that polls tend to revert to where they’d been previously. Over the course of the campaign, Clinton has been ahead by an average of about 5 percentage points, although she has ranged between being virtually tied with Trump at her worst moments to being ahead of him by as much as 10 points at her best moments. In fact, the election has tended to ebb and flow between these boundaries like a sine wave:

If we had a model like this — basically, a version of polls-plus that partially reverted the current polls to the long-term mean of Clinton +5 — it would show her as about a 70 percent favorite instead of a 60 percent favorite. But would that be a good set of assumptions? Does the fact that Clinton has usually led Trump by a larger margin than she leads him now mean anything?

This is a complicated question, and one that we might want to revisit over the next couple of weeks. But the short answer is… I don’t know. We know that many news events — most notably, the political conventions — produce short-term “bounces” in the polls that partly or wholly reverse themselves after a few weeks. There were also some examples of this in 2012. Mitt Romney’s position improved by several percentage points following his first debate in Denver against President Obama, but his gains soon proved fleeting. Media coverage of the campaign — which tends to rally behind whichever candidate is gaining in the polls until it tires of the story and switches to scrutinizing the frontrunner — could also contribute to such swings back and forth.

So it’s plausible that Clinton’s “bad weekend” could be one of those events that has a relatively short-lived impact on the campaign. As if to put to the question to the test, Trump upended the news cycle on Friday by relitigating the conspiracy theory that Obama wasn’t born in the United States. (Trump finally acknowledged that Obama was born here, but only after falsely accusing Clinton of having started the “birther” rumors.) If voters were reacting to the halo of negative coverage surrounding Clinton rather than to the substance of reporting about Clinton’s health or her “deplorables” comments, she could regain ground as Trump endures a few tough news cycles of his own. Over the course of the general election so far, whichever candidate has been the dominant subject in the news has tended to lose ground in the polls, according to an analysis by Larry Sabato, Kyle Kondik and Geoffrey Skelley.

All of this is tricky, though, because we still don’t have a great sense for where the long-term equilibrium of the race is, or even whether there’s an equilibrium at all — and we probably never will because of the unusual nature of Trump’s candidacy. Perhaps Trump isn’t that different from a “generic Republican” after all. Or perhaps (more plausibly in my view) he is very poor candidate who costs the Republicans substantially, but that Clinton is nearly as bad a candidate and mostly offsets this effect. Still, I’d advise waiting a week or so to see whether Clinton’s current dip in the polls sticks as the news moves on from her “bad weekend” to other subjects.

Election Update: Has Clinton’s ‘Bad Weekend’ Moved The Polls?

Want these election updates emailed to you right when they’re published? Sign up here .

Donald Trump has a 33 percent chance of winning the election according to our polls-only forecast and a 34 percent chance according to polls-plus. These roughly 1 in 3 odds are close to Trump’s highs since the party conventions.

Trump has had a reasonably strong couple of days in the polls, and the odds according to our forecast have resumed moving slowly but steadily toward him after having flattened out toward the end of last week. As is often the case, however, it’s hard to attribute causality. Hillary Clinton has had a series of negative news cycles — first after her Friday night remark that half of Trump supporters fall into a “basket of deplorables” and then after she abruptly left a Sept. 11 memorial event on Sunday and a video captured her appearing to stumble into her vehicle. It was later revealed that Clinton had been diagnosed with pneumonia.

If you want to make the case that the weekend’s news has moved the polls, there are a couple of them that you might cite prominently. First is the latest edition of the USC Dornsife/Los Angeles Times tracking poll; it shows a fairly dramatic swing, with the race going from a 1-percentage-point lead for Clinton to a 5-point lead for Trump over the past few days. Although the LA Times poll has had a strong pro-Trump house effect, the trend line in the poll can be useful because the poll re-interviews the same respondents over and over instead of finding a new sample of voters.

The other scary result for Clinton is a Selzer & Co. poll of Ohio, conducted Friday through Monday on behalf of Bloomberg Politics; it shows a 5-point lead for Trump there. Selzer hadn’t previously polled Ohio during this campaign, so the poll doesn’t have a trend line. But this is an unambiguously bad result for Clinton, coming from one of the highest-rated pollsters in one of the most important states.

But some other recent polls that include interviewing conducted over the weekend aren’t as bad for Clinton. The Ipsos-Reuters tracking poll, based on interviews through Monday, shows the race having moved to and fro but with no clear trend toward either candidate over the past week. Gallup’s tracking poll of candidate favorability ratings — not used in our model because it doesn’t contain a head-to-head result, but useful for context — shows Trump’s favorability rating having improved slightly over the past week but Clinton’s steady instead of declining. The Google Consumer Surveys national tracking poll showed little change, and YouGov’s national poll — the most recent of the bunch, having been conducted Saturday through Tuesday — had both Clinton and Trump gaining ground at the expense of third-party candidates.

My best guess on the effect of the weekend’s news, based on what the model shows so far, is that the race is continuing to trend moderately toward Trump, when the momentum toward him might have stalled out if not for the events of the weekend. But we can’t rule out a more acute shift toward Trump or that the “Hillary’s bad weekend” meme is a false alarm — there isn’t quite enough data yet.

