Tuesday, September 27, 2016

poll to poll

This is unbelievable to me, but here are the poll numbers that Trump and Hillary are tweeting, post-debate:

Breitbart (leans right): 76-24, Trump (yawn—no surprise)
Variety (leans left): 55-45, Trump (major surprise)
NJ.com (Jersey; leans...?): 57-39, Trump (not a major poll)
The Hill (leans...?): 59-36, Trump (major poll for pols)
Drudge (leans right): 80-20, Trump (I consider this very unreliable given heavy bias)
Time (leans left): 59-41, Trump (I'm flabbergasted)
CNBC (leans left): 61-39, Trump (this was most shocking of all)

Before we move on to what Hillary Clinton tweeted, let me say that I automatically discount the polls hosted by the rightie sites. Of course they're going to skew Trumpward. Duh. That means nothing. What horrifies and fascinates me, though, are the polls from unabashedly leftie sites like Time, Variety, and for God's sakes—CNBC.

In some of his videos, Styxhexenhammer666 makes the point that there are two parallel realities: there's the legacy-media reality, which is where most of the population probably gets its information; then there's the online reality, which has grown and metastasized, right under everyone's noses, into a formidable beast that now produces statistically significant psephological results. People only paying attention to the legacy data being gathered by folks like Nate Silver are completely missing the groundswell of online activity, most of which favors Trump at this point. Therefore, it's more likely that it's the online people who are madly clicking these legacy-media polls, which means that people focused on legacy media are, for the first time, seeing the concrete results of this ever-burgeoning online reality. Like a scenario from a fantasy novel, one universe is slowly, steadily, spilling into another.

Onward to Hillary's tweets.

CNN (leans left): 62-27, Hillary (no surprise)

...and that's it. That's the only poll that Hillary quoted. Amazing.

Could my perception of the debate—admittedly gained through the filter of both leftie and rightie commentary—really be that skewed? My post-debate impression, garnered in part from sober commentators on the right, some of whom openly support Trump, was that Hillary was more articulate, more substantive, and more coherent than her opponent. This was not only a testament to her debate prep, but also to Trump's failure to attack Hillary's more obvious weaknesses. Rightie talking heads like Stephen Green claim this is because Trump lacked the factual knowledge to actually attack Hillary on the facts... yet the above polls—poll after poll—seem to suggest that none of this mattered to great swaths of the American populace: instead, impression was everything.

Glenn Reynolds of Instapundit has been calling Trump a grandmaster-level troll for some time. Some of the commentators that Hairy Chasms visitor King Baeksu has pointed to have been talking about Trump's mastery of "psy-ops." Scott Adams of Dilbert fame half-jokingly refers to Trump as a "wizard," a master manipulator of mass perception. I'm left to wonder what, exactly, it is I'm witnessing. Is Trump really that good a manipulator of people's impressions? Can he really lose a debate and still seemingly win it, if by "winning" we mean "poll better than his opponent"? What sort of creature is this?

I've felt this way before. In fact, I felt this way all during the GOP primary debates. Trump would tout the polls; I'd wave them off. Trump ended up the GOP front-runner, then he became the GOP nominee for president. He's even received an open, explicit endorsement from Ted Cruz—the man Trump had branded as "Lyin' Ted." All I can say is that I'm in a state of disbelief. Can this really be happening?

If the map doesn't match the terrain, then a sane person, a scientific person, knows he must redraw the map. I'm not saying that I now plan to endorse Trump, but I am saying that I'm going to have to be less bearish about his electoral prospects.


Breitbart: "Who Won the First Presidential Debate?"
Variety: "Who Won the First Clinton-Trump Debate? Vote Now!"
NJ.com: "Who won the presidential debate (9/26/16)? How did Clinton, Trump do in the first debate?"
The Hill: "Who Won the Debate?"
Drudge: "WHO WON THE FIRST PRESIDENTIAL DEBATE?" (shouty all-caps)
Time: "Who Won the First Clinton-Trump Debate?"
CNBC: "Clinton or Trump: Who Do You Think Won the First Presidential Debate?"

CNN: "Regardless of which candidate you happen to support, who do you think did the best job in the debate – Hillary Clinton or Donald Trump?"

Note: unlike the polls Trump cited, the CNN poll was apparently administered differently. It wasn't an insta-click online poll; instead, it was a telephone-interview poll of 521 registered voters who all watched the debate. (LINK) The online polls have votes numbering from the hundreds of thousands to the millions.

Note 2: CNN (see link below) says its survey sample of 521 people skewed significantly Democrat. I appreciate CNN's honesty.

UPDATE: The Daily Mail surveys a bunch of online snap polls.


Charles said...

