How To Read 2020 Polls Like A FiveThirtyEighter

Beware: Reading polls can be hazardous to your health. Symptoms include cherry-picking, overconfidence, falling for junky numbers and rushing to judgment. Thankfully, we have a cure. Building on an old checklist from former FiveThirtyEight political writer Harry Enten, here are some guidelines you should bear in mind when you’re interpreting political polling — in primary season and beyond.

What to watch for during the primaries

People who try to discredit early primary polls by pointing out that, say, Jeb Bush led early polls of the GOP field in 2016 are being disingenuous. Should these polls be treated with caution? Sure, but national primary polls conducted in the calendar year before the election are actually somewhat predictive of who the eventual nominee will be. Earlier this year, fellow FiveThirtyEight analyst Geoffrey Skelley looked at early primary polling since 1972 and found that candidates who polled better in the months before the primaries wound up doing better in the eventual primaries. In fact, those who averaged 35 percent or higher in the polls rarely lost the nomination.

High polling averages foreshadowed lots of primary votes

Candidates’ share of the national primary vote by average polling level in the first half of the year before the presidential primaries and polling average in the second half of that year, 1972-2016

First half Second half
Poll Avg. Share who became nominee Avg. Primary Vote share Share who became nominee Avg. Primary Vote share
35%+ 75% 57% 83% 57%
20%-35% 36 27 25 25
10%-20% 9 8 9 12
5%-10% 3 7 10 10
2%-5% 5 5 0 4
Under 2% 1 2 1 1

We included everyone we had polling data for, no matter how likely or unlikely they were to run. If a candidate didn’t run or dropped out before voting began, they were counted as winning zero percent of the primary vote.

Sources: POLLS, CQ Roll call, DAVE LEIP’s atlas of u.s. presidential elections

And if we go one step further and account for a candidate’s level of name recognition, early national primary polls were even more telling of who might win the nomination. As you can see in the chart below, a low-name-recognition candidate whose polling average climbed past 10 percent in the first half of the year before the primaries had at least a 1 in 4 shot at winning, which actually put them ahead of a high-name-recognition candidate polling at the same level.

This is why we believe that national primary polls are useful (even this far out) despite the fact that they are technically measuring an election that will never happen — we don’t hold a national primary. For this reason, early-state polls are important, too, especially if they look different from national polls. History is littered with examples of national underdogs who pulled off surprising wins in Iowa or New Hampshire, then rode the momentum all the way to the nomination. And according to analysis from RealClearPolitics, shortly after Thanksgiving is historically when polls of Iowa and New Hampshire start to come into closer alignment with the eventual results.

But don’t put too much faith in early primary polls (or even late ones — they have a much higher error, on average, than general-election polls). Voters’ preferences are much more fluid in primaries than they are in general elections, in large part because partisanship, a reliable cue in general elections, is removed from the equation. And voters may vacillate between the multiple candidates they like and even change their mind at the last minute, perhaps in an effort to vote tactically (i.e., vote for their second choice because that candidate has a better chance of beating a third candidate whom the voter likes less than their first or second choice).

On the flip side, early general-election polls are pretty much worthless. They are hypothetical match-ups between candidates who haven’t had a chance to make their case to the public, who haven’t had to withstand tough attacks and who still aren’t on many Americans’ radar. And these polls aren’t terribly predictive of the eventual result either. From 1944 to 2012, polls that tested the eventual Democratic and Republican nominees about a year before the election (specifically, in November and December of the previous year) missed the final margin by almost 11 percentage points, on average — though it’s worth noting that they were more accurate in 2016, missing by around 3 points.1

Early general-election polls are usually way off the mark

Average error in general-election polls that tested the two eventual nominees in November and December of the year before the election, for presidential elections from 1944 to 2012

Polling Accuracy A Year Before The Election
Election Average GOP Poll Lead GOP Election Margin Absolute Error
1944 -14.0 -7.5 6.5
1948 -3.8 -4.5 0.7
1956 +22.0 +15.4 6.6
1960 +3.0 -0.2 3.2
1964 -50.3 -22.6 27.7
1980 -15.5 +9.7 25.2
1984 +7.2 +18.2 11.0
1988 +18.0 +7.7 10.3
1992 +21.0 -5.6 26.1
1996 -13.0 -8.5 4.5
2000 +11.9 -0.5 12.4
2004 +8.7 +2.5 6.2
2008 -0.3 -7.3 6.9
2012 -2.8 -3.9 1.0
Average 10.6

No odd-year November-December polling was available for the 1952, 1968, 1972 and 1976 elections.

Source: Roper Center for Public Opinion Research

In other words, at this stage in the cycle, primary polls can be useful but are by no means infallible, while general-election polls can safely be ignored. That may seem frustrating, but just remember that pollsters aren’t trying to make predictions; they’re simply trying to capture an accurate snapshot of public opinion at a given moment in time.

What to keep in mind generally

There are some guidelines you should remember at any time of the year, however. First, some pollsters are more accurate than others. We consider the gold standard of polling methodology to be pollsters that use live people (as opposed to robocalls) to conduct interviews over the phone, that call cell phones as well as landlines and that participate in the American Association for Public Opinion Research’s Transparency Initiative or the Roper Center for Public Opinion Research archive. That said, the polling industry is changing; there are some good online pollsters too. You can use FiveThirtyEight’s Pollster Ratings to check what methodology each pollster uses and how good its track record has been. (And if a pollster doesn’t show up in our Pollster Ratings, that might be a red flag.)

Another reason to pay attention to the pollster is for comparison purposes. Because pollsters sometimes have consistent house effects (their polls overestimate the same party over and over), it can be tricky to compare results from different pollsters. (For this reason, FiveThirtyEight’s models adjust polls to account for house effects.) When looking for trends in the data over time, it’s better to compare a poll to previous surveys done by that same pollster. Otherwise, what looks like a rise or fall in the numbers could just be the result of a different methodological decision or, especially for non-horse-race questions, the way the question is worded. The order in which questions are asked can matter too; for example, asking a bunch of questions about health care and then asking for whom respondents would vote might bias them to pick the candidate they think is best on health care.