Whether or not the race will continue to tighten is a guessing game, in other words. But my impression is that the commentariat has been slow to recognize how much the race has tightened already. It’s never a good idea to freak out over any one poll. But the trend toward Trump has been clear for a few weeks now, and it’s been just as clear in state polls as national polls. Yes, the data is noisy. Polls are all over the place in Ohio, for instance. But over the course of all of this, Trump has whittled down an 8-point lead for Clinton into about a 3-point lead instead — about a 5-point swing. With there having been several shifts of that magnitude since the primaries ended, with there being a large number of undecided voters, and with the debates still ahead, neither Clinton nor Trump should feel all that secure.

Putting Hillary Clinton’s Basket Of Deplorables In Context

 Subscribe: iTunes |ESPN App |Download |RSS |New to podcasts?

There are no breaks in the big news this election cycle. This week, our elections podcast tries to add some context and clarity to two big and complicated stories that broke over the weekend. First we discuss Hillary Clinton’s description of “half” of Donald Trump supporters as being in a “basket of deplorables.” Then, we talk about the coverage of Clinton’s health. Clinton felt ill at an event on Sunday morning and had to leave it early; later in the day, her campaign released a statement saying she’d been diagnosed with pneumonia on Friday. The incident played into long-running (and unfounded) conspiracy theories about her health, as well as the debate about how much candidates should reveal when it comes to their health. Also this week, we chat with Sasha Issenberg, a contributor to Bloomberg Politics, about a new project to provide real-time results data on Election Day.

Note: We have three live shows coming up this fall, two in New York and one in Chicago. Tickets are going fast. Get them here.

We’re experimenting with adding a transcript of a portion of the podcast here each week. Here’s some of our conversation about Clinton’s comment: “To just be grossly generalistic, you can put half of Trump supporters into what I call the basket of deplorables, right? Racist, sexist, homophobic, xenophobic, Islamophobic, you name it.” The transcript begins at the 7:15-minute mark and has been lightly edited for clarity.

Jody Avirgan: Harry, Clinton said half of Trump supporters fit into this basket of deplorables. What did you think when you saw that as a stat?

Harry Enten: Look, what we know is that Trump supporters are more likely — even than other Republicans who supported other candidates during the primary — to have less warm feelings towards African-Americans, Hispanics, Muslims, those who are transgender, gays and lesbians, feminists. So, we know that this is the case. This is not some shocking thing. I think where you might get into trouble is where you assign the 50 percent to it. Is it 20 percent? Is it 10 percent? Is it 30 percent? I think if we’re being honest with ourselves, it’s a significant portion of it.

Jody: Just to jump in here with some of the actual polling around this issue that shows the level of racism in Trump supporters, The Economist and YouGov did a poll of primary voters on racial resentment, measuring things like support for the idea that blacks are undeserving or want special assistance from government, and they found that 59 percent of Trump supporters in the Republican primary scored in the top quartile of racial resentment. A lot of people are sharing this Reuters poll to back up Clinton’s statements — Reuters asked voters to rate blacks and whites on character traits, and about 40 percent of Trump supporters placed whites higher on the “hardworking” scale than blacks.

Harry: And what percentage of Clinton supporters?

Jody: Well, that was the other part. Twenty-five percent of Clinton supporters said the same. So, since we are going to take her analysis and evaluate it on face value, she is really citing the fact that there’s real racism in this country and we’ve seen it kind of enabled by the Trump campaign, right, Nate?

Nate Silver: Trump right now is getting about 40 percent of or likely voters, so half of that is 20 percent of the country. … Can I believe that 20 percent of the electorate consists of white people — not that there can’t be other types of racism — but consists of white people that are more racist than not? Sure. That’s an imprecise estimate, but sure, it’s not ridiculous. Beyond that, I think you get into false precision, though. A couple of years ago, my colleague Allison McCann and I looked at the general social survey which asks about racial attitudes in a lot of different ways, and of the white population, you can get anywhere from five percent to 50 percent, depending on how you ask the question about anti-black racism. And there’s a question about how they split between Clinton supporters and Trump supporters, but if someone submitted that in an article at FiveThirtyEight, we’d say “needs citation.” And even if there was truth, we would probably end up writing the article to say there’s no one precise number. But it’s not ridiculous to say that 20 percent of the country is racist and that most of those people are voting for Trump.

Harry: Right. Depending on who you include in this basket, you’ll get different numbers. So, if you ask Americans, as Gallup did, the willingness to vote for a president of various backgrounds, you would see that 90 percent of Republicans said that they’d be willing to vote for a black president, yet only 45 percent of Republicans said that they’d be willing to vote for a Muslim president. So, this is like one of those things where it really kind of depends on your definition. The more groups of people you put into that basket of saying who you wouldn’t be willing to vote for, the more likely it is that you could get up to that 20 percent of the country or that half of Trump supporters. I don’t think it’s a ridiculous number when you look at all of these things, but to put a precise measurement on it is where you get into trouble.

Jody: And, Clare, this kind of made me think of the piece that you wrote about the state of the GOP or the end of the GOP, and how the GOP is grappling with this, because that was in many ways Clinton’s point. We see within the GOP and we see within this candidate that they are giving voice, and having to deal with as a party, a large element of racists in their midst.