Unless I missed something, those polls are all based on data collected prior to the debates, no? I mean, polls don't just pop up and immediately reflect the current political landscape. There is usually a lag time of a few days--this has always been the case.

So unless we have suddenly developed insta-polls that immediately tell us what people are thinking right now, the numbers quoted above do not reflect reactions to the debates. They reflect the political mood prior to the debates, and that has been moving toward Trump for quite some time now. I think we will have to wait a few days to see what affect the debates will have on public opinion.

Kevin Kim said...

The question on many of those polls was/is: "Who do you think won the debate?"—in that specific verb tense. The Drudge poll, for certain, was already at around 92-8 before the debate was halfway done, so yeah, the polls may indeed reflect sentiment "prior to the debate"—but more in the sense of "prior to the end of the debate," not sentiment from days before. (Drudge's poll wasn't even up until right before the debate started.)

Anyway, you'd be right to question the accuracy of the polls; I don't blame you. But I think we're seeing a phenomenon here, so I'm changing my view accordingly. I still don't think it's Trump by a landslide, but I'm prepared to be wrong.

These are all online polls, too, as far as I know, so 800,000 people can easily vote on any one of them in a short span of time. There's no logistical difficulty; data collection is instantaneous, as long as the server hosting the poll can handle the massive traffic.

Kevin Kim said...


I just went through every poll I mentioned to confirm wording. In every case, the poll question was worded in the past tense, implying the poll was an insta-survey conducted right around the time of the debate, not days before, and certainly not days after (we're still barely 12 hours from the end of the debate).

I've written down the poll questions as an addendum to this blog post, so you can see them for yourself.

Nathan B. said...

Kevin, are the poll results you've linked to even remotely scientific? I only checked one (the CNBC poll), and it had a prominent disclaimer about what it wasn't: rigorous and scientific.

Online polling is susceptible to various kinds of hacking, from massive-scale botnet-directed to something as simple as deleting one's browser's cookies and cache. Given what looks to be Russian involvement in the hacking of the DNC emails, I wouldn't put it past Russia to be meddling in online polling, too.

In short: I wouldn't put any stock in any online poll on this topic. It seems to me this post is too quick to label the better, at least semi-scientific polling as merely "legacy," while at the same time inflating the accuracy of polls that are by definition not at all scientific or accurate.

Kevin Kim said...


We're at an epistemic impasse, unless you can definitively prove that these online polls mean nothing and/or are tainted by Russian influence. To be honest, I'd like to believe you, but more and more, I'm starting to believe the likes of Scott Adams, not because I'm becoming a Trump fan, but because what Adams says does seem to match events in the real world.

It's always a question of whose map of reality more accurately matches the territory. I suppose we'll know more come November.

Nathan B. said...

Hi Kevin,

Well, it may be that Trump will win the election in November (I sincerely hope not, by the way); I just don't think that online polls are at all trustworthy when the stakes are so high.

Nathan B. said...

I should probably mention that I actually want the real polls to show close numbers: that will motivate the Democratic base to turn out and vote for Clinton. It seems like in general, the Republican base is more likely to vote than its counterpart.

(By the way, I don't actually like Ms. Clinton. I just really don't want Trump elected. But the two choices aren't very inspiring. Why couldn't the Democrats have nominated someone like Elizabeth Warren, and the Republicans someone like Colin Powell?)

Surprises Aplenty said...

i can only add that on my hobby-horse subject, evolution v creationism, I know people who seek out polls to skew. When a church or religious group creates a poll on the subject, these people would post the information at pharyngula - a blog devoted to evolution and atheism - and others - hundreds of others - would visit that poll.

There were enough of them to skew a local poll but probably not enough to mess up federal debate polls. Still, there must be others like this.

Any poll that has no restriction criteria (or whatever staticians call it) is not to be trusted.

Charles said...

Thanks for the clarification. I see what you mean now.

Although I kind of have to agree with Nathan in that I'm not sure how much weight I give to these polls--not because of any doubts about their legitimacy (which I also have), but because I'm not sure if they really matter. I think if we really want to see who won the debate, we'll wait to see how the polls play out in the coming days. If Clinton starts re-widening her lead, she'll have come out the victor. If the gap narrows or Trump surpasses Clinton, that will mean he was victorious. What people are saying now in online polls with regard to who "won" the debate doesn't mean as much in the long run as how the debate influenced perceptions of the candidates.

King Baeksu said...

"All I can say is that I'm in a state of disbelief. Can this really be happening?"

During the debate, $hillary invoked the concept of "implicit bias." I think your own implicit bias as a university-level educator, writer and holder of a master's degree is holding you back from viewing this election process on a more neutral or less self-interested basis.