In addition, note who is being polled and what the margin of error is. Polls conducted among likely voters are the best approximation of who will actually cast a ballot, although when you’re still several months away from an election, polls of registered voters are much more common, and that’s fine. For non-electoral public opinion questions, like the president’s approval rating, many polls use a sample that will try to match the demographic profile of all adults in the U.S., and that’s fine, too. As for margin of error … just remember that it exists! For example, if a poll of the 2018 Florida governor race showed former Tallahassee Mayor Andrew Gillum ahead of former Rep. Ron DeSantis 47 percent to 46 percent with a margin of error of plus or minus 4 points, you’d want to keep in mind that DeSantis may actually have been leading at the time. Remember, too, that the margin of error applies to each candidate’s polling number, not to the difference between the candidates. So if both numbers are off by the margin of error, the difference between them could be off by twice as much. In this case, that could mean Gillum dropping to 43 percent and DeSantis jumping up to 50 percent, going from a 1-point deficit to a 7-point lead.

Sample size is important too — a smaller sample means a larger margin of error — but good polling is expensive, so the best pollsters may wind up with smaller samples. And that’s OK. As long as you heed the margin of error, a poll with a sample size of, say, 300 isn’t inherently untrustworthy. That said, don’t dive too much into one poll’s crosstabs — that’s where sample sizes do get unacceptably small and margins of error get unacceptably big. This is one reason not to trust commentators who try to “unskew” a poll by tinkering with its demographic breakdown, or who say that a poll’s results among, say, black voters are unbelievable and therefore the whole poll is too. These people are usually trying to manufacture better results for their side, anyway.

Speaking of which, consider the motive of whoever is sharing the survey. Polls sponsored by a candidate or interest group will probably be overly favorable to their cause. You should be especially suspicious of internal polls that lack details on how they were conducted (e.g., when they were conducted, who was polled, their sample size and their pollster). If you get your news from a partisan outlet, it may also selectively cover only polls that are good for its side. And even the mainstream media might be inclined to overhype a poll as “shocking” or a margin as “razor-thin” because it makes for a better headline.

Next, beware of polls that have drastically different results from all the others. They often turn out to be outliers — although not always (every new trend starts with one poll), which is why you shouldn’t throw them out either. Instead, just use a polling average, which aggregates multiple polls and helps you put the outlier into proper context. We at FiveThirtyEight use averages for that very reason.

And even if a new trend does emerge, wait a bit before declaring it the new normal. Big events — candidate announcements, debates, conventions — can have dramatic effects on the polls, but they are often fleeting.

Finally, come to terms with the fact that polls won’t perfectly predict the final results. Polls are a lot more accurate than people sometimes give them credit for, but polling error is real. Since 1998, polls conducted within a few weeks of the election have missed by an average of 3-10 points, depending on the type of campaign. So trust the polls, but hold onto some uncertainty right up until the moment election results start rolling in.

Where Kamala Harris’s New Voters Came From

Last week, FiveThirtyEight partnered with the survey firm Morning Consult to poll how Democratic voters’ opinions changed as a result of last week’s two-night presidential primary debate. Before the debate, Morning Consult asked thousands of likely Democratic voters questions such as whom they supported and whether they had favorable or unfavorable views of each candidate; then, they asked the same voters the same questions after the debate was over, as well as who the voters thought performed the best.1 The poll yielded some fascinating findings, the toplines of which you can see at our “Who Won The First Democratic Debates?” page, and the details of which we’ll be writing about here. (Note: This, of course, is just one poll — others have been released since the debate.)

We’ll start simple: How did the debate (or coverage of the debate — as we shall see, not everyone in our poll actually watched it) change people’s prospective primary votes?

Roughly speaking, before the debate, our poll showed a clear front-runner in Joe Biden (35.0 percent support), a clear second tier in Bernie Sanders and Elizabeth Warren (both in the teens) and a single-digit bottom tier led by Kamala Harris.2 After the debate, Biden was still alone in the first tier, albeit much weakened. And Harris had joined Sanders and Warren right in the thick of the second tier. Everyone else, led by Pete Buttigieg at 5.3 percent, remained in the third tier.

Harris more than doubled her support; Biden took a hit

Share of support for candidates before and after the debate

Candidate Pre-Debate Post-Debate Change
Joe Biden 35.0% 28.2% -6.8
Bernie Sanders 19.2 18.3 -0.9
Kamala Harris 6.9 16.3 +9.4
Elizabeth Warren 13.2 14.1 +0.8
Pete Buttigieg 6.6 5.3 -1.3
Beto O’Rourke 4.1 2.7 -1.4
Cory Booker 2.4 2.3 -0.1
Andrew Yang 1.5 1.6 +0.1
Julián Castro 0.4 1.5 +1.1
Tulsi Gabbard 0.6 0.7 +0.1
Kirsten Gillibrand 0.8 0.6 -0.1
Amy Klobuchar 0.5 0.6 +0.1
John Hickenlooper 0.4 0.3 -0.1
Michael Bennet 0.2 0.3 +0.1
Tim Ryan 0.2 0.3 +0.0
John Delaney 0.3 0.2 -0.1
Marianne Williamson 0.2 0.2 +0.0
Bill de Blasio 0.0 0.2 +0.1
Jay Inslee 0.4 0.1 -0.2
Eric Swalwell 0.1 0.1 +0.0
Don’t know/no opinion 6.5 5.9 -0.6
Someone else 0.5 0.3 -0.2

Numbers may not add up due to rounding

Source: Morning Consult

In other words, perhaps because of how Harris and Biden’s exchange on race seemed to dominate impressions of the debate, Harris and Biden were the only two candidates whose vote shares changed significantly in the wake of the debate. But it’s not as simple as a big chunk of voters switching from Biden to Harris. Here’s whom people who supported Biden before the debate said they supported after the debate:

About one in 10 Biden supporters switched to Harris after the debate. That made Harris the biggest single beneficiary of Biden’s loss of support. But about twice as many people — about one in five Biden supporters — switched from Biden to a non-Harris candidate or became undecided. And as a reminder that debates usually don’t turn elections completely on their head, more than two-thirds of Biden supporters stuck with him after the debate.