Clare Malone: As the country gets more diverse ethnically, younger, and more educated — more people are going to college these days — the Republican Party has gotten whiter, a lot older, and a lot less educated than everyone else. What I say when I read things and I can’t quite wrap my head around it, sometimes I say it’s just my 1987 brain, where I can intellectualize it, but there’s something about me that just culturally doesn’t get it — and I think what the GOP is dealing with here is a lot of people who have 1940s brain, or there’s just this thing where they can’t quite move along and catch the drift of cultural things, and that surfaces in sexism and racism. … And the demographics, unfortunately, lend the GOP to have those people make up a bigger proportion of the party than the Democrats. But there are sexist and racist Democrats, too.

Jody: Let’s transition a little bit to evaluating Clinton’s statements as rhetoric and as campaign posture. So, Nate — was this a good choice of words on her part?

Nate: First of all, let me just say, I kind of resent the whole, “Well, we’re savvy journalists and we know maybe technically the statement is right, but how will it play with the public?” To me, that’s quite disdainful to the audience, and I’ll tell you my personal experience this year covering the campaign. I was one of those people who said, “Oh, sure, racism exists in America, but people are too quick to chalk up problems to racism,” right, and I think that’s one of the things that contributed to me being slow to recognize the Trump phenomenon and the breadth that it had within the Republican party. And of course, a lot of people are a little racist, some people are a lot racist — it’s a slippery kind of distinction at some point. I don’t know why it’s so un-PC to point out that, hey, you saw when Trump had the biggest rises in the Republican party primaries in the polls. It was when he was baiting people about Mexicans and Muslims and after terrorist incidents and whatever else. Is it smart politically? I guess I’d say the Clinton people seem to think it’s at least OK. You can’t call this a gaffe. They’ve been using this line before, and instead of totally backing down from it, they sort of doubled down and said, well, maybe I shouldn’t have said half, but the basic sentiment is right.

Harry: Fact is, most voters think that Donald Trump can be described as a racist or as a bigot. YouGov polled last week — 54 percent said that yes, they would use the word “racist” to describe Donald Trump. 52 percent said, yes, “bigoted” is a word that they would use to describe Donald Trump. This is not happening in a vacuum, folks. And if you look at the ad Donald Trump put out trying to capitalize on this, he’s talking to his base, he’s not talking to other voters. Fact is, at this point, if you’ve said you’re going to vote for Donald Trump, the polls indicate 90 percent of those voters at least are not going to be voters for Hillary Clinton. The question as to whether or not that this a gaffe or something that would hurt Hillary Clinton really comes down to whether or not you believe that those voters who are persuadable one way or another were actually affected by this in a negative way for Hillary Clinton, and I haven’t seen a single piece of evidence to suggest that is so.

Jody: I just want to add one thing that what Harry was saying made me think of: If she’s going to make this argument — and I think we want our leaders to call out racism when they see it — I wish she had not done it in this pseudo-polling kind of language of “half,” which sounds not very scientific. It sounds sort of rhetorical. She should have cited the polls that Harry just cited. Stand up there and say, “Look at this polling. This is scary. This is happening in our country.” And be precise, and be directed, and talk about it. But instead it comes off as dismissive about large swaths of voters as opposed to [addressing] a phenomenon in this country that we really need to reckon with in the way that we’re discussing.

Clare: Yeah, she was making a real statement, and I would encourage people to read Ta-Nehisi Coates and Jamelle Bouie — he’s written and talked about that. I think what’s bothering everyone, even if what she said was true, is that people are interpreting it as her saying it in a condescending way, and if you’re a Republican and you’re sort of like, “OK, I was supporting Cruz or Kasich, but maybe I’m going to vote for Trump,” I think the pushback you’re getting from a lot of people is: No one really wants to admit that they’re in a political coalition with racists even if they know it. This conversation made me think of a quote from a very thoughtful, conservative guy I talked to for our piece, “The End of the Republican Party,” and how he was sort of a Tea Party activist. He was talking with me about the atmosphere that he saw starting to stir up in 2010 and 2014, and he said to me, this isn’t the most artful way to say it, but it’s like — where do you go when the only people who seem to agree with you on taxes hate black people? And he basically said, I think I made the wrong choice. I didn’t call people out enough on being racist, but I agree with them on half this stuff. And so, I think for a lot of Republicans, it is an uncomfortable reckoning of saying, yeah, some of these people hold abhorrent views, but it’s kind of a rough thing to hear from the outside, even if it is true.

You can listen to the episode by clicking the “play” button above or by downloading it in iTunes, the ESPN app or your favorite podcast platform. If you are new to podcasts, learn how to listen.

The FiveThirtyEight Elections podcast publishes Monday evenings, with occasional special episodes throughout the week. Help new listeners discover the show by leaving us a rating and review on iTunes. Have a comment, question or suggestion for “good polling vs. bad polling"? Get in touch by email, on Twitter or in the comments.

Putting Hillary’s Basket Of Deplorables In Context

 Subscribe: iTunes |ESPN App |Download |RSS |New to podcasts?