Certainly $hillary was more articulate and better informed on policy. Naturally, such qualities will appeal to you as a highly educated individual yourself. But let's recall that only 11.5% of Americans hold a master's degree or higher. You are a member of the cognitive elite, in other words. However, in modern-day American democracy, the principle of egalitarianism is absolute, and the vote of a genius with a PhD holds equal weight with that of a pot-smoking high-school drop-out whose favorite intellectual activity is watching Sunday football at the corner bar on a Sunday afternoon. This system may not seem fair to you, but that's the way it's been set up for now, so one simply has to deal with it. (I also suggest that you not voice such criticisms as I do here for the sake of your own professional career, especially if you ever plan to return to the US, as you will immediately be attacked for failing to "check your privilege" or even worse.)

Two-thirds of Americans do not even possess a bachelor's degree. To them articulateness and policy mastery signal a social status that is beyond their reach and therefore somewhat alienating. No one likes to be made to feel stupid, or uninformed or inferior. If you recall the Bush-Gore debates of 2000, this truism is affirmed: Gore was perceived as snooty and condescending, whereas Bush was "the guy you'd like to have a beer with." In terms of sheer numbers, projecting a folksy regular-guy image is going to appeal to a whole lot more voters than the one third of Americans who posses at least a bachelor's degree and therefore value the thrust and parry of intellectual debate.

Imagine you have not paid attention to the present campaign up to now, and are a high-school educated union worker in Ohio or Pennsylvania. During the first debate, you are observing $hillary and Trump side-by-side for the first time. What do you see?

In the case of $hillary, the image she presents is of a smug policy wonk who has a tendency to lecture those she is addressing. We may indeed all have "implicit bias" to some degree, but no one actually likes be told that they are effectively racist or prejudiced. Even bad individuals tend to think that they are good people and that there are perfectly justified reasons for whatever they do. I also feel she pushed the "social justice" angle much too hard during the debate, and I think the general zeitgeist in the US these days is that the SJW phenomenon has gone much too far and needs to be reigned in.

What is the image that Trump presented during the debate? Essentially, a plainspoken businessman who may not be an obsessive policy wonk, but who possesses an even greater virtue: Common sense. At one point, he even stated that he'd rather have the endorsement of military leaders with real-world, practical experience than all the "brilliant" political hacks advising $hillary, That was his "common sense" pitch in a nutshell, and he made it repeatedly. And if you were paying attention, he also explicitly reached out to Sanders supporters and the African-American community, among others. In other words, while $hillary came off as something of a scold on the social-justice front, Trump appealed to the disenfranchised voter with a far more positive approach, and as we all know, the positive tends to trump the negative every time.

(Comment continued below)

King Baeksu said...

(Comment continued from above)

So from the perspective of someone not at all like you, which candidate was more appealing? Especially to an undecided voter who is not overly partisan or committed to a particular ideology, but rather is more worried about how to pay all their bills for the month and possibly avoid eviction? I'd have to say Trump, without question. I'd even say that his occasional inarticulateness was almost endearing, since it makes him "relatable" to the average, ordinary voter. Plus his message of "can-do optimism" is very American, as is the glamour of his wealth and celebrity, which may seem gauche to our elites both in the US and abroad, but in fact is also distinctly positive and worthy of emulation in the American psyche (America is traditionally a nation of strivers, not resentful "short poppies" as is generally the case in the UK or Australia and many other countries).

In short, the debate was a showdown between Nurse Ratched and Randle McMurphy (of "One Flew Over the Cuckoo's Nest"). Who would you rather have to see on TV or your smartphone screen every day for the next four years? I think the choice is obvious. McMurphy may be something of a scoundrel, but at least he's a fun guy to hang out with and has the bravery to say what everyone else is thinking: The system is totally fucked up and needs a damn good kick in the ass.

The thing about modern-day democracy is that you need to take a broader view that transcends your own particular viewpoint. The activist base on either side of the partisan divide made up its mind ages ago. The next level then, is for them to influence those less committed or involved voters who need some help in making up their own minds. (You've no doubt heard of the "white man's burden." Think of this obligation as the "woke man's burden.") Trump's message is solid: "Free-trade" in its present form is a rigged game that does not benefit American workers, and Open Borders is also a racket that is in the interest of our elites (the 1%) but not the average American (the 99%). In short, he is a US nationalist in an era in which globalism is seen to be a "false siren" by a great and growing many. So as far as these debates go, the trick is for Trump to "pivot" and expand his appeal to a more mainstream swathe of the American public. In the first round, he appeared restrained, sane and not at all bullying. In other words, he basically obliterated the image of him that $hillary and her campaign staff have been trying to sell up to this point. This is why I said in my previous comment: First round, advantage Trump. My personal opinion of how he performed on policy points or articulateness is irrelevant, because people like me were not his intended audience.