Those voters who switched from Biden to Harris may not have represented a big share of Biden’s support, but they made a big difference to Harris. Almost a quarter of Harris’s post-debate supporters were Biden converts — by far her biggest single source of added support. However, she also picked up support — more than two-fifths of her post-debate support — from former backers of Warren, Buttigieg, Sanders and several other candidates. Only about a third of her post-debate support was also with her before the debate.

We were also curious about how reaction to the debate might have changed minds independent of the debate itself, so we took a look at how candidate support differed between respondents who watched the debate3 and respondents who didn’t.4 Unsurprisingly, but importantly for our conception of how and why debates matter, those who did not tune in were less likely to switch their vote. For example, Harris gained just 3.9 points among non-watchers, but her support increased by an impressive 13.3 points among those who watched her performance firsthand. The discrepancy was not as stark for Biden. Although he lost more support among debate watchers than among non-watchers, he still dropped 5.9 points among those who missed the debate — the biggest change of heart that non-watchers had.

How did debate watchers and non-watchers differ?

Share of debate watchers and non-watchers who said they would vote for each candidate before and after the debate

Debate watchers Non-watchers
Candidate Before After Change Before After Change
Joe Biden 35.8% 28.0% -7.8 35.2% 29.2% -5.9
Kamala Harris 7.7 21.0 +13.3 5.7 9.6 +3.9
Bernie Sanders 17.8 16.5 -1.3 20.0 21.4 +1.4
Elizabeth Warren 14.7 14.9 +0.2 11.3 12.7 +1.4
Pete Buttigieg 7.6 6.0 -1.6 4.3 3.6 -0.7
Julián Castro 0.6 2.2 +1.7 0.2 0.4 +0.2
Cory Booker 2.6 2.0 -0.5 2.5 2.1 -0.4
Beto O’Rourke 3.9 1.8 -2.1 4.5 3.9 -0.5
Andrew Yang 1.4 1.5 +0.2 1.2 1.3 +0.2
Amy Klobuchar 0.6 0.9 +0.3 0.4 0.3 -0.1
Tulsi Gabbard 0.5 0.6 +0.0 0.8 0.7 -0.2
Kirsten Gillibrand 0.7 0.5 -0.2 0.7 0.6 -0.1
Tim Ryan 0.3 0.3 +0.0 0.2 0.1 -0.1
Marianne Williamson 0.3 0.3 +0.0 0.0 0.1 +0.0
John Delaney 0.2 0.2 +0.0 0.4 0.1 -0.2
John Hickenlooper 0.3 0.2 -0.1 0.7 0.4 -0.4
Eric Swalwell 0.1 0.1 +0.1 0.2 0.0 +0.0
Michael Bennet 0.1 0.1 +0.1 0.4 0.7 +0.3
Jay Inslee 0.3 0.1 -0.2 0.4 0.3 -0.2
Bill de Blasio 0.1 0.0 +0.0 0.0 0.1 +0.0
Don’t know/no opinion 4.1 2.6 -1.4 10.2 11.8 +1.7
Someone else 0.5 0.1 -0.4 0.7 0.4 -0.3

Source: Morning Consult

This could suggest one of three things (or two of three, or maybe even all three). First, hearing about a debate from a secondary source is less likely to change voters’ minds than something they see with their own eyes. Second, those secondary sources — be they news coverage, social media reaction, word of mouth, etc. — have communicated a fairly superficial “Biden lost” and, to a lesser extent, “Harris won” message. This second theory is further supported by the fact that other candidates, such as Julián Castro and Beto O’Rourke, experienced small gains and losses in support among debate watchers, but virtually no change among non-watchers. Third, debate-watchers and non-watchers might be different in ways we’re not capturing here. (For example, the latter may be less politically engaged.) The different effects therefore also may be due to these underlying differences between the two samples.

But the horse-race question alone understates the extent to which voters’ opinions of candidates shifted, as Castro and O’Rourke can attest. Several candidates other than Biden and Harris saw significant changes in their favorable or unfavorable ratings. (That’s important because changes in how voters feel about the candidates can lead to changes in vote preference later on.) In fact, Castro and O’Rourke experienced the biggest changes in net favorability (favorable rating minus unfavorable rating) of any of the candidates — perhaps as a byproduct of their disagreement over immigration on the first night of the debate. Castro’s ratings improved by 15.5 points, and O’Rourke’s fell by 11.6 points; essentially, they switched places.

Other than Castro and O’Rourke (and Harris and Biden), Warren and author Marianne Williamson also saw meaningful increases and decreases, respectively, in their net favorability ratings. In Williamson’s case, it was enough to have more people view her negatively than positively, which is a rare (and unenviable) position to be in among members of your own party.

Overall, the debate was the biggest deal for two candidates — Biden and Harris — above all and may have bunched the top of the field closer together. And it may have downstream implications for Castro and Warren (in a good way) as well as O’Rourke and Williamson (in a bad way). While the debate did not turn the race totally on its head, it certainly appears to have been one of the few opinion-moving events of the campaign so far.

Should We Take These Early General Election Polls Seriously? $#!% No!

Welcome to Pollapalooza, our weekly polling roundup.

Poll of the week

A lot of people are talking about hypothetical 2020 general election polls, including the president of the United States.

A national Quinnipiac University poll released this week, for example, showed Joe Biden with a 53-40 lead against President Trump. But it wasn’t just Biden — all the Democratic contenders Quinnipiac included in matchups with Trump were significantly ahead of the president: Bernie Sanders 51-42, Kamala Harris 49-41, Elizabeth Warren 49-42, Cory Booker 47-42 and Pete Buttigieg 47-42. Meanwhile, in an interview with ABC News’s George Stephanopoulos, Trump dismissed any polls that showed him trailing the Democrats.12 “No, my polls show that I’m winning everywhere,” Trump said.

So, just how seriously should we take hypothetical general election polls more than a year out and before the Democratic nominee has been selected?

Not seriously.