There are no breaks in the big news this election cycle. This week, our elections podcast tries to add some context and clarity to two big and complicated stories that broke over the weekend. First we discuss Hillary Clinton’s description of “half” of Donald Trump supporters as being in a “basket of deplorables.” Then, we talk about the coverage of Clinton’s health. Clinton felt ill at an event on Sunday morning and had to leave it early; later in the day, her campaign released a statement saying she’d been diagnosed with pneumonia on Friday. The incident played into long-running (and unfounded) conspiracy theories about her health, as well as the debate about how much candidates should reveal when it comes to their health. Also this week, we chat with Sasha Issenberg, a contributor to Bloomberg Politics, about a new project to provide real-time results data on Election Day.

Note: We have three live shows coming up this fall, two in New York and one in Chicago. Tickets are going fast. Get them here.

And, we’re experimenting with adding a transcript of a portion of the podcast here each week. Check back on this page soon.

You can listen to the episode by clicking the “play” button above or by downloading it in iTunes, the ESPN app or your favorite podcast platform. If you are new to podcasts, learn how to listen.

The FiveThirtyEight Elections podcast publishes Monday evenings, with occasional special episodes throughout the week. Help new listeners discover the show by leaving us a rating and review on iTunes. Have a comment, question or suggestion for “good polling vs. bad polling"? Get in touch by email, on Twitter or in the comments.

What Makes A Tipping-Point State

 Subscribe: iTunes |ESPN App |Download |RSS |New to podcasts?

In our latest forecast update podcast, Nate Silver and politics editor Micah Cohen talk about what makes a state more or less likely to be the tipping point in deciding which presidential candidate wins the Electoral College. Nate also fields listener questions about how the model works.

Tickets are now on sale for the elections podcast live show in Chicago on Oct. 7.

You can listen to the episode by clicking the “play” button above or by downloading it in iTunes, the ESPN app or your favorite podcast platform. If you are new to podcasts, learn how to listen.

The FiveThirtyEight Elections podcast publishes Monday evenings, with occasional special episodes throughout the week. Help new listeners discover the show by leaving us a rating and review on iTunes. Have a comment, question or suggestion for “good polling vs. bad polling"? Get in touch by email, on Twitter or in the comments.

Is A 50-State Poll As Good As 50 State Polls?

It sounds like a riddle of sorts: Is one giant poll of all 50 states the same thing as 50 small polls, one for each state, added together?

If this seems like an odd question, it’s because it hadn’t really come up before this year. Sure, technically speaking, any national poll is composed of interviews from all 50 states. For instance, we’d expect a 1,000-person national poll to include about 100 respondents from California, 30 from Virginia, and 5 from Idaho, assuming that the number of people interviewed in each state was roughly proportional to turnout in 2012. But pollsters almost never report those state-by-state breakouts in the same way they do other sorts of demographic splits. That’s probably for good reason: The margins of error on those subsamples would be astronomical for all but the most populous states.

But what if instead of using a sample size of 1,000, your poll interviewed 50,000 people? Now you’d have around 5,000 respondents from California and 1,500 from Virginia — more than enough to go around. Even your Idaho sample size — about 250 people — is semi-respectable.

Several online pollsters are now doing this, interviewing tens of thousand of people nationally per week or over the course of several weeks, as part of their national polling. And they’re increasingly reporting their results on a state-by-state basis. SurveyMonkey, Ipsos and Morning Consult have all released 50-state surveys, projecting the outcome in each state along with the overall Electoral College result. Google Consumer Surveys, which interviews around 20,000 people per week, has a crosstab showing their state-by-state results.

FiveThirtyEight has been using the state-by-state results from SurveyMonkey and Ipsos in its forecasts, and we’re in the midst of incorporating the data from Morning Consult and Google. (This has already attracted a fair amount of attention; Donald Trump’s campaign erroneously attributed Ipsos polls of Ohio and Iowa to FiveThirtyEight.) Which brings me back to my earlier question: Is a 500-person subsample of Colorado voters from a 20,000-person national poll the same thing as a 500-person poll that was dedicated to Colorado, specifically?

After thinking and researching my way through the problem, my answer is that these polls aren’t quite the same. The Colorado-specific poll is likely to provide a more reliable estimate of what’s going on in that particular state. And it deserves a higher weight in our model as a result.

One reason to give the 50-state technique a lower weight is that hasn’t really been empirically tested. There have been cases in the past where pollsters commissioned simultaneous polls of all 50 states — surveying 600 voters in each state, for example — but for reasons I’ll explain in a moment, that’s potentially different from commissioning a huge national poll and reporting the results of state-by-state subsamples.

One potential source of error has to do with demographic weighting. Polls of all kinds engage in extensive demographic weighting because people aren’t equally likely to respond to polls Typically, for instance, white voters are more likely to respond to telephone polls than black voters. Pollsters attempt to counteract this by giving extra weight to the black voters they reach until the demographics of their poll matches that of Census data or other reliable sources.

