Nate Silver's Blog, page 125

Election Update: Clinton’s Lead Keeps Shrinking

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

I, for one, welcome the unofficial end of summer. I’ll miss the Olympics and fancy tomato salads. But it’s an election year, and Labor Day is usually accompanied by a return to more substantive news cycles — along with a significant increase in the amount of polling.

That was certainly true Tuesday morning, which brought a bevy of new data, including about a half-dozen new national polls and a 50-state poll from SurveyMonkey (conducted in conjunction with The Washington Post). People are focusing on the flashier results among these polls: that CNN’s poll shows Donald Trump narrowly ahead among likely voters, for instance, while SurveyMonkey has Hillary Clinton tied with Trump in Texas. At times like these, though, it’s especially useful to zoom out and take a more holistic approach.

The clearest pattern is simply that Trump has regained ground since Clinton’s post-convention peak. He now has a 31 percent chance of winning the election according to our polls-only model, and a 33 percent chance according to polls-plus. For a deeper look, let’s run through our set of 10 framing questions about the election in light of the most recent polling:

1. Who’s ahead in the polls right now?

Clinton’s ahead, by a margin of about 3 percentage points in an average of national polls, or 4 points in our popular vote composite, which is based on both national polls and state polls. While the race has tightened, be wary of claims that the election is too close to call — that isn’t where the preponderance of the evidence lies, at least for the moment. If one candidate is ahead by 3 or 4 percentage points, there will be occasional polls showing a tied race or her opponent narrowly ahead, along with others showing the candidate with a mid- to high single-digit lead. We’ve seen multiple examples of both of those recently.

In swing states, the race ranges from showing Trump up by 1 point in Iowa to a Clinton lead of about 6 points in her best states, such as Virginia. That’s a reasonably good position for Clinton, but it isn’t quite as safe as it might sound. That’s because the swing states tend to rise and fall together. A further shift of a few points in Trump’s favor, or a polling error of that magnitude, would make the Electoral College highly competitive.

2. What’s the degree of uncertainty?

Higher than people might assume. Between the unusually early conventions and the late election — Nov. 8 is the latest possible date on which Election Day can occur — it’s a long campaign this year. But just as important, many voters — close to 20 percent — either say they’re undecided or that they plan to vote for third-party candidates. At a comparable point four years ago, only 5 to 10 percent of voters fell into those categories.

High numbers of undecided and third-party voters are associated with higher volatility and larger polling errors. Put another way, elections are harder to predict when fewer people have made up their minds. Because FiveThirtyEight’s models account for this property, we show a relatively wide range of possible outcomes, giving Trump better odds of winning than most other statistically based models, but also a significant chance of a Clinton landslide if those undecideds break in her favor.

3. What’s the short-term trend in the polls?

It’s been toward Trump over the past few weeks. Clinton’s lead peaked at about 8.5 percentage points in early August, according to our models, and Trump has since sliced that figure roughly in half. Of Trump’s roughly 4-point gain since then, about 2 points come from Trump’s having gained ground, while the other 2 points come from Clinton’s having lost ground — possibly a sign that her lofty numbers in early August were inflated by a convention bounce.

One slight caveat: If you’re talking about the very short term, it’s not quite as clear who’s gaining, as the most recent daily and weekly tracking polls have been flat lately instead of showing continued gains for Trump. By late this week, we should have a better sense of whether Trump’s position is still improving.

4. What’s the medium-term trend in the polls?

It depends on where you measure it from. Clinton had a lead of 6 to 7 percentage points when we launched our forecast in June. That dwindled to about 3 percentage points just before the conventions got underway, and then a tie once Trump got a modest bounce after the Republican convention. Clinton then got a comparatively large bounce after her convention, bringing her lead to about 8 points, but it’s receded some. Overall, her current lead of 4 percentage points is close to or slightly below where the race has been on average throughout the campaign.

5. Which states shape up as most important?

It’s still early enough — and we’re lacking recent, high-quality polling in enough states — that I’d discourage you from fixating on any one exact combination of states that Clinton or Trump might win to clinch the Electoral College. Instead, you might think of this election as a battle between the Big Ten states and the ACC states, either of which offer a plausible path to victory for Clinton. If she holds on to most of the Big Ten states that President Obama won four years ago, such as Pennsylvania, Ohio, Michigan and Wisconsin, she can afford to lose ACC states such as Florida, Virginia and North Carolina. If she can win either Florida or both Virginia and North Carolina — and certainly if she sweeps all three ACC states — she can sacrifice quite a bit of ground in the Big Ten. The handful of competitive states outside of these groups, such as Nevada and New Hampshire, have few enough electoral votes that they’ll serve as tiebreakers only in the event of an extremely close race.

According to our tipping-point index, however, the single most important state is Florida. That’s because its 29 electoral votes are as much as many combinations of two and three swing states put together.

6. Does one candidate appear to have an overall edge in the Electoral College, relative to his or her position in the popular vote?

Our models, somewhat in contrast to the conventional wisdom, have usually found that Trump is more likely to win the Electoral College while losing the popular vote than the other way around. Some of this is for a quirky reason: Trump is underperforming recent Republican nominees in polls of deeply red states. Last week, for instance, there were new polls of Kansas and Alaska that showed Trump winning by 7 points and 10 points, respectively. By comparison, Mitt Romney won Kansas by 22 points and Alaska by 14. Losing states like those by 10 points instead of 20 would yield a better popular vote margin for Clinton, but wouldn’t help in the Electoral College.

The SurveyMonkey poll showing a tied race in Texas is in line with this theme. The race probably isn’t really tied there, as other recent polls in Texas have Trump ahead. But a close call — Clinton losing Texas by only 5 percentage points — could yield wasted votes for Clinton in terms of their impact on the Electoral College. It’s plausible that Clinton gains among Hispanic voters are contributing to this pattern, since most Hispanics are not concentrated in swing states. (Almost half the Hispanic population is in Texas or California alone.)

7. How do the “fundamentals” look?

Some “fundamentals”-based models, which look at economic data and other nonpolling factors to forecast the election, suggest that a generic Republican candidate should be a slight favorite over a generic Democrat in this election. Our polls-plus model also contains a fundamentals model based on an economic index, and it slightly disagrees, finding that the economy is about average or, based on more recent data, very slightly above average — conveying just the slightest re-election edge to Democrats.

This is literally something of an academic debate, however. Overall, the fundamentals imply that the election ought to be close. If Clinton or Trump win by a significant margin, it probably has more to do with the peculiarities of the candidates than the underlying conditions of the race.

8. How do FiveThirtyEight’s forecasts compare against prediction markets?

Trump’s chances are currently about 30 percent in betting markets, a close match for FiveThirtyEight’s forecasts.