In the runup to the 2016 presidential election, this same question came up, and FiveThirtyEight analyzed general election polls from 1944 to 2012 that tested the eventual nominees and were conducted in the last two months of the year before the election (so for 2012, that would be November and December of 2011). On average, these polls missed the final result by 11 percentage points.13

Polling Accuracy A Year Before The Election
Election Average GOP Poll Lead GOP Election Margin Absolute Error
1964 -50.3 -22.6 27.7
1992 +21.0 -5.6 26.1
1980 -15.5 +9.7 25.2
2000 +11.9 -0.5 12.4
1984 +7.2 +18.2 11.0
1988 +18.0 +7.7 10.3
2008 -0.3 -7.3 6.9
1956 +22.0 +15.4 6.6
1944 -14.0 -7.5 6.5
2004 +8.7 +2.5 6.2
1996 -13.0 -8.5 4.5
1960 +3.0 -0.2 3.2
2012 -2.8 -3.9 1.0
1948 -3.8 -4.5 0.7
Average 10.6

The last presidential election featured one of the more accurate sets of early polls for this point in the cycle: Hillary Clinton led Donald Trump 46.2 percent to 41.2 percent in an average of all polls conducted in November and December 2015, missing the eventual national popular vote margin by about 3 points.14 (The actual result was Clinton 48.0 percent, Trump 46.0 percent.)

But that’s more the exception than the rule, as the table above shows. And remember, these are polls conducted at least five months later in the cycle than where we are now. Jump back to roughly this point in the 2016 cycle, for example, and Clinton was ahead of all eight of her hypothetical GOP opponents in a May 2015 Quinnipiac poll, with a whopping 50-32 advantage over Trump.

There’s just soooo much that can and will change. To take the two biggest ones: Democrats have an entire primary to get through and a nominee to pick. And we really have no idea what the economy will look like by Election Day 2020.

OK, maybe you’re not shocked that very early general election polling isn’t particularly predictive. Do these numbers tell us anything at all? Maybe. I think they hint at two things.

First, the Republican Party under Trump has had a ceiling so far — and it’s south of 50 percent of American voters. The president won 46 percent of the vote in 2016. House Republicans won 45 percent of the national House vote in 2018. Trump’s approval rating for the past two years has been between 37 percent and 43 percent. I doubt that Trump will get just 42 percent of the national vote (and most other national polls pitting him against the Democratic candidates have him in the mid-40s). At the same time, it’s pretty hard right now to see Trump getting the majority of the electorate behind him.

That doesn’t mean he can’t win. But Trump may need, like in 2016, to overperform in the Electoral College relative to the popular vote and for third-party candidates (perhaps Justin Amash or Howard Schultz) to take some of the anti-Trump vote from the Democratic nominee.

Secondly, this poll is more evidence that Trump should probably spend less time courting his political base and more time appealing to voters outside of it. He’s getting more than 90 percent of the Republican vote in head-to-head matchups against these Democratic candidates (even against Biden), according to the Quinnipiac survey. And that’s consistent with other data. Gallup polling suggests that Trump’s approval rating among self-identified Republicans is around 90 percent. In the 2018 midterms, exit polls suggested that about 94 percent of self-identified Republicans backed the GOP House candidate, as did 88 percent of those who approve of President Trump.

Trump’s real political problem is self-identified independents and voters who don’t love him or hate him. In the 2018 midterms, independents broke heavily for the Democrats in U.S. House elections (+12), as did voters who “somewhat” disapproved of the president (+29), according to exit polls. In this Quinnipiac survey, all the Democratic candidates had double digit leads over Trump among independents, and those are the numbers that should worry the president and his political team.

Other polling bites

  • 63 percent of Americans favor allowing transgender people to serve in the U.S. military, according to a Public Religion Research Institute survey released this week.
  • According to a new Monmouth University poll, Biden leads among Democratic presidential candidates with 36 percent of the vote in Nevada, which votes third in the Democratic nomination process. Warren (19 percent) and Sanders (13 percent) are the only other two candidates polling in double digits.
  • 79 percent of Iowa Democrats said that to get their vote, a Democratic presidential candidate must support “a woman’s right to abortion,” according to a recently released by Selzer & Co. for the Des Moines Register/CNN/Mediacom Iowa poll; 12 percent said that position was not a must-have.
  • In the same poll, 23 percent of Iowa Democrats said that a candidate must support offering all Americans free tuition to a public four-year college; 15 percent said they would oppose a candidate who took that position.
  • 33 percent of Americans favor Congress starting impeachment proceedings against Trump, compared with 61 percent who do not, according to a Quinnipiac survey released this week; 44 percent believe he “deserves to be impeached,” while 50 percent do not.
  • 72 percent of Americans oppose a proposal to increase the salaries of members of Congress by $4,500, compared with 14 percent who support the idea, according to a Politico/Morning Consult poll released this week. The proposal would have meant members’ salaries were $178,500 per year.
  • 49 percent of Americans support the Hyde Amendment, which bars federal funding of abortions, compared with 32 percent who oppose it, according to the Politico/Morning Consult survey.

Trump approval

According to FiveThirtyEight’s presidential approval tracker, 42.3 percent of Americans approve of the job Trump is doing as president, while 52.9 percent disapprove (a net approval rating of -10.6 points). At this time last week, 41.5 percent approved and 53.6 percent disapproved (for a net approval rating of -12.1 points). One month ago, Trump had an approval rating of 42.4 percent and a disapproval rating of 52.5 percent (for a net approval rating of -10.1 points).

Congressional generic ballot

According to FiveThirtyEight’s congressional generic ballot tracker — which returned this week! — about 46.1 percent of Americans would vote for a Democratic candidate for Congress, compared with 39.9 percent who would choose a Republican.

Check out the polls we’ve been collecting ahead of the 2020 elections, including all the Democratic primary polls.

Most Americans Agree That WWII Was Justified. Recent Conflicts Are More Divisive.

Welcome to Pollapalooza, our weekly polling roundup.

Poll of the week

On Thursday, Americans both at home and abroad commemorated the 75th anniversary of D-Day, the seaborne invasion of France that helped usher in an Allied victory in World War II. The widespread observance of the occasion was hardly surprising; most Americans see World War II as a proud moment in U.S. history. A full 66 percent of Americans said they thought the U.S. role in World War II was “completely” or “somewhat” justified, according to a YouGov poll out this week, while just 14 percent said it was “not very” or “not at all” justified.