But establishing these weights is not easy because voters are not monolithic within these demographic groups. White voters in Oregon are much more likely to vote Democratic than white voters in Mississippi, for instance. If you’re taking a poll just of Oregon or Mississippi, you’ll optimize your demographic weights to match the makeup of those states specifically. But if you’re conducting a national poll that includes interviews from Oregon and Mississippi along with the other 48 states, you might not pay as much attention to how the results shake out in individual states. Perhaps you’ll overestimate the Democratic vote in Mississippi, where whites are especially conservative, and underestimate it in Oregon — but those differences will likely cancel out in the national result.

Another potential problem is misidentifying the state a poll respondent votes in. With online polls, the problem is that IP addresses aren’t 100 percent reliable — for instance, a website would think I’m in Connecticut right now because that’s where ESPN’s internet connection is based, even though I’m writing from the FiveThirtyEight office in New York City. Someone filling out an online survey at their office in Washington, D.C., might actually live in Virginia or Maryland. With telephone polls, the issue is that people carry their mobile phone numbers around when they move from state to state, making it harder to identify a voter’s residence based on her phone number alone.

Pollsters spend a lot of time thinking about problems like these when they’re conducting surveys of a particular state, and they can employ some good workarounds (for instance, asking the voter where they’re registered to vote). But in a national poll, the pollster doesn’t need to be as precise. If you misidentify me as a Connecticut voter when I’m really registered in New York, that won’t affect the topline margin in the national poll, even though it could skew the Connecticut and New York results.

We’ve noticed, anecdotally, that the 50-state polls sometimes produce weird results in cases like these, in states that are either demographically idiosyncratic (such as Mississippi) or in small states (such as New Hampshire) where the sample is potentially contaminated by voters from another state. But this is FiveThirtyEight, and we’re not big on anecdotes. So we looked at the best predecessor for the 50-state polls that we could find: results from the 2012 Cooperative Congressional Election Study (CCES), a project conducted jointly by the online pollster YouGov and a consortium of universities. In 2012, the study surveyed around 50,000 voters, asking them about their presidential vote along with a long battery of demographic and political questions.

This is an incredibly useful dataset that we use all the time at FiveThirtyEight. But what if you use the CCES to estimate the presidential vote in each state? To be clear, this is not a use the authors of the CCES necessarily intended or would recommend. But it’s the closest approximation I could think of for what Ipsos or SurveyMonkey are doing with their 50-state surveys.

It turns out that the CCES didn’t produce very reliable estimates of the state-by-state results, even when applying the demographic weights recommended by the survey. For example, it had Barack Obama beating Mitt Romney by 10 percentage points in Florida (Obama actually won by just 1 point) and narrowly winning Georgia (he lost by 8 points), while it had Romney easily beating Obama in New Hampshire (Obama won there by nearly 6 points). Overall, the poll missed the final result in each state by an average of 7.3 percentage points, a much higher margin than state-specific election polls. That includes 3.1 percentage points of what we call pollster-induced error, meaning that which can’t be explained by sampling error or the difference in timing between the poll and the election.

As a result of this analysis, we’ll continue to use the state-by-state breakouts from Ipsos and other pollsters, but we will significantly lower the weight assigned to them in our polling averages. These polls will continue to have some influence around the margin — we don’t want to discard their data entirely — but state-specific polls will have more influence on the forecast. Note that this penalty won’t apply when a pollster such as Ipsos decides to survey one state specifically, only to state subsamples from their national polls.

The overall result of this change is minimal for now. It very slightly lowers Hillary Clinton’s projected popular vote margin over Trump (by about two-tenths of a percentage point) and very slightly increases her projected chances of winning the electoral college (by about 1 percentage point). Even if this change in our algorithm hasn’t had a major effect so far, it’s intended mostly as a preventive measure against allowing our polling averages to be flooded by subsamples from the 50-state polls. This is useful data that we’re pleased to have at our disposal, but not quite the same thing as getting 50 separate surveys from every state.

Election Update: The Swing States Are Tightening, Too

Want these election updates emailed to you right when they’re published? Sign up here.

In the more poll-obsessed corners of the internet, we’ve been arguing about Hillary Clinton’s decline in the polls against Donald Trump. Everyone seems to agree that Clinton’s lead is down quite a bit in national polls, to an average of around 3 percentage points from a peak of about 8 points shortly after the Democratic convention. But there’s a debate about how this translates to the state level.

My position is that a decline in Clinton’s national polls necessarily means that she’s declined in the states. There’s just no way around this; as we learned on Schoolhouse Rock, the United States is composed of 50 states and the District of Columbia. Perhaps it’s possible Clinton’s declined more in noncompetitive states than competitive ones — for instance, if Trump’s gains have mostly come from Republicans, widening his margins in red states but less in purple states. But that sort of conclusion is usually wishful thinking.

Still, there’s no better way to prove or disprove this than to look at polls of swing states directly. We’ve gotten a lot more of those polls recently, and they show pretty much just what the national polls do: Clinton’s lead in the states most likely to tip the balance of the election is somewhere around 3 percentage points.

Take the set of polls that Quinnipiac University released on Thursday afternoon. In the versions of the polls that include third-party candidates, Clinton led Trump by 5 percentage points in Pennsylvania and 4 points in North Carolina, but was tied with him in Florida and trailed him by 4 points in Ohio. The North Carolina result is slightly better for Clinton than our model expected, and the Ohio result was slightly worse — but overall, these results are consistent with the model’s hypothesis of a 3- or 4-percentage point national lead for Clinton.