9. What would keep me up late at night if I were Clinton?

My first question would be whether the race has settled into a 4-point Clinton lead, as the polls have it now, or is continuing to trend toward Trump. If I’m still ahead by 4 points or more at the time of the first debate on Sept. 26, I’ll feel reasonably good about my position: A Trump comeback would be toward the outer edges of how much trailing candidates have historically been able to move the polls with the debates. If the race gets much closer, though, my list of concerns gets a lot longer. It would include geopolitical events that could work in Trump’s favor, third-party candidates who seem to be taking more votes from me than from Trump, and the tendency for incumbent candidates (since Clinton is a quasi-incumbent) to lose ground in the polls after the first debate.

10. What would keep me up late at night if I were Trump?

As the polls have ebbed and flowed, I’ve been 8 or 10 points behind Clinton at my worst moments, but only tied with her at my best moments. I’ve also never gotten much above 40 percent in national polls, at least not on a consistent basis, and I’ve alienated a lot of voters who would allow me to climb higher than that. In other words, maybe that dreaded Trump ceiling is there after all, in which case I’ll have to get awfully lucky to win the election, probably needing both a favorable flow of news in the weeks leading up to Nov. 8 and a large third-party vote that works against Clinton.

Election Update: As The Race Tightens, Don’t Assume The Electoral College Will Save Clinton

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

The race between Hillary Clinton and Donald Trump has tightened. Clinton, whose lead over Trump exceeded 8 percentage points at her peak following the Democratic convention, is ahead by 4 or 5 percentage points today, according to our polls-only forecast.

The tighter margins in the polls, which reflect a loss of support for Clinton along with a modest improvement for Trump, have come gradually over the past few weeks. The evidence of a tightening has become more widespread, however, and it’s particularly clear in polls that surveyed the race just after the conventions and are retaking its temperature now. Fox News’s national poll, for instance, had Clinton up by 9 points just after the conventions (in the version of the poll that included third-party candidates) and has her up by 2 points now.

There isn’t any guarantee that Trump will continue to gain ground. Over the course of the year, polls have oscillated between showing a dead heat at Trump’s best moments and a lead of 8 to 10 percentage points for Clinton at her peaks. We’re about halfway between those goal posts now. It’s plausible that the recent shift reflects Clinton’s convention bounce wearing off — reversion to the mean — as much as it does momentum for Trump per se. Most importantly, Clinton is still ahead, with a 74 percent chance of winning according to the polls-only model and a 70 percent chance according to polls-plus.

But what if the race continues to tighten? I’ve often heard Democrats express a belief that Clinton’s position in the swing states will protect her in the Electoral College even if the race draws to a dead heat overall. But this is potentially mistaken. Although it’s plausible that Clinton’s superior field operation will eventually pay dividends, so far her swing state results have ebbed and flowed with her national numbers.

Take Wisconsin, for example. At her peak, Clinton had a double-digit lead there, according to our polls-only forecast. By Wednesday morning, it had declined to an estimated 7 points, as a result of our model’s trend line adjustment — which adjusts polls in all states based on shifts it detects in the race overall — along with data from the Ipsos 50-state tracking poll.

We know that some readers don’t like the trend line adjustment. But if anything, the model hadn’t been aggressive enough. Two highly rated, traditional telephone polls of Wisconsin came out Wednesday, and they showed Clinton up by just 3 points and 5 points. The 3-point lead was in a poll from Marquette University, which had Clinton up by 13 points just after the Democratic National Convention. (As of Thursday morning, Clinton is projected to win by 5 points in Wisconsin, according to the polls-only model.)

Usually, the trend line adjustment helps the model peg what forthcoming polls will look like in a state even if there haven’t been many of them recently. When Clinton established a roughly 8-point lead nationally in August, for example, it figured we’d see polls showing her with leads of 10 to 12 percentage points in some of her better swing states, such as Michigan and Colorado, along with leads of 5 to 6 percentage points in swing states that are just slightly redder than the country as a whole, such as Ohio and Florida. And that’s pretty much what we saw, at least on average. Now that the race has tightened to 4 or 5 points nationally, the model expects to see narrower leads — along with some polls showing a tie or Trump slightly ahead in the more red-leaning swing states:

ADJUSTED POLLING AVERAGESTATEAUG. 14AUG. 31Arizona+1.6-3.9Colorado+10.5+6.1Florida+6.3+3.1Georgia+0.7-2.3Iowa+5.0+0.7Michigan+11.8+6.0Minnesota+12.6+7.7Nevada+4.2+0.8New Hampshire+9.0+4.2North Carolina+5.5+1.2Ohio+6.4+2.7Pennsylvania+9.3+4.8Virginia+9.1+7.1Wisconsin+12.6+4.8Clinton’s projected leads have shrunk in swing states

Keep in mind that these numbers are self-correcting. For instance, the model expects new Ohio polls to show Clinton ahead by 2 or 3 percentage points, provided they don’t have a strong Trump-leaning or Clinton-leaning house effect. If new surveys deviate significantly from that range, the model will adjust itself accordingly. But usually this method gets things about right. Swing states are swing states for a reason — they closely follow the overall national trend.

The other thing to notice about Clinton’s swing state polls is that they aren’t especially strong (or weak) relative to her national polls. At her post-convention peak, Clinton’s path of least resistance to 270 electoral votes appeared to run through a set of states that included Pennsylvania, Wisconsin and New Hampshire, among others. But in Pennsylvania, the most recent polls have Clinton ahead by margins ranging from 3 to 8 percentage points — perfectly fine, but not that different from her national numbers. We haven’t gotten much data recently from New Hampshire, but it can be swingy, and the most recent numbers from the Ipsos poll (caveat: very small sample size) showed Trump ahead. We did get some high-quality polls from Wisconsin, and, as I mentioned, they weren’t that good for Clinton.

Overall, Clinton’s leads in the tipping-point states — the ones most likely to determine the Electoral College winner in a close election — average about 4 percentage points, close to her numbers in national polls.

With a tighter race, the model’s expectations for Clinton are lower. A new poll showing her up by 5 or 6 points in Florida or Ohio — which would have been a ho-hum result a few weeks ago — could be a terrific one for her today, depending on the pollster.

What In The Hell Is Trump Doing On Immigration?

In this week’s politics chat, we wonder what the Trump’s campaign recent rhetoric on immigration might mean. The transcript below has been lightly edited.

micah (Micah Cohen, politics editor): Hey, everyone. Today’s topic: Donald Trump’s shifting rhetoric on immigration policy. Trump is scheduled to give a speech on this topic tomorrow, and he will ostensibly lay out some concrete proposals. (Maybe?) But in the meantime, I wanted to talk about the politics of immigration and what the Trump campaign is doing.