That makes World War II the most popular U.S. military engagement of the eight that the poll asked about, as measured by the net share of people who said it was justified (+52). The American Revolution came in second (+47); 61 percent said it was justified and 14 percent said it was not justified. (Yes, apparently after nearly 250 years, there are still some Loyalists out there!) Americans’ perceptions of more recent wars are more complicated. For some of those conflicts, a greater share of people said they see them as unjustified than see them as justified. One of those is the Vietnam War, which 55 percent said was not justified and 22 percent said was justified.

The Civil War is the third-most-supported war in the poll, with 54 percent of Americans saying it was justified and 22 percent saying it wasn’t. Perhaps surprisingly, about the same share of Southerners said the war was unjustified as residents of other regions. Respondents from the South said the Civil War was justified by a 52-percent-to-23-percent margin. As anyone who has seen a Confederate flag in Maine can tell you, Confederate sympathies can be found in all corners of the country.

Similarly, there weren’t big partisan differences on the U.S.’s involvement in the Civil War. About the same share of Republicans (61 percent) as Democrats (57 percent) said they thought it was justified.

But there’s a catch. These numbers may have less to do with the historical reasons behind the war and more to do with Americans’ philosophies about going to war in general. In answer to a different question, 59 percent of Republicans told YouGov that there is “often” or “always” a justification for war, while 29 percent said there is “rarely” or “never” a justification. Among Democrats, those numbers were 21 percent and 66 percent, respectively. 10 Unsurprisingly, then, Republicans are more supportive on net than Democrats of every conflict YouGov asked about — but the gap between the parties is smallest on the Civil War.

More Republicans than Democrats think wars are justified

Share of each party that said a given conflict is “completely” or “somewhat” justified vs. “not very” or “not at all” justified, according to a May 21-22 poll

Democrats Republicans
Conflict Justified Not justified Net Justified Not justified Net Diff.
Afghanistan 26% 55% -29 56% 27% +29 R+58
First Gulf War 28 47 -19 58 20 +38 R+57
Korean War 30 40 -10 56 17 +39 R+49
Vietnam War 14 68 -54 37 44 -7 R+47
World War I 51 27 +24 68 14 +54 R+30
American Revolution 60 18 +42 74 7 +67 R+25
World War II 67 17 +50 77 9 +68 R+18
Civil War 57 24 +33 61 18 +43 R+10

Source: YouGov

Overall, Democrats tended to view older wars (World War II and earlier) as justified, bringing them in closer agreement with Republicans. After the Civil War, the parties are closest together on World War II and the American Revolution. But there is more partisan polarization over more recent wars. For example, Republicans said by a +29 net margin that the current military engagement in Afghanistan is justified, while Democrats said it was not justified by the same margin (-29). The Vietnam War is the only conflict in the poll that both parties said was not justified. But a large majority of Democrats felt very strongly about it while Republicans were divided, so there remains a significant gap between the parties.

Other polling bites

  • In a press conference on May 29, then-special counsel Robert Mueller said, “If we had had confidence that the president clearly did not commit a crime, we would have said so.” In a survey released this Monday, HuffPost and YouGov asked Americans who were aware of that statement whether they thought Mueller’s report had or had not cleared President Trump of any wrongdoing. Seventy-four percent of Trump voters said they thought it had. They were more likely to say that than Trump voters who were not aware of Mueller’s statement.
  • So far in 2019, several states have passed “fetal heartbeat” laws that ban abortions after fetal cardiac activity can be detected, which can be as early as six weeks into a pregnancy. A new USA Today/Ipsos poll found that Americans oppose these laws 55 percent to 45 percent. In addition, as Missouri may soon become the first state without an abortion clinic, Americans said 73 percent to 27 percent that they don’t want every abortion clinic in their state to close.
  • Morning Consult’s continuous poll of the Democratic presidential primary found that the share of Democrats who say “women’s issues” are their top priority rose from 6 percent to 14 percent in the month of May.
  • In a new poll from Monmouth University, 27 percent of Americans said someone in their household did not receive needed medical care in the past two years because of the cost. And 20 percent reported that the need to hold onto their health insurance plan had prevented them from trying to change jobs at least once in the past 10 years.
  • An Opinium poll in the United Kingdom has put the nascent Brexit Party in first place for U.K. parliamentary elections for the first time. When asked which party they would support, 26 percent said the Brexit Party, 22 percent said Labour, 17 said the Conservatives and 16 percent said the Liberal Democrats. Multiple other polls in the past couple of weeks have shown a similar four-way pileup at the top, a remarkable shift from the traditional two-party dominance of Labour and the Conservatives. It reflects how much Brexit has come to dominate U.K. politics — the Liberal Democrats and Brexit are explicitly pro- and anti-European Union, respectively, while Labour and Conservative opinion is muddled.

Trump approval

According to FiveThirtyEight’s presidential approval tracker, 41.5 percent of Americans approve of the job Trump is doing as president, while 53.6 percent disapprove (a net approval rating of -12.1 points). At this time last week, 41.2 percent approved and 54.2 percent disapproved (for a net approval rating of -13.0 points). One month ago, Trump had an approval rating of 42.7 percent and a disapproval rating of 52.4 percent (for a net approval rating of -9.7 points).

Check out all the polls we’ve been collecting ahead of the 2020 elections.

A Lot Of Americans Say They Don’t Want A President Who Is Over 70. Really?

Welcome to Pollapalooza, our weekly polling roundup.

Poll of the week

Gallup recently released new data on Americans’ willingness to vote for presidential candidates with certain traits. About 1,000 adults were asked3 whether they’d vote for a well-qualified candidate who was nominated by their party and was black, gay or had one of 10 other characteristics that are rarely or never seen in presidential nominees.

Almost all Americans said they’d be comfortable voting for a woman (94 percent), or a Catholic (95 percent), Hispanic (95 percent) or black (96 percent) candidate. But there are characteristics that big swaths of Americans said would be disqualifying — in particular being older than 70, being an atheist and being a socialist.

What types of candidates would Americans NOT vote for?