(Comment continued below)

King Baeksu said...

(Comment continued from above)

In any case, the next stage in this process is for the hard-core influencers to manage the "spin" or "framing" of the aftermath of the debate, and this is where the polls come in. I won't get into the mess of how accurate the online polling is, except to say two things: First of all, my own assessment of how well Trump succeeded in selling himself to the mainstream voter largely accords with the online polls we are seeing. (Insightful observers like Scott Adams also agree, for what it's worth.) More to the point, however, is the fact that Trump is the candidate of the Internet, and this is why he is winning both in the online polling, and why all the momentum is on his side at this point. I believe the following passage, from the journal "n+1," explains both the dominance of Trump during this election cycle and the inability of the legacy media to keep up with the hydra-headed monster that is the Internet today:

"Unwilling to go to my parents’ house and unable to sleep in the office tilt-and-swivel-chair, I picked up the book I’d brought from New York: 'The Theory of Poker,' a how-to classic of 1987 written by David Sklansky, a native of Teaneck, dropout from U Penn’s Wharton School of Business (where he just missed overlapping with Trump), winner of three World Series of Poker bracelets, and arguably the greatest Draw and Hold’em player of all time. In the very first pages of his book—which I must’ve read a dozen times before, for a reliable soporific—Sklansky lays out his Fundamental Theorem, which in my amped-up wakefulness now hit me like a law on the level of gravity’s: “Every time you play a hand differently from the way you would have played it if you could see all your opponents’ cards, they gain; and every time you play your hand the same way you would have played it if you could see all their cards, they lose.”

"Here, presented in sane, rationalist fashion, was the insane truth behind this race: that if Trump just keeps on being Trump, and if Clinton keeps pivoting and responding to his every move, he wins. The only way that Clinton can win, according to Sklansky’s schema, is to force Trump to become inconsistent, but since Trump is already inconsistent—since he’s consistently inconsistent—that’s impossible."

I said many months ago that Trump would win in a landslide and I still stand by the prediction, because he understands that modern-day democracy is all about marketing and he's a genius on that front (in part because he is simply sui generis, a one-of-a-kind wonder). Put another way, $hillary is trying to sell us a plate of steamed, unseasoned vegetables, whereas Trump is not only selling the proverbial "sizzle" of Marketing 101, but the steak itself is big, fat and juicy as well. It's not even a contest, really, and I'll say it again: Trump in a landslide, mark my words.

King Baeksu said...

This NY Post article pretty much accords with what I wrote above:

"How Trump Won over a Bar Full of Undecideds and Democrats"

"Reed, 35, is a registered Democrat and small businessman. 'By the end of the debate, Clinton never said a thing to persuade me that she had anything to offer me or my family or my community,' he said, sitting at the same bar that has boasted local icons as regulars, such as the late Fred Rogers, and Arnold Palmer, who had his own stash of PM Whiskey hidden behind newer bottles of whiskey for his regular visits.

“'Have to say Trump had the edge this evening, he came out swinging but also talked about specifics on jobs and the economy,' Reed said.

"Reed said Clinton came across as either smug or as though she was reading her résumé, adding there was nothing on her résumé that touched on his life. 'I am a small businessman, a farmer, come from a long line of farmers and coal miners. The policies she talked about tonight ultimately either hurt me or ignore me,' he said."

It looks like more and more Americans can see through $hillary Clinton's well-polished lies, and understand all too well that she's simply not on their side at this point. Trump, on the other hand, is talking about the kind of issues that working Americans actually care about.

As I said before, advantage Trump this time around. You can expect him to refocus and refine his message next time around, and most likely go for the jugular. And it will be a glorious thing to see.

Nathan B. said...

For what it's worth, this is how the polling issue of who won is being reported up here in Canada. The CBC article notes that Trump's "poll victories" are unrepresentative, and links to several major polls that come down on the side of Clinton.

Kevin Kim said...


I think the CBC article concludes fairly (debate impact unknown for now), and it also notes some problems with the supposedly more "legitimate" polls that it cites (although it does minimize the Democrat-stacking in the CNN poll). I think the article's methodological criticisms are fair, but they don't really address the fundamental issue raised by Scott Adams et al. as to whether (1) Trump truly is an effective persuader, and (2) a groundswell coming primarily from online, and not the legacy media, in fact exists and is having significant statistical impact. I wish there were a clear way to measure these phenomena; right now, what we've got are gurus and intuitions. Oh, and Nate Silver's algorithms.