First, though, our dear colleague Clare Malone is vacationing in Cleveland this week, so we’re instead joined today by Anna, who mostly reports on public health for us but has covered the immigration debate pretty extensively too — welcome, Anna!

anna (Anna Maria Barry-Jester, senior writer): Hello!

natesilver (Nate Silver, editor in chief): I was going to make fun of “vacationing in Cleveland.” But people from Michigan actually vacation in Cleveland, or at least Sandusky (Cedar Point! Woo-hoo!).

harry (Harry Enten, senior political writer): I’m vacationing in NYC this week. The subway is the best.

anna: I vacationed in both Cleveland and Sandusky just about every year of my youth. Going to Cedar Point as part of a physics class field trip was somewhat of a life-defining moment for me.

micah: So, NPR has a good rundown of all the disparate, sometimes-conflicting things Trump and his campaign have said about immigration in the past couple weeks, but here’s the gist: Trump ran on a hardline immigration position during the Republican primary, calling for the immediate deportation of all undocumented immigrants, for example. Lately, though, he’s made noise about seeking a more “humane” position.

What’s going on? Is Trump making a late play for Latino voters?

natesilver: I think Trump is getting advice from five different people, and they’re all telling him different things, and he repeats the last thing about immigration he heard.

anna: I don’t think he’s just making a play for Latino voters, Micah. His build-a-wall-and-remove-11-million-people stance isn’t popular with a lot of voters.

harry: When someone is losing, they start throwing stuff up against the wall. That’s what’s going on, in my opinion.

micah: Yeah, it seems like even if he shits positions, Anna, voters won’t forget about his old one.

* “Shifts,” that should say.

natesilver: I think “shits” is more accurate tbh.

micah: We can’t publish that.

harry: Trump is playing right into the Clinton critique that he isn’t ready to be commander in chief — shifting positions, then getting angry that people are saying he is flip-flopping.

anna: A lot has been written about the fact that flip-flopping or shifting or whatever you want to call it doesn’t seem to be held against Trump in the same way it has with previous candidates. But I do wonder if this might hurt his image with his early core of supporters.

micah: So, that question is interesting…. We’ve found that immigration has become a really important issue in GOP politics — it’s one of the most predictive in who identifies as a Republican and who supports Trump. Doesn’t that suggest that if he softens on immigration, some of his core supporters might abandon him?

natesilver: Trump doesn’t have very many core supporters — his favorability rating is only about 35 percent — so the ones he does have are a pretty hearty lot.

micah: But his hardline immigration stance seems like one of the main reasons that 35 percent like him.

natesilver: But I’m not sure if those supporters give a damn about the substance of immigration policy. If you ask people substantive policy questions about immigration — even if you ask Republicans — the answers are surprisingly centrist.

Most of Trump’s hardcore supporters live in areas where there aren’t very many immigrants. Immigration is a symbolic issue, instead, that candidates wield to prove they’re not part of a corrupt establishment.

anna: Right, so if he’s seen as pandering to the rest, couldn’t that hurt him with those hardcore supporters?

micah: I guess, intellectually, I would think this would hurt him. Yet … I don’t think it will.

natesilver: Look, he’s doubled and tripled and quadrupled down on appealing to that 35 percent — probably to his ultimate detriment. But it also means he has some credibility to lose.

harry: Trump has two problems. The first is that an unusually large segment of Republicans dislike him. The second is that an even larger segment of general election voters dislike him. And many general election voters have an unfavorable view of Trump because of his divisive comments on race and immigration — 59 percent, according to a recent Quinnipiac poll, for example, said “the way Donald Trump talks appeals to bigotry.” And so what happens is Trump is trying to play both sides, and to me seems like he’s playing none. He is losing the general electorate and getting a smaller percentage of Republicans than Clinton is of Democrats.

natesilver: And he’s also doubled and tripled and quadrupled down on bigotry — which is why he probably won’t convince anyone with this shift, if there’s even a shift.

anna: On the other hand, his early talk of building a physical wall and evicting 11 million people is … absurd, at best, when you think about making it a reality, so, as Nate was saying, this could be seen as a positive evolution for him by Republicans who were turned off by some of his early remarks on immigration. I keep reading quotes from voters saying “that was just a starting point for negotiating….”

micah: Yeah, maybe his supporters will view his shifting positions in that more positive light — i.e., “he’s a negotiator.”

Let’s say he does soften his position: Will it have any chance of bringing wayward Republicans (let’s say, suburban white women) back into the GOP’s camp?

natesilver: “Any chance” is a low bar, but I basically think he won’t convince anyone, especially since he’s handled it so clumsily.

harry: Who can say? But the fact that his favorable ratings have been so bad for so long suggests to me that it will be very tough.

micah: IDK … he’s been a bit more scripted/traditional over the past couple weeks and has gained in our election forecasts. Moreover, there’s like, say, 10 percent of voters who are conservative and normally vote Republican but who so far are saying they’ll vote third party or are undecided. Aren’t those voters kind of low-hanging fruit? If Trump can just convince people he’s not a bigot?

anna: Yeah, isn’t Clinton’s favorability rating pretty bad as well? If he can convince people that it’s just part of the Trump show, couldn’t that win some people over?

harry: Wait a minute. He hasn’t really gained per se as much as Clinton has lost.

natesilver: Yeah, exactly. Trump’s at 37.7 percent in our national polling average. Where was he three weeks ago? 37.8 percent. No change. Clinton’s declined some, though.

I think what you see is her convention bounce wearing off, along with maybe a couple of very bad news cycles for Trump just after the convention given his comments about the Khan family.

It’s not obvious that his “turnaround” has actually been successful per se.

micah: OK, but then let’s play Trump adviser for second: What would you have him do on immigration? Anna, is there more of a consensus position on immigration policy that Trump can adopt?

anna: I might have said yes back in 2013 when the Gang of Eight was together pushing immigration reform. During an election season? I’m not sure there is. There’s a safer route, but he’s made this a key rallying issue, which probably helped him early on.

micah: I mean, I guess Trump’s winning the nomination suggests that his hardline stance was the GOP primary vote’s consensus on immigration — maybe he can’t square the circle between that group and the Gang of Eight folks.

harry: I think he just shouldn’t talk about it.

natesilver: I’m not sure that immigration is the issue I’d have pivoted on, especially if I were going to do it so clumsily.