Share of respondents to an April survey who said they would not vote for a “generally well-qualified” presidential candidate from their own party if the candidate had each of the following characteristics

Democrats Independents Republicans Overall
Socialist 24% 48% 80% 51%
Atheist 28 33 56 39
Older than 70 35 37 37 37
Muslim 14 26 62 33
Younger than 40 21 28 34 28
Gay or lesbian 17 18 39 24
Evangelical Christian 27 20 6 18
Jewish 5 9 5 7
Woman 3 6 9 6
Catholic 4 6 3 5
Hispanic 3 3 8 5
Black 1 4 5 3

Source: Gallup

These results are fairly similar to what Gallup found when it previously asked this question, in 2015. There were a couple of interesting exceptions, however. Americans in 2019 said they were slightly more comfortable with a candidate who is an evangelical Christian (the share who said they’d vote for such a candidate rose from 73 percent in 2015 to 80 percent this year) or a Muslim (from 60 percent to 66 percent). Socialists, meanwhile, remained unpopular (47 percent in both 2015 and 2019).

So with Democrats obsessed with finding an “electable” candidate, does this mean that Bernie Sanders (who’s over 70 and identifies as a democratic socialist) and Joe Biden (who’s over 70) have big problems? Not so fast. So how seriously am I taking these numbers?

For the 2020 presidential election, I’m not taking them too seriously. Thirty-seven percent of Republicans said they would not back a GOP presidential candidate over the age of 70. Well … yep, President Trump was 70 on Election Day in 2016, and he’ll be 74 in 2020. I’ll bet that more than 63 percent of Republicans will vote for him — his job approval rating among GOP voters is currently in the 90s. In short, it’s important to remember that the survey question asks about categories of people, not individuals. The negative feelings that some Americans might have toward the idea of a gay or socialist presidential candidate, for example, might not apply to Pete Buttigieg or Sanders specifically.

On the other hand, these numbers could be understating some Americans’ resistance to certain characteristics. In particular, I’d view the numbers on ethnicity, race and gender skeptically. It could be true that virtually all Americans are comfortable with a black, female or Hispanic president, as the Gallup data implies. But I’d expect Americans who aren’t comfortable to be unlikely to express that view to a pollster. So I wouldn’t use this data to suggest that, say, Julian Castro wouldn’t run into electoral problems caused by racism or Elizabeth Warren because of sexism if either were the Democratic nominee.

In terms of which groups might face overt discrimation in the U.S., I’m taking these numbers more seriously. The results generally lined up with my expectations of which categories of people Americans are both somewhat wary of and willing to say so to another person.

Being a socialist is an expression of left-wing political views, so it’s natural and unsurprising that a lot of Americans, particularly Republicans, would openly oppose a socialist candidate. Similarly, it’s not surprising that some Americans wouldn’t want a president who is in her 70s as president (maybe they suspect that person wouldn’t have the energy for the job) or who is younger than 40 (a lack of experience). This is also a view that is perhaps not particularly controversial to express — columns suggesting that Biden (76) and Sanders (77) are too old to be running for president are published regularly.

What views about candidates are more controversial? Disqualifying people based on gender, race, ethnicity or sexuality. Again, I’d expect some Americans with negative attitudes toward certain religious groups, racial groups and sexual orientations not to admit that to a pollster.

Here’s where it gets interesting, however: The share of Americans who were willing to tell a pollster that they would not back an atheist, evangelical Christian, gay or Muslim presidential candidate was nonetheless fairly high. That lines up with how these four groups are treated in American culture — they face open, direct criticism based on their identities. (I don’t want to cast all parties as equal here — Republicans’ high level of opposition to an atheist or Muslim candidate jumps out.)

In terms of understanding the diversity of the Democratic Party, I’m taking these numbers very seriously. I’ve written that Biden is essentially the candidate of the un-woke Democrat (or maybe “less woke” is more accurate) and that those voters still represent a substantial bloc of the Democratic Party. This data is more evidence of that bloc’s existence. I was surprised that the share of Democrats who are uncomfortable with an evangelical Christian president was matched by about an equal share wary of a president who is an atheist or a socialist, since the Democratic Party is often characterized as becoming less religious and more liberal on economic issues. The share of Democrats who said they would not vote for a gay or Muslim candidate was also larger than I anticipated.

Other polling bites

  • 46 percent of likely Democratic primary voters in South Carolina say they would vote for Biden, according to a new Post and Courier/Change Research poll, with only two of his rivals reaching double digits. Sanders (15 percent) and Kamala Harris (10 percent) are far behind the former vice president, as is the rest of the 2020 Democratic field.
  • Biden leads in Pennsylvania too, with 39 percent of the vote, according to a new Quinnipiac University survey. The only other candidate in double digits was Sanders (13 percent).
  • The Quinnipiac survey also found Biden leading Trump 53 percent to 42 percent in Pennsylvania in a hypothetical general election matchup. Sanders also bested Trump (50-43).
  • In the Republican nomination contest, Trump leads former Massachusetts Gov. William Weld 72 percent to 12 percent in New Hampshire, according to a recent Monmouth University survey.
  • 61 percent of Americans support same-sex marriage, and 31 percent oppose it, according to a Pew Research Center survey. Support for same-sex marriage varied by party (75 percent of Democrats and Democratic-leaning independents, compared with 44 percent of Republicans and GOP-leaning independents). It varied by race (62 percent of white Americans, 58 percent of Hispanic Americans, 51 percent of black Americans). And it varied by religion (79 percent of those who are religiously unaffiliated, 66 percent of white mainline Protestants, 61 percent of Catholics, 29 percent of white evangelical Protestants).
  • 47 percent of registered voters rated the economy as “excellent” or “good,” according to a new Fox News poll.
  • Also from that Fox News poll: The share of voters who said Trump hasn’t been tough enough with North Korea is up to 50 percent; that number was 19 percent in September 2017.

Trump approval

According to FiveThirtyEight’s presidential approval tracker, 42.0 percent of Americans approve of the job Trump is doing as president, while 53.1 percent disapprove (a net approval rating of -11.1 points). At this time last week, 42.4 percent approved and 52.7 percent disapproved (for a net approval rating of -10.3 points). One month ago, Trump had an approval rating of 42.1 percent and a disapproval rating of 52.3 percent, for a net approval rating of -10.2 points.

Check out all the polls we’ve been collecting ahead of the 2020 elections.

Abolishing The Electoral College Used To Be A Bipartisan Position. Not Anymore.

Twice in the past five presidential elections, a Republican has won the presidency despite losing the popular vote. Now Democratic Sen. Brian Schatz of Hawaii has introduced a constitutional amendment to abolish the Electoral College and use the national popular vote to decide who becomes president. His proposal is among the latest efforts by Democrats and those on the left to push for structural changes to the American political system.

