One Poll Doesn’t Equal “Polls” & It Doesn’t Make This Clintonite’s Case Either [Updated Below]

Medium Dept. - "Dude Kembro," yet another crackpot Clinton cultist, has authored yet another rubbish article--"Why does media ignore polls showing Dem base still prefers Hillary not Bernie?"--that tries to craft a fantasy of how golly-gee great it would be if Hillary Clinton were to launch a 2020 bid for the presidency, a prospect that no one but Clinton, her diminishing cult and Donald Trump would find remotely appealing. I'm not going to deal with the bulk of his loony ravings but he makes some comments about polling, so I decided to tackle that narrow part of it.

Kembro describes the notion that Bernie Sanders is "the nation's ‘most popular senator'" as an "outdated fiction" based, in part,

"on pushing disgruntled ex-Clinton pollster Mark Penn's monthly Harvard-Harris poll, notoriously unreliable due to its online methodology."

Well, let's see...

--The very YouGov poll on which you're relying as an alternative to Harvard/Harris is also an online poll:

--You provide nothing to support your fanciful assertion that Harvard/Harris is "notoriously unreliable," because there is, in fact, nothing to support that claim.

--While your narrative suggests Mark Penn is some fellow with an axe to grind--some sort of motive for his faking a poll in an anti-Clinton direction, whatever would be gained by that--he is, in fact, a rightist Clintonite loyalist who has been with the Clintons for 23 years. When Democrats lost the 1994 congressional elections, he was the guy Bill hired to try to get things back on track. He worked for Bill right to the end of his administration, worked on the '96 reelection campaign, worked on both of Hillary's Senate races and worked on her 2008 presidential campaign. Only this Summer, he co-authored an op-ed in the New York Times and this was his advice to Democrats, re:the Sanders/Elizabeth Warren challenge from the left:

"The path back to power for the Democratic Party today, as it was in the 1990s, is unquestionably to move to the center and reject the siren calls of the left, whose policies and ideas have weakened the party."

That delusional piece was uniformly derided across progressive media. Penn is both stupid and his views delusional, but any motive he may have to fake a poll--an asinine suggestion anyway--is in the other direction, toward making Clinton look better and Sanders worse.

--The Harris poll isn't some wet-behind-the-ears upstart; it's one of the longest-running established pollsters in the U.S., founded over 50 years ago (when, I'll note in passing, Mark Penn was 9 years old). That doesn't mean it's right, of course. It does mean it isn't some amateur in the field.

--While Harvard/Harris is conducted every month, your YouGov poll is just that, a single poll, and could be utterly anomalous, which happens in polling all the time (H/H isn't immune to it either).

--In line with that last, there are some huge warning-signs within the YouGov results, principally the often-vast number of people who express no opinion. For example, 21% of independents report no opinion of Clinton while a staggering 34% report no opinion on Sanders. A huge warning that there's something very wrong with the data. And you make it even better. You cite the two politicians' relative standing among minorities in the poll and make a lot of fanciful assertions based on it:

"Hillary has a 78 percent favorable rating with Democrats, seven points higher than Bernie's 71 percent. Her 69 percent favorable rating with blacks remains a whopping sixteen points better than Sanders's 53 percent rating with that determinative demographic. Among Hispanics, Hillary's 42 percent rating comfortably bests Sanders's dismal 36 percent."

But the poll shows that nearly 1 in 4 black respondents offered no opinion of Sanders, whereas this was only the case with 8% when it came to Clinton. Clinton only leads in this demographic by 16% and three times as many black folks are reporting no opinion of Sanders as reported no opinion of Clinton. This same dynamic repeats itself among Hispanics. It's astonishing that 24% of Hispanic respondents offer no opinion of Clinton but utterly incredible that 36% offer no opinion of Sanders.

It's also worth noting that your assertion regarding Clinton's lead among Hispanics is comical. While you insist she "comfortably bests" him, Clinton's lead among Hispanics is only 6% in a poll with a reported margin of error of 4.1%, meaning less than 2% separates them, yet you call his standing "dismal."