Pivoting to the center on economic issues might have been more interesting in some ways.

harry: He should talk about Social Security.

anna: Agree, Nate, but isn’t it a big part of what got him out ahead of the rest early on in a busy race? Can he just stop talking about it now and inspire the same fervor at rallies? Actually, yeah, he probably could. And silence could probably be an effective Band-Aid on that wound while he talks about economic policy.

micah: Well, he’s doing the exact opposite, giving a “MAJOR!!!” speech on immigration Wednesday; what will you all be watching for?

natesilver: I think I have a fantasy football draft so I’ll be watching for whether there’s a run on wide receivers.

harry: I’m less interested in what he says than how he says it. That’s part of what got Trump in so much trouble in the first place. It’s one thing to say we’re going to build a wall. It’s another to talk about Mexicans as rapists.

micah: Nate, you’re playing into people’s stereotypes about how we don’t care about issues.

natesilver: 180 degrees wrong! It’s not a fucking policy speech!

micah: It might be.

natesilver: It would be disrespectful to policy speeches to see it as a policy speech.

harry: This chat is rated R.

micah: The speech hasn’t even happened yet! How will you know if it’s a policy speech if you don’t watch?

natesilver: Anything he says carries no substantive weight because he’s already taken every available position on the issue.

anna: There’s been a lot of talk of “humane” and “fair” policy recently from his camp. The wall, deportation, ending birthright citizenship … that’s all against immigrants. I wonder if he’s really going to address the humanity of immigrants as has been suggested.

micah: What would that look/sound like?

anna: What the process would be for a touchback system, for example.

To be clear, I’m not expecting humanity! But all of his ideas (Nate’s right, they aren’t policy) so far are so completely absurd. They aren’t feasible.

harry: Again, this isn’t about substance. It’s about tone.

anna: You’re probably right for the horse race that it doesn’t matter what Trump actually says, Harry, but I don’t know how he walks his current immigration stance back without injecting some policy into the rhetoric. For voters who care about immigration, he’ll need some viable policy to make them believe what he’s saying. Then again, as you’ve both said, most voters don’t care that much about immigration.

natesilver: If nothing else, I think we’ll learn something about who has his ear from the speech.

The General Election Is About To Hit The Home Stretch

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

We’ve been saying for months that the general election won’t begin in earnest until after Labor Day. Well, Labor Day is almost here, so our elections podcast crew checks in on what we know, what we don’t know and what we want to know heading into the fall. We discuss which states are really in play, which demographic groups will be crucial in determining who wins, how accurate the polls are and lots more.

Like live shows?!
-Wed @atlasobscura! https://t.co/hJgM4zfIf5
-Thur @askroulette! https://t.co/Bgk7gSS3cy
-538 tix! https://t.co/Xd9mZvH00p

— Jody Avirgan (@jodyavirgan) August 29, 2016

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.

Election Update: It’s Too Soon For Clinton To Run Out The Clock

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

Last week, Politico reported that Hillary Clinton’s campaign was set to employ a “run out the clock” strategy, declining to respond to recurring controversies even at the risk of seeming nonresponsive. In the abstract, such a strategy could make sense. Clinton has a fairly clear lead in the polls. There are only 10 weeks to go until the Nov. 8 election — and less than that until early voting, which begins in late September in some states.

But Clinton shouldn’t get too complacent. After mixed evidence before, it’s become clearer, at least according to our forecast models, that Donald Trump has regained some ground on her. Clinton’s national lead in our polls-only forecast has gone from a peak of about 8.5 percentage points two weeks ago to 6.5 percentage points as of Sunday evening — that is, a 2-point gain for Trump over two weeks. Correspondingly, Trump’s chances of winning the election have improved from a low of 11 percent to 19 percent.

Trump’s gains have been more modest in our polls-plus forecast, which discounted Clinton’s early August polls because of a potential convention bounce and which anticipated that the race would tighten. In polls-plus, which forecasts that Clinton’s margin over Trump will narrow to roughly 4 percentage points by Election Day, the clock is more of an ally to Clinton and an enemy to Trump. Still, Trump is keeping slightly ahead of the pace of improvement that polls-plus expected of him. His chances of winning are 27 percent according to polls-plus, up slightly from 25 percent a week ago and from a low of 21 percent earlier this month.

None of this is to say that there’s been some game-changing shift toward Trump. There probably hasn’t been, and you wouldn’t necessarily anticipate one given that, after such an action-packed period earlier this summer, there have been a lot of slow news days amid the August doldrums. Generally speaking, polls don’t move suddenly without a good reason.

But the polls can move gradually, whether because they’re reverting to a previous mean (a convention bounce wears off) or because of the cumulative effects of the campaign (an undecided voter finally makes up her mind). Lately, that movement seems to be slightly toward Trump.

The clearest case for this is in a set of daily and weekly tracking polls, several of which — including the Morning Consult poll, the UPI/CVoter poll, the Gravis Marketing poll, and the Ipsos/Reuters national tracking poll — show Clinton at or near her post-convention lows (and in some cases, within the margin of error of Trump overall). There are also a couple of exceptions: The Republican-leaning Rasmussen Reports poll showed Clinton gaining ground this week, and the Ipsos/Reuters 50-state tracking poll, which has a considerably larger sample size than Ipsos’s national tracking poll, showed Clinton steady to slightly improving from the previous week.

But a candidate doesn’t need to gain ground in every poll to gain ground in our forecast, just in the preponderance of them. If the race shifts by 7 points toward a candidate essentially overnight, as it did toward Clinton following the Democratic convention, it’s going to be obvious in almost every poll. But such moves are rare, and a gradual shift — like Trump gaining 2 points over two weeks — will manifest itself in fits and starts in the averages.

A concern for Trump is that these gains haven’t been as apparent outside of those national tracking polls, all of which are conducted online or by automated script and several of which have a Trump-leaning house effect. The most recent traditional telephone poll, from Quinnipiac University, had Trump down by 7 percentage points — or down 10 points without third-party candidates — a poor result for him given that Quinnipiac had been one of the better traditional pollsters for Trump earlier in the cycle. There haven’t been very many of those traditional polls lately; it’s been a few weeks since we got numbers from any of the five major national surveys that will be used to determine eligibility for the debates, for example.

Polls in swing states were all over the place last week, meanwhile. In Florida, for example, we saw surveys showing everything from a 14-point lead for Clinton to a 3-point lead for Trump. If you squint, you can perhaps perceive some movement back toward Trump in some of the red-tinged swing states, such as North Carolina. But it’s hard to say for sure. A lot of the swing state polls released last week were from pollsters surveying the states for the first time, meaning that they didn’t have trend lines, or if they did have trend lines, they didn’t show much change from the previous version of the survey. The pollster that had Clinton up by 14 points in Florida, for example, had her ahead by 13 points when it previously surveyed the state in June.