Quinnipiac’s aren’t the only polls out there, of course. The table below contains a simple average of recent polls in what we call “states to watch” — the set of 14 states that are most important to the national outcome. The average includes all polls in each state with a median field date of Aug. 21 or later — that is, polls conducted over the past two or three weeks. In some states, such as Nevada, the only polls that qualify by this definition are the 50-state online polls from Ipsos and SurveyMonkey, but in most others, such as Pennsylvania and North Carolina, there’s quite a lot of data to work with.

STATE (NO. OF POLLS)SIMPLE AVERAGEMinnesota (2)Clinton +8.0Virginia (4)Clinton +5.8Nevada (2)Clinton +5.5Pennsylvania (8)Clinton +5.5New Hampshire (5)Clinton +5.3Colorado (3)Clinton +3.7Wisconsin (5)Clinton +3.3Michigan (4)Clinton +3.0Florida (7)Clinton +1.6North Carolina (9)Clinton +1.2Ohio (5)Trump +0.4Arizona (7)Trump +2.4Iowa (4)Trump +2.5Georgia (2)Trump +3.5Recent swing state polls show Clinton narrowly ahead

Based on polls with a median field date of Aug. 21 or later. Michigan would be the tipping-point state if recent polls are right.

I’ve highlighted Michigan in the table because it would be the tipping-point state if the recent polls are right — that is, the state that would get Clinton to 270 electoral votes if she wins it along with all the states above it. She leads in Michigan by 3 percentage points in the simple average of recent polls, almost exactly matching her lead in national polls. That’s further confirmation that national polls and state polls tell pretty much the same story.

It’s interesting that Michigan shapes up as the tipping-point state in this analysis, since it’s one that had been considered relatively safe for Clinton before. But Clinton’s decline has been steeper recently in the Midwest, including in Michigan, Wisconsin, Ohio and Iowa. Conversely, her numbers have held up a little better elsewhere, such as in New Hampshire and Pennsylvania. That’s not to say there’s been no decline in these states, however. Recent Pennsylvania polls have had Clinton ahead by an average of 5 or 6 percentage points, which is nothing to complain about, but is down from the roughly 9-point lead she held in the state in mid-August.

Clinton’s Florida and North Carolina numbers have also held up comparatively well (although there’s a lot of disagreement among pollsters in both states). As a result, they’ve moved slightly closer to the tipping point and have become more important for Clinton, serving as a potential hedge for her in the event of a further deterioration of her numbers in the Midwest. On the flip side, some states presumed to be safe for Clinton might not be. Her lead in Colorado has fallen quite a bit, for instance, although this is somewhat counteracted by stronger polls in Nevada.

Some of these results look a bit different if you use the FiveThirtyEight model’s fancy-schmancy averaging methods, instead of the simple one I used above. For example, our adjusted polling average in Iowa has Clinton down 1 percentage point there, instead of 2 or 3 points. But the results don’t look that much different. Either way, Clinton is up by 3 points, give or take, in the states that matter the most.

Overall, Clinton’s chances of winning the Electoral College are 70 percent, according to our polls-only forecast, and 68 percent according to polls-plus. That’s a slight improvement in both cases from Wednesday, when her numbers were 67 percent and 66 percent, respectively.

Why Is Trump Gaining On Clinton?

In this week’s politics chat, we talk about why the presidential race has tightened. The transcript below has been lightly edited.

micah (Micah Cohen, politics editor): Welcome, everyone, to our first post-Labor Day politics chat. Things are afoot! And we’re here to talk about them. Our question for today: Why is Donald Trump gaining on Hillary Clinton in the polls?

To set us up, Harry, briefly describe how the race has shifted since the conventions.

clare.malone (Clare Malone, senior political writer): I hope no one is wearing linen and/or white to this chat.

harry (Harry Enten, senior political writer): Clinton got a large convention bounce. In our polls-only model, for example, Clinton got as high as an 89 percent chance of winning and was projected to win the national vote by 8.6 percentage points. But since mid-August, her chances have fallen steadily. Right now, she’s at a 69 percent chance of winning and only is projected to win the national vote by 3.9 percentage points.

micah: So, let’s talk causes, then we can talk about where Trump’s gains have come from. Can we point to any one or two things to explain the race tightening?

clare.malone: I think it’s not so much what Trump has been doing, so much as Clinton has been the top story in most news cycles these last couple of weeks, and she’s not being portrayed in a flattering light, generally speaking.