But Schatz’s amendment is sure to meet defeat in the Republican-controlled Senate. Today, attitudes toward the Electoral College are polarized by party, with Democrats far more likely to support a change and Republicans much more likely to defend the current system — but it wasn’t always like that.

While the controversial 2000 election was still being decided, Gallup found that 61 percent of Americans — including 73 percent of Democrats and 46 percent of Republicans5 — preferred amending the Constitution to elect the popular vote winner. Only 35 percent of respondents preferred the current system. The partisan gap widened even further after the 2016 election: A few weeks after President Trump won the presidency while losing the popular vote, Gallup found that 49 percent of Americans preferred changing to a popular vote system, compared to 47 percent who wanted to keep the Electoral College, with 81 percent of Democrats supporting a change compared to just 19 percent of Republicans.6 Even given some space after that heated election, there remains a major partisan gap in opinion over how to elect a president — Pew Research found in March 2018 that 75 percent of Democrats supported moving to a popular-vote system versus only 32 percent of Republicans.

But 50 years ago, moving on from the Electoral College had bipartisan support. In May 1968, 66 percent of American approved of the idea of amending the constitution to replace the Electoral College with a popular vote system, according to Gallup. And there was no partisan divide: 66 percent of Republicans and 64 percent of Democrats approved. Six months later, Republican Richard Nixon defeated Democrat Hubert Humphrey while only winning the popular vote by less than 1 percentage point, and a post-election Gallup survey found 80 percent of Americans approved of changing the electoral system. The bipartisan support among voters and the fact that the 1968 election nearly produced a split between the popular vote and the Electoral College7 explain why there was bipartisan support in Congress in 1969 for a constitutional amendment to elect presidents based on the popular vote. The House passed it 339 to 70, with more than 80 percent of each party’s voting members lending their support. But small-state senators from both parties filibustered the amendment and it never got an up-or-down vote in the upper chamber.

As long as one side feels disadvantaged by the Electoral College, it will be far more likely to push for a popular-vote system. Right now, that’s the Democrats. Reforming how the country elects presidents falls into the broad effort on the left to reform aspects of our electoral system, including voting access and how campaign finance works. But some who want reform believe abolishing the Electoral College should be a secondary goal. “There’s a bunch of stuff to do without amending the constitution that would have the end result of making institutions and elections more fair,” said David Faris, a political scientist at Roosevelt University, who recently argued in his book “It’s Time To Fight Dirty” that Democrats should be challenging the structural and legal boundaries of the American political system to better gain and hold power. Nonetheless, Faris sees discussion over the electoral system as a good thing in that it could soften up public opinion and make people more willing to consider alternatives to the status quo.

But we may not see a true shift in public opinion unless a Republican loses in the Electoral College while winning the popular vote. As FiveThirtyEight has argued in the past, the system is not inherently biased against either party, with one side’s seeming advantage lasting for just an election or two before it flips to the other party. But as the 1969-1970 example shows, it seems likely that only serious bipartisan support for abolishing the Electoral College system could ever change how we elect a president. Although states may figure out a way around the Electoral College with the National Popular Vote interstate compact, it would not seem as permanent as a constitutional amendment, given that only one amendment has ever been repealed. And as Faris argues, using the interstate compact method might precipitate a crisis because an outcome might be seen as illegitimate and be subject to legal challenges if it delivers a result that contravenes what the Electoral College would otherwise do.

Schatz’s proposal is unlikely to pass the Senate, but it may be a symbolic effort to influence the conversation about what we want our electoral system to look like. Nonetheless, without broader agreement, a constitutional amendment to abolish the Electoral College will pass when pigs fly.

From ABC News:
Sen. Elizabeth Warren wants to get rid of the Electoral College

Trump’s National Emergency Policy Is Unpopular, But Not Really Unpopular

Welcome to Pollapalooza, our weekly polling roundup.

Poll(s) of the week

President Trump’s decision to declare a national emergency in order to build more physical barriers on the U.S.-Mexico border was generally unpopular, but polls suggest the move has very high support among Republicans. That dynamic could be important as Trump seeks to overcome challenges to his new policy both on Capitol Hill and in the courts.

Two polls conducted entirely after the emergency declaration show a majority of Americans don’t like it: An NPR/PBS Newshour/Marist poll that came out Tuesday showed a 61-36 split against Trump’s policy, and a Morning Consult/Politico poll released on Wednesday found 39 percent in support, 51 percent opposed. A HuffPost/YouGov survey conducted the day before and the day of the emergency declaration found similar results — 37 percent of Americans said they approved of the move, compared with 55 percent who disapproved.

These numbers don’t surprise me — they generally mirror Trump’s overall job approval ratings. For much of the past two years, around 40 percent of Americans have approved of the president’s performance, while a clear majority has disapproved. Similarly, overall support for the national emergency declaration is in the upper 30s in the polls we have so far. That’s because Republicans have lined up solidly behind it, according to both polls conducted after the declaration was made — the NPR poll found 85 percent support within the GOP, and the Morning Consult survey found 77 percent support. The HuffPost/YouGov poll found that 84 percent of Trump voters supported the declaration, although that poll was already underway when the declaration was made, so some respondents were asked about the move before it became official while others were asked after the announcement.

It’s not surprising that large numbers of Republicans supported Trump’s decision to declare a national emergency — GOP voters overwhelmingly approve of him. But high party support for a Trump policy is not always a given. For example, the policy of separating immigrant children from their parents at the U.S.-Mexico border was significantly more unpopular within the party than the emergency declaration is — a FiveThirtyEight average of polls found that only about half of Republicans were on board with the separations. And while a majority of Republicans supported both the failed 2017 health care bill meant to replace Obamacare (67 percent) and the GOP tax plan passed the same year (64 percent), they did so at rates 10 to 20 points lower than we’re seeing on the national emergency policy.