How the clock does and doesn’t help Clinton

It’s nice to have a model at times like these, instead of just throwing up your hands (or worse, cherry-picking polls to suit your case). And that model, as I said, shows Trump as having gained about 2 points over two weeks. If Trump keeps gaining 1 percentage point a week, he’ll beat Clinton by a couple of percentage points on Nov. 8. Hence, Clinton should probably not be picking out the White House drapes just yet.

Continued gains may not be so easy for Trump, however. He’s still at only 37 or 38 percent in national polls that include third-party candidates. That might seem like an easy number to improve upon, but his favorability rating is only about 35 percent, meaning that he’s already relying on support from a few voters who don’t like him but may vote for him to prevent a Clinton presidency.

Clinton also has some work to do. She’s at about 43 percent in national polls and in polls of key swing states — not enough to clinch victory, even if 6 to 10 percent of the vote eventually goes to third-party candidates, as appears increasingly likely. She’ll need to persuade a few undecided voters toward her side or get some of those third-party voters — more of whom have Clinton than Trump as their second choice — to turn out for her.

Our models rely on both the number of days until the election and on the number of undecided and third-party voters to calculate the uncertainty in the race. Because the conventions were held early this year, it can be easy to forget that it’s still just August. (On this date four years ago — Aug. 28, 2012 — Republicans were holding their convention in Tampa and the Democratic convention hadn’t even started yet.) Meanwhile, there hasn’t been much of a decline in the undecided or third-party vote. As compared with July 17, the date before this year’s Republican convention, Clinton has gained only 2 to 3 percentage points in the polls and Trump hasn’t gained at all.

Toward the end of the 2012 campaign, we frequently emphasized the distinction between closeness and uncertainty. President Obama led Mitt Romney by just 1 or 2 percentage points nationally, according to our models, throughout much of the stretch run of that campaign — a close race. But between Obama’s consistently strong numbers in the swing states, the low number of undecided voters, and a strong alignment between polls and economic “fundamentals,” there was a narrow range of plausible outcomes for that election, with most of them resulting in a second Obama term.

Related:

The Elections Podcast Checks In On The 2016 Forecasts

 Subscribe: iTunes |Download |RSS |Video

We talk about our elections forecasts endlessly, so why not do so in podcast form? Every other Friday in the regular elections podcast feed, editor-in-chief Nate Silver and politics editor Micah Cohen will sit down to talk about how the forecast model’s predictions have changed and why. It’ll get wonky. This week, they looked at the gaps between state and national polls and between online and live-interview polls. In each installment we’ll cover these basics:

What are the latest odds?What has shifted and why?Listener questionsWhat story are we watching? Which news is going to affect the polls?

This week we also announce the first of our fall live shows, slated for September 19th in New York City.

You can stream or download the full episode above. You can also find us by searching “fivethirtyeight” in your favorite podcast app, or subscribe using the RSS feed. Check out all our other shows.

If you’re a fan of the elections podcast, leave us a rating and review on iTunes, which helps other people discover the show. Have a comment, want to suggest something for “good polling vs. bad polling” or want to ask a question? Get in touch by email, on Twitter or in the comments.

Election Update: The Blue State Polling Abyss

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

Public Policy Polling on Wednesday released a poll of New Mexico, a state that hasn’t attracted much attention. Hillary Clinton led Donald Trump by 9 percentage points, while the Libertarian candidate and former New Mexico Gov. Gary Johnson had 16 percent of the vote.

The margin was a little closer than you might expect given Clinton’s 6 or 7 percentage point national lead and New Mexico’s demographics. But we don’t know very much about how New Mexico and a lot of other states like it are voting this year. The PPP survey was the first poll of New Mexico since PPP’s previous poll of the state in May. Other blue-leaning states have also received scant polling: Minnesota has barely been polled, for instance, and Maine hasn’t been polled much given that at least one of its electoral votes could be highly competitive.1

“Who cares?” you might protest. A 9 percentage point margin isn’t exactly close. Pollsters seem to think it’s more fun to poll traditionally red states such as South Carolina and Missouri — and, in fact, those states have been tight in recent surveys.

But while South Carolina and Missouri could allow Clinton to run up the score in the Electoral College, they’re very unlikely to determine the winner. In any election in which she wins South Carolina, for example, Clinton will almost certainly have already won North Carolina and probably also Georgia, meaning that she’ll be on track for 300-plus electoral votes with or without the Palmetto State. South Carolina and Missouri are unlikely to be tipping-point states, in FiveThirtyEight parlance.

By contrast, states such as New Mexico, Minnesota and Maine potentially could be tipping-point states if Trump makes a comeback. The polling there — showing Clinton leads in the high single digits or low double digits — isn’t far removed from the numbers in the more glamorous battleground states, such as Colorado, Virginia, Wisconsin and Pennsylvania, which have received far more attention.

And although they aren’t remotely competitive, the polling in populous blue states such as New York and California is also more important than you might think. These states have a lot of votes, making them important for understanding the relationship between the popular vote and the Electoral College. Recently, Clinton’s numbers have seemed to be a bit stronger in state polls than in national polls. Could that reflect her advantages in battleground states, such as her greater advertising presence and her better field operation?

Well, perhaps. But it’s not just battleground states where Clinton’s gotten good numbers; she’s also considerably overperforming Barack Obama in red states. Blue states are the missing part of the puzzle. Could it be that Clinton is underperforming in those? In the table below, you’ll find a comparison of Obama’s margin of victory in 2012 against Clinton’s current adjusted polling average in each blue state where we have at least one poll. By blue state, I mean everything Obama won by a wider margin than he did Wisconsin in 2012 (7 percentage points).

STATEOBAMA MARGIN OF VICTORY, 2012CLINTON LEAD IN ADJUSTED POLLING AVERAGE, 2016Minnesota+7.7+9.3Michigan+9.5+8.4New Mexico+10.1+9.4Oregon+12.1+9.3Washington+14.8+15.8Maine+15.3+8.0Illinois+16.8+21.4Connecticut+17.3+10.5New Jersey+17.7+13.7Delaware+18.6+14.7California+23.1+25.0Massachusetts+23.1+24.9Maryland+26.1+30.6New York+28.2+20.4Vermont+35.6+21.9How big are Clinton’s leads in blue states?

Adjusted polling average is per FiveThirtyEight polls-only model as of Aug. 25

This is a mixed bag of results. Clinton’s generally underperforming Obama in the Northeast, including in Maine, Connecticut, New York, New Jersey and Vermont. She’s running slightly ahead of him in California, however, and perhaps more surprisingly — given that it was Obama’s home state — well ahead of him in Illinois.2 Overall, weighted by each state’s 2012 turnout, she’s running about half a percentage point behind Obama in these states. So perhaps there’s something to the notion that Clinton’s underperformance in blue states can help explain some of the seeming differences between state and national polls.

