As was predictable, the polling from Hoax Poll organizations has shown considerable tightening, with Donald Trump going up comfortably in Georgia (as he should be) and taking the lead in North Carolina (which, in reality, he never had relinquished). A new UK Battleground Poll last week showed Trump up in every single battleground state, including Minnesota, and leading nationally. Personally, I think this is where the race is today.
It is critical to remember who is doing the polls, what their objectives are, and why, even if they are honest, they are likely still substantially undercounting Trump support.
Polls are expensive. Depending on whether it is a telephone poll or an online poll, on the number of questions, and the number of respondents, a poll can run from $20,000 to $75,000 or more. Pollsters, for the most part, don’t just poll because they feel like it. Someone is paying them. In the case of university polls, the pollster is usually an adjunct of a center, the political science department, or possibly the communications department. Private pollsters like Gallup earn their living by conducting year-round polling for corporations, private philanthropies, and other private sector groups. Say you are Planned Parenthood and you are under fire from a new law in, say, Arkansas. PP will hire a pollster to “show” that in fact, Arkansas residents favor abortion by a significant majority. And believe me, that is the result the pollster will get. Some of the margins may be closer than PP wants, but in the end, the customer is always right. This is true for any organization: the NRA does the same thing with polling.
With that in mind, when you see polling by MSNBC, NBC, CBS, Reuters, and so on, the customer is the network or “news” organization conducting the poll. None of them are conservative and none will suggest to the pollster that it, you know, just might be good to show Donald Trump losing. Don’t kid yourself: this is true for Fox, which has had some of the worst polling out there for years. In 2016 Fox polls routinely were worse than some of the polls from more liberal organizations.
How bad were these polls? I’ve been over this at length many times before, but at the national level, not one of the top five pollsters had Donald Trump winning in 2016. The polling was far, far worse in individual state polling. The average error in Wisconsin was six points. (To get that average, some pollsters were off by a galaxy). The average number for Donald Trump in Ohio was around four—yet he won by more than double that. Strangely, USC/Dornsife, which had the correct results (Trump winning) was also wrong in that it had Trump winning the popular vote by about one percent. As frequently mentioned, only the state-level pollster Trafalgar and Richard Baris’s (then-called) People’s Pundit Daily had Trump winning Pennsylvania and Michigan. (Baris actually had Trump ahead in Wisconsin but told me he didn’t trust his own numbers and on election eve put Wisconsin in Hillary’s column). Rasmussen had Trump up nationally—but he lost by over a point.
The reason Trafalgar was able to “call” Michigan and Pennsylvania when no one else but Baris could, was that Trafalgar used an interesting polling technique. In addition to asking “Who do you plan to vote for?” they asked “Who do you think your neighbor will vote for?” and reconciled the two. This gave rise to the “shy voter” effect, where people projected onto their neighbors their own plans to vote for Trump, but did not want to tell a pollster of their intent. Baris achieved the same end but through much more in-depth “likely voter” analysis.
So, most polls are done with an objective in mind—to please the client—and if the client is MSNBC, CNN, NBC and so on you know what will please the client. It won’t be a Trump lead.
But the most critical element of polling is even when companies are honest—like Trafalgar or Baris’s new Big Data Poll—there are still profound difficulties in polling, most notably finding the “likely” voter as opposed to just someone who is registered to vote. By this point in the campaign, most pollsters are shifting to “likely” rather than “registered” voters. The problem here is finding out who will actually go vote. It will be far more difficult this year than ever before. Both William Cahaly of Trafalgar and Richard Baris have said publicly that they are having much more difficulty getting Trump voters to come forward. In some cases, people will tell them upfront they will vote for Trump but won’t say so in the poll. Obviously, Black Lives Matter, with its penchant for burning down cities, and the fascist organization called “Antifa” with (now) murder on its hands are strong motivators for people to keep their intentions about voting quiet until election day.
But even beyond finding the shy Trump voter—which both Baris and Cahaly say is twice as hard this cycle—there are other issues plaguing the samples. Most notably, the China Virus has caused massive voter shifts. The generally accepted view is that Democrats, being more fearful of the virus than Republicans, will vote by mail or absentee and that Republicans (who traditionally do better on election day) will have a virtual tsunami on November 3. When it comes to voting by mail—leaving fraud aside for a moment—this is fraught with peril for the Democrats. While it seems easier, it turns out to be a little more difficult to get people to actually sit down and mail in their ballots. It remains to be seen if 2020 offers a significant departure from that history.
An even larger problem for the Democrats, and one that I’ve written about, will be the inability to mobilize and organize the 14 million student voters because large numbers of American campuses will be closed. By my estimates here in an earlier column I estimated the Democrat student vote would be down 30%. Many campuses, however, are already closed to on-campus life entirely (Michigan State, Pomona, Harvard, Princeton, the University of California system, the Cal-State system, University of Southern California, and others). My estimates on student turnout may be way off. The “youts” could be down 50% from their 2018 numbers. Belatedly, two other outlets figured this out, but still not most of them.
And today, the betting odds—which rely heavily on polls—shifted for the first time in Trump’s favor. The result is that the left is finally figuring out that election night will be bloody for them, which is the topic of my Friday column
Larry Schweikart is the co-author with Michael Allen of the New York Times #1 bestseller, A Patriot’s History of the United States, author of Reagan: The American President, and founder of Wild World of History, a history curriculum site for grades 9-12 that features full curricula in US and World history that includes teacher guides, student workbooks, tests, maps/graphs, and video lessons coordinated with the reading (www.wildworldofhistory.com).