Sunday, July 10, 2016

Reuters-Ipsos polling more Democrats, fewer Republicans now than it was two months ago

Cicatrizatic notices that nearly half of those who participated in the latest formal Reuters-Ipsos presidential preference poll are Democrats while just one-third are Republicans.

Reuters-Ipsos tracks the figures daily. It has done so for a head-to-head match up between Hillary and Trump since May 1. The following graph shows the partisan affiliation of "likely general election voter[s]" through July 8:

Back in May when Trump enjoyed a slight lead in the RCP average, Reuters-Ipsos' Democrat-to-Republican ratio was considerably closer to 1-to-1--and consequently closer to what other surveys have found the country's partisan distribution to be--than it has been for the last several weeks.

That could be a consequence of Republican recalcitrance in the face of Trump's refusal to morph into John Kasich, the tendency for people to identify with whatever party the candidate they intend to vote for is from, a problem with Reuters-Ipsos' sampling methodology (either inadvertently, or, more sinisterly, intentionally), some combination of these things, or something else entirely.

Whatever the explanation, the shift is noteworthy. As the sample has become more heavily Democrat, Hillary's support has correspondingly increased while Trump's has decreased.


The Z Blog said...

I tend to assume good old fashioned confirmation bias in these situations. As their sampling becomes more Democratic, the results appear more in line with what their instincts tell them is correct. We saw this in 2012 with GOP polling outfits. They will never own up to it, but that's my hunch.

Interestingly, the final data on primary turnout shows the GOP way up and Democrats down significantly. That does not correlate strongly with general results, but it does provide a window into *who* is strongly invested in this election. If I were to guess, Bernie Bros propped up Democrat turnout in the primary so where they go is a big concern for Team Hillary.

Similarly, the boost on the GOP side was clearly the result of Trump. First time voters were a big factor. These are the so-called lost white voters Sean Trende documented. There's also the remnants of the working class Democrats that never made it over to the GOP in the 80's and 90's.

I think the pollsters will have a miserable time modeling the general election this time. You have too many outliers. Will Bernie Bros come home and vote Hillary? Will Bush Republicans come out and vote Trump? Will blacks both to vote? People forget that the last days of rage stuff depressed the black vote.

Ron Guhname said...

Thanks! I've been using their website a lot and noticed that the gap between Hillary and Trump has gotten much bigger--12 points--than the RCP average of about 5 points. They seem to be following some methodology that diverges from the pack. Maybe sampling too many Dems. I assume they're off and the others are closer to the truth.

Audacious Epigone said...


Yes, it even seems like that from general election polling now. Comparing the margins across polls is huge. The two most recent show Trump +2 and Hillary +9, respectively, an 11 point margin. In '12 by contrast, of the last nine polls reported, the margin only ranges 4 points, from Obama +3 to Romney +1. That seems to corroborate your sense that pollsters are having a tougher time than usual.


I've followed R-I for several months now because they conduct their polls daily and because of how user-friendly the cross-tabbing. During the GOP primary it was pretty consistently the one that was most favorable towards Trump. The easy explanation is that it over samples Dems/Independents. It was also one that showed for a brief time in May that Trump was ahead of Hillary, however. That was when it was polling much closer to 1:1 Reps and Dems though, so who knows? We can have fun speculating at any rate.

Mil-Tech Bard said...

Audacious Epigone

Significant denunciation of voters' opinions by media and political elites only gets the latter to go dark, but doesn't influence their opinions in the slightest.

Only hostile environment polling - polls of secondary opinions which when properly analyzed reveal real opinions on other issues - can work now. This is vastly expensive, incompatible with phone polling, etc, etc, etc.

This is several orders of magnitude greater than the Bradley Effect.

It is also different from, but related to, systematic bias by the polling industry. Basically the polling industry is a faction of the MSM and political elites in general.

With Trump, all of those factors are working simultaneously.

BREXIT was a damned good example of the "Go Dark" phenomena.

Politico thought a low turn out election meant REMAIN would win because young voters would bury LEAVE.