But I wouldn’t take this too far. The relative lack of polling in these states means that the data is noisy. It doesn’t make a lot of sense that Clinton is outperforming Obama in Washington but underperforming him in Oregon, for instance, as polls suggest. In some cases, the polls don’t match the demographics of the states very well either. Our polls-only model “thinks” that Clinton should be ahead in New Mexico by about 14 points, for example, based on the patterns it’s seeing in other states, and not just by 9. It’s less skeptical of the polls in overwhelmingly white Maine and Minnesota, however. Minnesota’s worth keeping an eye on provided that — because somehow this has become a problem for him — Trump can get on the ballot there.

Related:

Has The Hillary Clinton Campaign Been Lucky Or Good?

In this week’s politics chat, we talk about whether Hillary Clinton has run a “good” campaign. The transcript below has been lightly edited.

micah (Micah Cohen, politics editor): Welcome, everyone. Today’s topic: Has Hillary Clinton’s campaign been lucky or good? Or, if both, in what proportions? For instance, you could describe the election so far like so:

Hillary Clinton was such a formidable candidate that she cleared most of the field for the Democratic primary. Her main opponent, Bernie Sanders, fell behind in the delegate race in the third contest and never recovered. Then, once she was the nominee, she opened up a pretty big lead over Donald Trump despite the fundamentals suggesting the race should be close or should even favor the GOP slightly.

Or, you could summarize the election to date thusly:

Hillary Clinton won the Democratic primary by default. Her main opponent was a 74-year-old Democratic socialist, from Vermont, and he still managed to get 43 percent of the vote. Now, she’s running against perhaps the worst presidential nominee in modern history and isn’t even leading him by that much.

So, which is it?

natesilver (Nate Silver, editor in chief): To a first approximation, No. 1 is true for the primary and No. 2 is true for the general election.

clare.malone (Clare Malone, senior political writer): Hah. It’s No. 2.

natesilver: Abso-fucking-lootley not No. 2 in the primary

clare.malone: You don’t think she was helped in some measure by the fact that Sanders struck a lot of Democrats as attractive but fundamentally untenable? Also, Nate segmented this out! I think if we’re going to pick a sweeping “good or lucky” she is more lucky.

natesilver: First of all, she deserves 100 percent credit for clearing the field of candidates other than Sanders. That reflects her formidability as a candidate, not just luck.

harry (Harry Enten, senior political writer): Yeah, Clinton was a strong primary candidate. The reason no one ran against her was because she locked down support from a majority of the party leaders before the primary started. That made it difficult for any opponent to gain a foothold.

But it’s not only that. When her back was against the wall, she won. Consider the state of Nevada. If Sanders had won there, it could have created at least a momentary disaster for Clinton. But Clinton used her connection with Nevada Sen. Harry Reid, ensured the casino workers got time off to caucus, and won the caucuses by about 5 percentage points. That’s not lucky. That’s good.

natesilver: And the Bernie constituency consisted of people who are by and large not Democratic base voters. Instead, lots of independents and young voters. Those voters might be important to the future of the Democratic Party and/or the future of left-wing politics, and it was really impressive that Sanders was able to mobilize them. But Clinton crushed Sanders with the voters she needed to win.

micah: OK, but you think she’s gotten lucky in the general? Hasn’t she run a pretty top-notch campaign?

natesilver: Hold on, shouldn’t we talk about the primary more? Because I think the conventional wisdom is that she had a close call in the primary and was lucky to survive it.

micah: I don’t think we should talk about the primary that much … it’s over, and has been for a while.

harry: FIGHT FIGHT FIGHT.

clare.malone: I was under the impression we were talking about the general the whole time.

micah: Also, we already wrote about how the Democratic primary wasn’t all that close.

natesilver: Just wanted to get Sanders fans mad at us one more time, for old time’s sake.

clare.malone: Clinton is lucky in the general election in the sense that if she were running against a regular Republican candidate, she would be in the news a whooooole lot more for her email foibles. As it stands, she gets off the hook a lot because Trump has some campaign shakeup or does something out of the blue to take attention away from Clinton. And that’s obvious, but I think it’s a huge factor this year. She can disappear for a couple days at a time.

micah: Totally agree. I think the email thing is the clearest example of her not running a good campaign. They’ve handled that very poorly.

natesilver: You mean Clinton handled her emails poorly (agree) or the campaign’s handled the issue poorly (not as clear)?

micah: Both.

clare.malone: I think they have handled it poorly, yeah. The campaign hasn’t helped her hone her answer about the emails, or to be apologetic in any genuine way. I just think it’s a failure of crisis management in a major way.

micah: Clare and I are on the same page (the correct page). What’s the argument that they haven’t totally botched the email issue?

natesilver: I think she was going to get shit about it either way. “It’s not the crime, it’s the cover-up” is one of those political clichés that political reporters think to be true, but which actually isn’t true a lot of the time. And it’s partly because they can judge the campaign’s handling of a problem as sort of a performance, whereas the underlying issue requires deeper reporting and analysis.

clare.malone: Uh, I think the deeper analysis concluded that her activities were irresponsible and unethical?

harry: Most voters think Clinton has not been honest with how she handled the email situation. In fact, 64 percent in the recent Monmouth poll said so. That speaks to her handling of the controversy more than her handling of the emails. And just 38 percent in the most recent ABC News/Washington Post polls say Clinton is honest and trustworthy overall.

natesilver: But I’m saying that her sins were committed as secretary of state, and alternate ways of handling it on the campaign might not have mattered very much. There are also lots of other Clinton “trust” issues, like the Clinton Foundation and her paid speeches and — for her Republican critics — Benghazi.

micah: Even so, I still think the fact that her answers on the email issue haven’t really improved in a year — whether improved answers would make a difference or not — goes in the “not a great campaign” category.

natesilver: But what’s the “right” answer on email?

micah: “I messed up. It won’t happen again.” FULL STOP

natesilver: Micah, I watched the Anthony Weiner documentary the other day. And when he began running for New York City mayor, that’s exactly what he did. I messed up, full stop.