Being backed only by Republican voters still isn’t great for the president. His base alone likely won’t be sufficient to win re-election. But in terms of policy, Trump tends to reverse himself only if there is a breadth of opposition that encompasses more than just Democrats and independents. That kind of opposition tends to create a feedback loop that’s hard to ignore — so, for example, the media criticizes something Trump does or says, establishment Republicans join in, and then the media prominently features those GOP critics in its coverage. Some Republican elected officials were initially wary of Trump declaring a national emergency, but I wonder if they will reconsider that posture after seeing these polls. And with few prominent Republicans willing to cast the national emergency policy as an “extraordinary violation of constitutional norms,” as The New York Times described it last week, I suspect the media will feel pressured to cover this debate as a traditional partisan dispute and so will back off from sharper condemnations of Trump.

Like the media, the courts are sometimes hesitant to take strong stands on partisan disputes. So they may be more reluctant to strike down Trump’s policy than they would be if it had gotten more of a mixed reaction from both sides of the aisle.

But the biggest reason these polls matter is they can affect what happens on Capitol Hill. House Speaker Nancy Pelosi announced this week that the House will likely hold a vote to overturn the emergency declaration. If such a measure passed both houses of Congress but was vetoed by Trump, Congress would need a two-thirds majority of both chambers to override the veto. That would require 53 Republicans in the House and 20 in the Senate to break with the president. I thought that was unlikely even before these polls came out. Now, seeing almost universal support for Trump’s declaration among Republican voters, it’s even harder to imagine a large bloc of Republicans in Congress breaking with the president, which means this policy is likely to survive.

Other polling nuggets

  • Former Vice President Joe Biden, who has not announced whether he will enter the 2020 Democratic presidential race, leads in South Carolina with 36 percent of the vote, according to a new Change Research survey that was provided to the Palmetto State-based Post and Courier newspaper. Also in double digits were Bernie Sanders (14 percent), Kamala Harris (13 percent) and Cory Booker (10 percent). Twelve other Democrats who either have officially entered the race or are rumored to be considering presidential runs were in single digits.
  • Biden is also ahead in New Hampshire, with 28 percent of the vote, according to a University of Massachusetts, Amherst, poll. Sanders is in second place with 20 percent, and Harris is in third with 14 percent. Seven other Democrats that were included in the survey are in single digits.
  • An average of 54 percent of white Democrats identified as politically “liberal” during the six-year period from 2013 to 2018, according to data released by Gallup this week. That compares with 38 percent of Latino Democrats and 33 percent of black Democrats. There was also variation by education level — Democrats with postgraduate degrees were the most likely to describe themselves as liberal (65 percent), followed by Democrats with undergraduate degrees (58 percent), those who attended college but don’t have degrees (45 percent) and those with high school educations or less (32 percent).
  • The same Gallup survey found major differences among liberal and conservative Democrats on a few issues: 81 percent of liberal Democrats think marijuana should be legal, for example, compared with 44 percent of conservative Democrats. Sixty-four percent of liberal Democrats oppose the death penalty for people convicted of murder, compared with 39 percent of conservative Democrats.
  • 45 percent of Democrats said they would be somewhat or very unhappy if their son or daughter married a supporter of the Republican Party, according to a PPRI/Atlantic survey released this week. Thirty-five percent of Republicans said they would be unhappy if their child married a Democrat. Higher shares of Republicans were concerned about their child marrying someone of the same gender (58 percent unhappy) or someone who identified as transgender (70 percent).
  • More than 60 percent of Americans said that the government should pursue policies to reduce the wealth gap and that they support a 2 percent tax on wealth above $50 million, according to a survey conducted by SurveyMonkey that was published by The New York Times this week. Opinion is more divided (51 percent support, 45 percent oppose) on a marginal tax rate of 70 percent on income above $10 million a year.
  • Just 17 percent of Virginians said they approved of the job performance of the state’s embattled governor, Ralph Northam (who has denied that he was in a racist photo on his page in his medical school yearbook but has admitted to wearing blackface in the 1980s), according to an Ipsos/University of Virginia Center for Politics survey released this week. Thirty-four percent said they disapproved of Northam, while 44 percent said they neither approved nor disapproved. The good news for Northam is that only 31 percent of Virginians said they think he should resign, compared with 43 percent who said they don’t think he should resign. Lt. Gov. Justin Fairfax, who has been accused by two women of sexual assault, has a slightly worse standing — 35 percent of Virginia residents said they think he should resign, while 25 percent said that he shouldn’t. (The other 40 percent are in neither camp.)
  • A new Quinnipiac University survey of Virginia voters found better job approval numbers for Northam (39 percent approve, 44 percent disapprove). And in this poll too, a plurality (48 percent) of Virginians said he shouldn’t resign.

Trump approval

According to FiveThirtyEight’s presidential approval tracker, 42.5 percent approve of the job Trump is doing as president, while 53.2 percent disapprove (a net approval rating of -10.7 points). At this time last week, 41.5 percent approved and 54.1 percent disapproved (for a net approval rating of -12.6 points). One month ago, Trump had an approval rating of 40.0 percent and a disapproval rating of 55.3 percent, for a net approval rating of -15.3 points.

From ABC News:

We’re Tracking 2020 Polls

Presidential election polls for 2020 are rolling in, and we’re collecting them all in one place for you!

You can check out the latest on our polls page, which we’ll update every day with polls of the Democratic and Republican presidential primaries, as well as head-to-head matchups for the general election. You can download the data at the bottom of the page or at And if you see any polls that we missed, please shoot us an email at [email protected].

While we have you here, though, we thought we’d explain a little bit about the choices we made for displaying 2020 primary polls. Showing polls for a race that is still (relatively) far in the future creates some challenges. In particular, we had to come up with a design that works for polls that ask respondents about a lot of candidates — sometimes 25 or more. And those results have to be displayed along with the results of surveys, such as presidential job approval or generic congressional ballot polls, that have only two main answers.

So in the updated version of our polls page, the default view will show only the most popular choice in polls in which respondents were asked about multiple candidates. To see the rest of the candidates in the poll, click “more” to expand the list of results.

It’s not a perfect solution. This early in the primary season, any of the Democratic candidates could realistically become the front-runner, and we’d prefer not to display one and hide others. But this strategy allows us to keep all the polls in one central location where they are easily discovered.

As the volume of polling picks up ahead of the 2020 election, we’ll expand how we track the latest survey releases, but this page will fuel all FiveThirtyEight’s 2020 work to come.