6. National turnout

By around 4 a.m, pundits will be able to give a rough estimate of the turnout — Downing Street’s major concern.

Vote Leave believe anything less than 60 percent and victory is theirs. North of 70 percent favors a vote to Remain, because it suggests younger voters and middle-class professionals have turned out.

A freak turnout like in the Scottish referendum, when 85 percent voted, could see a swing back to Brexit, because it points to working-class traditional non-voters being motivated to join in — one of the most anti-EU demographics.

In fact, BREXIT turn out was north of 70% and LEAVE still won by 52% to 48%. The increased turn above 60% included large swaths of Labour working class whites in Northern England.

In fact, an 85% turn out might have favored REMAIN.

Immigration was an existential threat to the UK working class, and given an option to vote against it, they will turned out in disproportionate numbers vice younger voters.

Audacious Epigone said...

Mil-Tech Bard,

The conventional wisdom in the US is that higher turnout is good for Democrats and bad for Republicans. I don't even think the professional punditry class is going to assume that will be the case this time. First-time voters will go to Trump, something that is unprecedented in recent American political history.

Mil-Tech Bard said...

Audacious Epigone

It isn't the 1st time voter that makes or breaks 2016.

It is the voter who gave up, but will show up for a "real nationalist" candidate or cause that is key

There are a lot more of the latter than the former and those are the voters Trump is bringing to the table.

Just like the North English Labour white working class -- abandoned by both the Tory government and the cosmopolitan "New Labour" -- was for UKIP's Brexit vote.

The Z Blog said...

Here we have another poll showing that Clinton is under-performing with a key demo.

That's been the story of this election. She is well behind with every key Democrat coalition member. According to a recent poll, 56% of Americans believe she should have been indicted. What percentage of that 56% just want to see her get her day in court so she can prove her innocence? My bet is it rounds to zero.

I fully admit to having a strong bias, but there's not a lot of good news for Team Clinton at the moment.

Anonymous said...

What the fuck?!?!


Clinton extends lead over Trump to 13 points: Reuters/Ipsos

Cicatrizatic said...

There is an Iowa poll out today that includes Trump, Clinton, Johnson, and Stein, and has Trump +2.

Their sample is 37 % D / 35 % R / 28 % I - which looks like a much more accurate election day sample than most of the other polls. 2012 was 38D/32R/30I nationally.

Mil-Tech Bard said...

Trump supporters have repeatedly pay a price for publicly expressing their beliefs.

And the most fervent Trump supporters have not voted for President in the last several election cycles.

They won't talk to pollsters any more than BREXIT supporters did in the UK.

NB - The whiter and more working class the American state, the bigger the Trump voting margin over the polls in those states.

See this Salena Zito column dated May 21, 2016

Pollster Wes Anderson of OnMessage, a Washington-based Republican strategy group, says Politco's article finds that the vast majority of Trump primary supporters had voted in at least one of the last four presidential elections: “Really? If a voter skipped the 2008 (or) 2012 general election but now voted in a GOP primary, isn't that evidence that Trump is expanding the GOP vote?”

Anderson pointed to North Carolina's early-spring primary to prove his point: There in 2012, “inactive” voters (those who skipped one or more of the previous two presidential elections) made up just 2 percent of the vote. In 2016, they represented 12 percent.

“That's right, 12 percent of the primary had skipped one or both of the last two presidential general elections,” he explained. “We're talking about voters we would normally give up on because history would say they have given up on voting.”

This year they decided to vote in a primary. How does that not say something significant about what Trump may be doing to the voter pool? The number of truly “new” voters may be small but the number of independents participating in GOP primaries for the first time is big.

The same is true of Republicans who never participated in a primary, until now.

Politico's article is classic half-truth that stretches or distorts data to prove a preconceived point. If you miscalculate the voter universe now and don't understand what will drive it to vote, then you likely never understood this entire election cycle and will continue making the same miscalculations.

Meaning that November may shock the hell out of a lot of political professionals.