And by the way, he was running close to the top of the polls for a while.

micah: And wasn’t it working until new transgressions came to light?

natesilver: Right. And the contrition was a disaster, because the story was ongoing.

micah: OK, Nate, but then that precedent only holds if Clinton is currently doing something similar, emailwise.

harry: Let this be a lesson: Don’t take pictures of your naked self and send it to other folks. This goes for all of you out there.

clare.malone: I think the “right” answer for other candidates if they had an email problem would be really opening up on a different front — trying to sell a narrative of the candidate as a full human being, a person with a great personal story. A person who can command a crowd. Clinton can’t or won’t do that. She has basically made it clear that we only get to know her so much, and in some ways I think that inaccessibility (which is central to her persona now) has made the emails worse than it would for another candidate.

harry: Could it be argued that Clinton has run a good campaign, but isn’t a good candidate?

clare.malone: Yes.

micah: Interesting. I think that’s right, although — and this is my personal opinion — I think the things that make Clinton a “bad” candidate are mostly not within her control and largely gender-related (i.e., sexism).

clare.malone: Some for sure, but not all.

micah: Not all. But the hatred of Clinton is hard to explain otherwise.

clare.malone: I think her issues with the press date back to her husband and obviously, I suppose, a lot of her mistrust of the media centers around her feeling buffeted around by them, stereotyped as nontraditional … OK, maybe the root causes are a lot of gender stuff!

But she still decided to develop calluses rather than try to start anew with her persona. She sort of retreated into this world with a lot of locks and keys around her, and voters are smart enough to know that she’s been living this way for a long time.

natesilver: It might be worth considering Clinton’s poor ratings among millennials. If they liked her as much as they liked President Obama, she’d be running — I haven’t run the math so I’m estimating — at about break-even favorables instead of being more than 10 points underwater.

harry: Here’s what I know: Clinton currently has a favorable rating around 43 percent, and she is winning rather easily. That, to me, indicates that she’s doing something right. Part of that may be that Republicans selected Trump. Fine, OK. But she also raised a ton of money and has spent it wisely.

micah: So, would she be losing against another, more traditional Republican?

clare.malone: I mean, I think that was John Kasich’s whole argument for his existence in the race, right? That he was a general election threat, even though he couldn’t get out of the primary cage match alive.

harry: I think Kasich would be doing better than Trump, but Kasich too would have faced an onslaught of ads from the Clinton campaign.

natesilver: It’s impossible to prove, but yeah, I tend to think a race against a different, non-insane Republican would be a toss-up, although Kasich would have a nice advantage from having a big home-state effect in Ohio. Not all that many politicians have been through the ringer and survived with above-average favorability ratings these days. Kasich — like Sanders — benefited from sort of being this hypothetical candidate that nobody really had reason to scrutinize too closely.

clare.malone: Yes … I mean, to go back to our initial question, is she more lucky or good, the woman is currently leading the polls by a pretty steep margin; this is just a thing that wouldn’t have happened without Trump. This election has taken on contours that default to her advantage. She would be fighting really hard right now if it were a moderate Republican.

micah: Here’s another test: If Clinton vs. a generic Republican would be a toss-up, how about Trump vs. a generic Democrat?

natesilver: I guess I’d just say that Clinton won her primary big and now it looks like she’s going to win the general election, possibly also big. And, by the way, she’s going to be the first woman in American history to accomplish this in 240 years of the republic.

Obviously, she’s helped by the presence of Trump in the race. But I think the burden of proof is sort of on the lucky side.

clare.malone: I mean, she’s been good in the sense that, she’s qualified as hell. She’s just not a natural at campaigns.

micah: Yeah, obviously you can’t become president without being both lucky and good. That’s doubly or triply true for a woman trying to become the first female president.

clare.malone: Right.

micah: So, is this whole line of questioning sexist?

clare.malone: No. I don’t think anything is that black and white — when we say “is she good or lucky?” we’re talking about as a campaigner, as a political character. She’s obviously qualified.

BUT

I think we do have to acknowledge that when we say that Clinton is a historically disliked candidate (as measured by her unfavorables), she’s historically disliked in no small part because she’s lived a life antithetical to what a lot of Americans expect out of women. She’s not here to make you feel good and cuddly — she’s a woman of ambition who makes a lot of men feel emasculated, quite frankly (and to be clear, that’s their problem).

She’s also, in the past, not done a great job about waving the flag for choice feminism (remember that comment about baking cookies? You can be damn sure she offended a lot of stay-at-home moms … and their now-adult kids who were grateful for those moms). In short (though this is long for a chat) Hillary Clinton is not a woman who’s lived a life amenable to anything other than a complicated retelling of it; she’s lived a lot of shades of gray. But I think she’s lived a remarkable life — as have most candidates for president, regardless of gender.

But yes, to distill it: A lot of people hate women who have their shit together.

harry: I want to build off of what Clare said. Let’s take a look at Clinton’s candidate record: She cleared the Democratic field in her 2000 Senate race. Then Rudy Giuliani, citing health concerns, left the race after Clinton took a lead in the polls. Clinton then beat Rick Lazio by double digits. In 2006, Clinton won re-election by more than 35 percentage points. Then in 2008, she lost to perhaps the greatest politician of our era in Obama and actually got more votes than he did. This year, Clinton cleared the Democratic field and now she leads Trump by a considerable margin in the key swing states. To me, she’s either the luckiest politician since Chris Coons or she may actually be pretty damn good.

natesilver: I think she’s definitely very fortunate to have drawn Trump as an opponent, but otherwise I think she’s underrated as a politician. And I don’t necessarily buy that she’d be losing to Kasich or Marco Rubio or Mitt Romney right now, although that’s a whole ’nother discussion.

Obviously if she loses to Trump, though — still a chance of that — we’ll need to revisit our assessment. And same probably goes if she only wins by a point or two.

Related:

Election Update: Leave The LA Times Poll Alone!

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

I’m tired of hearing about the USC Dornsife/Los Angeles Times tracking poll.

I’m tired of hearing about the poll from Donald Trump fans such as Reince Priebus, Matt Drudge and Donald Trump himself.1 They frequently cherry-pick that poll because it consistently shows much better results for Trump than the other surveys. As of Tuesday morning, for example, the poll showed the race as virtually tied — Hillary Clinton 44.2 percent, Trump 44.0 percent — even when the national poll average has Clinton up by about 6 percentage points instead.

This has been a fairly consistent difference between this poll and most others. Take the LA Times poll, add 6 points to Clinton, and you usually wind up with something close to the FiveThirtyEight or RealClearPolitics national polling average. What’s the source of the LA Times poll’s Trump lean? There are good “explainers” from The New York Times’s Nate Cohn and Huffington Post Pollster’s David Rothschild. Long story short: The poll’s results are weighted based on how people said they voted in 2012. That’s probably a mistake, because people often misstate or misremember their vote from previous elections.2

The poll does some other things differently also, some of which I like. For instance, it allows people to assign themselves a probability of voting for either candidate instead of saying they’re 100 percent sure. And the poll surveys the same panel of roughly 3,000 people over and over instead of recruiting new respondents. That creates a more stable baseline and can therefore be a good way to detect trends in voter preferences, although it also means that if the panel happened to be more Trump-leaning or Clinton-leaning than the population as a whole, you’d be stuck with it for the rest of the year.