Trump had a rough go after the conventions, and Clinton just sorta laid back and let him run his mouth for a while. But now, the Clinton Foundation stuff is raising itself as a specter in her campaign — whodathunk a charitable foundation would prove to have such nefarious connotations! But things like the AP story that went viral about her supposed “pay to play” with people at the State Department didn’t play well.

natesilver (Nate Silver, editor in chief): I don’t think there’s any one cause — and if you look at the data, it suggests both some gains for Trump and some decline for Clinton.

harry: To Clare’s point, I think it’s what we’ve spoken about before: Both of these candidates are flawed. The American public isn’t in love with either one. When they are reminded of a candidate’s flaws, they seem to oscillate back to undecided or towards the other candidate. In this case, we’re back to where we were pre-conventions. But I haven’t seen Trump breakthrough yet in ways he hasn’t before.

clare.malone: Is the number of undecideds right now on-track historically? Or is it out of the norm?

natesilver: No, it’s way higher, at least compared to recent elections. You have 18-20 percent of the electorate that’s either undecided or voting for one of the (largely anonymous) third-party candidates. That figure was like 5-10 percent at a comparable point four years ago. People largely ignore that, because they get focused on the margin between Clinton and Trump, when it’s maybe like the most important thing right now.

micah: So, this seems like a good point to bring up the Gary Johnson gaffe…

clare.malone: Mitt Romney tweeted this yesterday:

I hope voters get to see former GOP Governors Gary Johnson and Bill Weld on the debate stages this fall.

— Mitt Romney (@MittRomney) September 7, 2016

And then….

Gary Johnson asked “What is Aleppo?” in an interview … and not in response to a “Jeopardy!” question. I also think we should be fair to him, though — he released this statement earlier this morning where he basically said his brain had been in a space where he thought “Aleppo” was an acronym. That kind of brain fart seems pretty plausible to me, especially since he’s not engaging in a full-court press campaign, where he’s constantly guarded — it’s a much more laid-back operation.

harry: According to Huffington Post/Pollster.com, 53 percent of Americans are undecided when asked whether they have a favorable or unfavorable view of Johnson. The Aleppo “gaffe” probably doesn’t help. But the third-party stuff at this point seems as much about the major party candidates than anything Johnson or Jill Stein has done themselves.

natesilver: Trump’s whole strategy in the primaries was basically to make gaffes like that and then get more media coverage as a result. So maybe it will help Johnson? I’m being flip, but he and Stein are basically getting votes as no-name alternatives right now, and not because people like their policies per se.

micah: So, let’s put a bow on the Johnson thing: Any predictions on whether it will affect the horse race? Or if so, how?

natesilver: It’s plausible that some Johnson/Stein voters could begin to drift back into the Clinton camp if the race looks like it’s tightened up and she might need their votes. I’m not sure it will have that much to do with Aleppo-gate, though.

harry: I tend to think it won’t impact the race.

clare.malone: I will third that. I think people don’t know that much about Johnson yet, and they won’t really unless he gets on the debate stage.

micah: OK, so back to Trump-Clinton: As Harry said, Clinton’s margin is essentially back to where it was before the conventions — do you think this is where the natural equilibrium of the race is? (Clinton up 3-4 points.)

natesilver: I don’t make a lot of presumptions about where the equilibrium is, although it’s worth noting that over the very long term, Clinton’s lead is a bit larger than it is now — it has been about 5 points, on average, throughout the campaign.

clare.malone: I think she’s at a low ebb right now… I kinda feel like the debates could be good for her. In the same way that the convention was good for her. She’s the kind of person who preps like crazy for those sorts of things and I think she, unlike Matt Lauer, will certainly be calling BS on a lot of Trump statements.

micah: The Lauer stuff was ridic.

I’ll be the presumptuous one though: The fundamentals of this race, as Harry wrote yesterday, suggest a Clinton edge of about 1 point. Then I think you add 3-4 points because Trump is a historically awful candidate. So, all else being equal, I think the race oscillates around a +4-5 Clinton edge.

natesilver: Well, maybe. I think you start at Clinton +1, add 10 points for Trump being a really awful candidate, then subtract 7 for Clinton being a bad candidate too.

But there’s a margin of uncertainty around all those things, which is why anything from a narrow Trump win to a 10-point Clinton win wouldn’t be that surprising.

micah: Clinton has had some bad news cycles, as Clare said, so she’s a bit below the default now.

clare.malone: This sounds like that “math exercise” about how old or young you’re allowed to date.

micah: What’s that math, Clare?

harry: (28/2) +7 = 21.

clare.malone: That thing.

natesilver: I do think there’s something to the notion that — at least in terms of the media coverage — it’s maybe not the worst thing for the media to have “gotten this out of their system” on Clinton, so they might be ready for a Clinton comeback story by the time of the debate.

clare.malone: Hm. Well, I think that David Axelrod made a legit point on Twitter earlier — he said this Lauer screw up might put the fear of God in the other debate moderators:

.@mlauer forum performance could aid @HillaryClinton in debate. If @LesterHoltNBC reads critiques, he won't give @realDonaldTrump same pass.

— David Axelrod (@davidaxelrod) September 8, 2016

The idea that they need to come prepared to fact-check like crazy.

natesilver: Here’s the thing: Historically, the challenging-party candidate (Trump in this case) has gained ground after the first debate. But the first debate is usually held about a month after the incumbent party’s convention. There’s a much bigger gap this year. So maybe the challenging party’s debate bounce is actually the incumbent party’s convention bounce wearing off? And this year, without a debate, the convention bounce wore off gradually instead?

micah: Nate, does that apply when there’s no incumbent running too?

clare.malone: I just don’t see any way how the debate is good for Trump. Does anyone have a theory?

natesilver: THE EXPECTATIONS GAME. Journalists will use phrases like “the expectations game” to basically justify a biased interpretation of the debate.

micah: That, and: The first debate typically hurts the incumbent president because it puts the challenger on the same level as the president — something like that could apply to Trump too, right?

clare.malone: But… the expectations game is usually about the finer points of debating. (Which I know you know about, Nate!) Trump literally lacks a handle on basic facts.