--The overwhelming majority of "independents" always vote for their favored party and are just Republicans and Democrats by another name. Pollsters call independents who always vote for the same party "leaners," while those who vary their votes among the different parties are called "true independents." Whereas Harvard/Harris combines the leaners with the people who identify with the parties--a methodology that makes sense--the YouGov poll you cite divides these categories, which can seriously distort the data, as the leaners always make up a large share of the vote for each of the parties.

--The YouGov poll used only 1,000 respondents. That's not unusual for these sorts of routine polls where laserlike precision isn't essential. Some pollsters use even less. Harvard/Harris is a massive survey compared to these sorts of workaday polls. The just-released November H/H poll used 2,350 respondents.

--The YouGov poll appears absent context. Whereas H/H has been surveying on the same questions every month all year, YouGov hasn't, as far as I can tell, conducted a single poll on these particular matters for 9 or 10 months before the one you've cited. H/H monthly polls provide a context wherein one can see the various politicians' ratings changing over time. Sometimes--relatively rarely--there are anomalous jumps or dips but they usually seem to work themselves out in the next one. By lining up their polls, we can see that, for example, Sanders' popularity has been in slow decline throughout the year, a pattern veteran poll-watchers will find familiar. Earlier this year, when other pollsters were polling on Sanders' and Clinton's popularity, their results mirrored those of H/H. No other polls have mirrored this YouGov poll.

This is Sanders' current standing in the new Harvard/Harris survey:

And for comparison, this is Clinton's standing in the same poll, tied with Donald Trump's abysmal 38% overall favorable rating:

Sanders leads her by significant numbers in every category except among African Americans, where the two are statistically tied. He even leads her among people who voted for her last year.

Most of the rest of your piece is a Clinton-cult fever-swamp fantasy that has as little connection to reality as the Lord of the Rings. When one reads that kind of unhinged raving about a strange and mysterious place where the press isn't a virtual monolith of reflexive pro-Clinton sentiment, where Joe Biden is somehow unacceptable to Clinton's cult rather than one of the politicians to which elements of that cult are now turning to beat back any challenge from the left, where Clinton is some sort of counter-Establishment figure, where math is such that a single poll equals "polls" but multiple polls that disagree apparently don't and where Bernie Sanders, the lifelong civil rights activist and feminist, is somehow a racist and sexist, one may wonder what color the sky must be in that parallel universe but one isn't going to mistake it for anything resembling our own. To any informed, reasonably intelligent reader, your piece makes clear in virtually every line that nothing you say need be taken remotely seriously. As your nonsense regarding this poll is the only foundation upon which you've built the rest of your fantasy, I just decided I'd correct it.

--j.

---

UPDATE (Wed., 29 Nov., 2017) - Kembro had written that the Harvard/Harris poll was "notoriously unreliable due to its online methodology," but after I and others who commented on his article noted that the YouGov poll he, himself, was using was also an online poll, Kembro turned up and removed this wording from his article, replacing it with a description of H/H as "a survey given a poor C- in FiveThirtyEight's pollster ratings due to its unreliability." A little later, he added language that describes YouGov as "a pollster with a solid B rating from FiveThirtyEight." For whatever reason, he also changed his listed name from "Dude Kembro" to "DK Kembro." The current version of his article gives no indication that he's changed any wording.

The FiveThirtyEight pollster ratings chart on which he bases these newly-inserted assertions is long out-of-date--the page indicates it hasn't been updated since 5 August, 2016, nearly 16 months ago. While the chart rates Harris Interactive and that's the rating Kembro cites, the Harvard/Harris collaboration didn't exist until this year. The chart compares the outcome of elections to the final polls preceding them by various pollsters. Kembro apparently didn't read beyond the simple letter-ratings he cites. While he wants to make a case that YouGov is more accurate, YouGov's simple average error is 6.7%, compared to only 5.5% for Harris. YouGov's better letter rating is based on the fact that it accurately called the results of 93% of the races it covered. Harris, the pollster Kembro says is "unreliable," accurately called 86% of the races it covered--hardly a chasmic difference. Perhaps just as significantly, Harris is only ranked based on 135 polls; YouGov is ranked based on 707.