But I’m also tired of hearing from the LA Times poll’s critics. I’m not a fan of litigating individual polls, for several reasons. First, in my experience, these critiques tend to involve their own form of cherry-picking. Clinton fans will pick apart the LA Times poll and find a few things wanting — in this case, with good reason (in my opinion). But they’ll give a free pass to a poll like this one that shows Clinton ahead by 16 percentage points in Virginia, even though it’s also something of an outlier. You can almost always find something “wrong” with a poll you don’t like, even if you might have approved of its methodology before you saw its result.

It’s probably also harmful for the profession as a whole when poll-watchers are constantly trying to browbeat “outlier” polls into submission. That can encourage herding — pollsters rallying around a narrow consensus to avoid sticking out — which is bad news, since herding reduces the benefit of averaging polls and makes them less accurate overall.

Furthermore, the trend from LA Times poll still provides useful information, even if the level is off. Before the conventions, the poll had Trump ahead by an average of 2 or 3 percentage points. Trump then got a modest convention bounce in the poll and pulled ahead by 6 or 7 percentage points. But Clinton got a bigger bounce, and she’s been ahead by an average of 1 or 2 percentage points in the poll since the conventions, although it’s been a bit less than that recently, with Trump narrowly leading the poll at times. All of this follows the trend from other polls almost perfectly, as long as you remember that you have to shift things to Clinton by about 6 points.

And that’s pretty much what FiveThirtyEight’s forecast models do through their house effects adjustment. A pollster’s house effect is a persistent lean toward one candidate or another, relative to other polls. House effects are not the same thing as statistical bias — how the poll compares against actual results — which can be assessed only after the fact. Nor do they necessarily indicate partisan bias. For example, Public Policy Polling, a Democratic polling firm, has a very mild pro-Trump house effect this year.

Calculating house effects is simple, in principle — you compare a poll’s results against the average of other surveys of the same states (treating national polls as their own “state”). In practice, there are a few challenges, which you can read more about in our methodology primer. One of the important ones is defining what the average is. In the case of FiveThirtyEight’s forecasts, the average is weighted based on our pollster ratings.

Put another way, the house effects adjustment seeks to determine what the best pollsters are saying and not just what the most prolific ones are saying. In 2012, that made a difference: the higher-quality pollsters generally projected better results for Obama than the lower-quality ones. This year, any such effects are very minor,3 and neither Trump nor Clinton benefits much from the house effects adjustment overall, although it can matter more in individual states. Polls in Nevada happen to be a Trump-leaning bunch, for instance, so the house effects adjustment slightly helps Clinton there.

Which polls have a big house effect?

In the midst of an election, I’m sometimes reluctant to fixate on the house effects for individual polling firms because I don’t necessarily want to imply that a poll with a strong house effect is wrong. A house effect is sometimes the sign of a problem and sometimes not; it’s hard to know for sure until after the election has taken place. I also don’t want to encourage herding. Instead, I’d rather pollsters stick with what they’re doing, even if they stand out a bit, than to change methodology in midstream, as at least one pollster (Ipsos/Reuters, which previously had a Clinton-leaning house effect) has already done.

Nonetheless, we talk about polls being Clinton-leaning or Trump-leaning all the time — so here’s some more detail about what that means. In the table below, you’ll see the house effects for all firms that have conducted at least 5 national polls this year or conducted surveys in at least 5 states. A couple of technical points: First, although it’s not shown in the table, our models calculate house effects for Clinton and Trump (and Gary Johnson) separately. A poll could be deemed to have both a pro-Clinton and a pro-Trump house effect if it tended to show few undecided voters, for instance. The numbers in the table are net figures. Also, you’ll see the house effects presented in two ways: as a raw figure and a discounted one. The raw figure reflects the magnitude of the house effect so far, while the discounted one is essentially what the model predicts the house effect will be going forward. The less data we have from a given firm, the more the raw house effect is discounted, since it may reflect statistical noise rather than anything systemic.

Here’s the data,4 with pollsters sorted into three major groupings based on their methodology: internet polls, automated polls (robopolls) and traditional live-caller telephone polls:

As you can see, the LA Times poll has the strongest house effect of any major pollster: a raw house effect of about 6 points in Trump’s direction, or a discounted one of about 4 points. Other Internet-based polls have been a mixed bag. The UPI/CVoter tracking poll has also been Trump-leaning. Ipsos/Reuters formerly had a strong Clinton-leaning house effect but, after a methodology change, it has pretty much gone away.5 Other prolific online polling firms, such as Morning Consult, YouGov and SurveyMonkey, don’t have strong house effects.

All the major automated polling firms6 have Trump-leaning house effects, ranging from moderate to severe, especially in the case of Rasmussen Reports and Gravis Marketing, which have longstanding GOP-leaning house effects. You might also notice that the various daily and weekly tracking polls, which are either online or automated polls, are mostly a Trump-leaning bunch. We haven’t had a lot of national polls lately other than the tracking polls, so that’s one reason our national polling average and others that adjust for house effects show a slightly wider margin for Clinton right now than those that don’t.

By contrast, traditional landline telephone polls have been Clinton-leaning as a group, although not uniformly. Quinnipiac University polls had a strong Trump lean earlier in the cycle, for example, although it has dissipated recently. It’s worth keeping these patterns in mind when you evaluate new surveys. Accounting for house effects, our model thinks a Quinnipiac poll showing Clinton up 8 in Colorado is roughly equivalent to a Marist College poll showing her up 12 there, since Marist’s polls have been Clinton-leaning while Quinnipiac’s have been Trump-leaning.

Related:

Clinton Is Spending A Lot More Money Than Trump

 Subscribe: iTunes |Download |RSS |Video

Donald Trump released his first TV advertisement of the general election last week. Meanwhile, Hillary Clinton and the main super PAC supporting her have already spent $65.8 million combined on TV ads in nine battleground states. The podcast crew talks about how much TV ads matter and how the campaigns are spending their money on the whole. The Trump campaign’s recent Federal Election Commission filings show that it has spent relatively little money on expanding its staff and ground game, leaving it to rely largely on the Republican National Committee.

The team also looks at Trump’s decision to hire a pollster, Kellyanne Conway, as his campaign manager.

You can stream or download the full episode above. You can also find us by searching “fivethirtyeight” in your favorite podcast app, or subscribe using the RSS feed. Check out all our other shows.

If you’re a fan of the elections podcast, leave us a rating and review on iTunes, which helps other people discover the show. Have a comment, want to suggest something for “good polling vs. bad polling” or want to ask a question? Get in touch by email, on Twitter or in the comments.