He’s good on his feet, but he’s got cotton-candy talking points, not steak-and-potato ones.

harry: Things happen. In this case, Trump says nothing racist, pronounces the name of a foreign country correctly and boom!

natesilver: I’m just saying, the media could “call” the first debate for Trump even if that’s false.

clare.malone: Doesn’t that become apparent during a two-hour television event?

micah: I’m not sure the debates are that effective at sussing out substance.

clare.malone: But I think general election debates are very different from primary season ones, especially this primary season’s. Ted Cruz was drafting off Trump for most of the primary and not calling him out.

micah: Sarah Palin and Joe Biden tied in their debate in 2008, according to the media.

natesilver: Actually, I think the media was ready to declare Palin the winner of that debate, until polls started saying otherwise.

micah: We have to save this debate talk though.

clare.malone: lol, like it’s cake and ice cream for a very special occasion.

micah: Back to the tightening polls! Nate, you said earlier that Clinton’s support has dipped a little and Trump’s support has risen a little. Can we say anything about which types of voters they’re losing and gaining, respectively?

natesilver: Perhaps not a lot. There’s some evidence that Trump is regaining ground with Republicans more than with other groups.

harry: Trump’s doing a little better with Republicans, while Clinton is doing a little worse among Democrats if you look at the Pollster.com aggregates.

clare.malone: Is this the whole people tuning in after the summer thing? I feel like I’ve heard countless people say, “We just have to wait until voters get their kids back to school and then they’ll pay attention to the election.” Once they see what they’ve been missing, it might turn into “tune in, drop out, man.”

natesilver: I can tell you from looking at FiveThirtyEight’s traffic over the years that interest usually takes a big step up as of the conventions, and then steadily increases from there. But again, maybe the timing has been a bit thrown off by the early conventions.

clare.malone: And the Olympics.

harry: Freaking Olympics.

clare.malone: And Taylor Swift breaking up AGAIN.

natesilver: What is Taylor Swift? I’m just kidding. I know who she is.

harry: That was the 49ers’ secret weapon. The John Taylor swift.

natesilver: I did at one point think Zephyr Teachout was a thing and not a person.

clare.malone: hah! The best name in politics, hands down.

natesilver: Like: Let’s go to the Zephyr Teachout in Park Slope tonight! It’ll be groovy.

clare.malone: It’s even better than a kid who went to my high school whose name was “Cleveland Brown.”

harry: One thing I’ll point out is we’re after Labor Day now. We’re after the conventions now. And Clinton still holds the lead and a significant one at that. Trump still has work to do. He’s got room to improve among Republicans, but he’s a very unpopular politician. He also has no real ground game compared to Clinton. He can win, but I wonder how he would win.

clare.malone: PBS NewsHour had a thing on his ground game a little while back. She has 291 campaign offices in battleground states; he has 88.

natesilver: Well, if Trump only has to get 44 percent of the vote to win, instead of 50 percent, because there’s a big third-party vote, that makes it a lot easier.

How does Trump win? It’s on the margin rather than with some brilliant strategy, I think. Stay relatively gaffe-free and you’ll probably get a few reluctant Republicans to come home to you. Keep that trust question about Clinton forefront in voters’ minds. Maybe you both go into Election Day with a 40-ish percent favorability rating. And then the third-party voters and the swing states break in a way that’s favorable for you.

Now, that marginal strategy would work a lot better if Trump had a better ground game. But still — it’s not that hard to imagine how he could win.

clare.malone: Yeah, that about tracks.

harry: Trump has only a 4 percent chance of getting a majority of the vote right now, according to the polls-only forecast.

natesilver: And Clinton has only a 21 percent chance. We’re probably not going to see anyone hit 50 percent this year.

harry: You’re always

2016 NFL Predictions

How The Media Have Covered The 2016 Campaign

 Subscribe: iTunes |ESPN App |Download |RSS |New to podcasts?

This week, the Elections podcast team looks at the challenges the press faces in covering the 2016 campaign. With new reports on the dealings of the Clinton Foundation, liberal commentators in particular have accused the media of falsely equating Hillary Clinton’s weaknesses with Donald Trump’s. The gang also takes stock of the tightening in recent national and swing state polls.

We have two live shows coming up this fall in New York and Chicago. More details here.

You can listen to the episode by clicking the “play” button above or by downloading it in iTunes, the ESPN app, or your favorite podcast platform. If you are new to podcasts, learn how to listen.

The FiveThirtyEight Elections podcast publishes Monday evenings, with occasional special episodes throughout the week. Help new listeners discover the show by leaving us a rating and review on iTunes. Have a comment, question or suggestion for “good polling vs. bad polling?” Get in touch by email, on Twitter or in the comments.