In the 2016 the major polls were horrifically wrong. All the national polls except Rasmussen had Hillary Clinton winning the election (both popular vote and electoral college). But Rasmussen was wrong as well, in that it had Trump winning the popular vote. Only the Big Data Poll by People’s Pundit Daily had Trump winning the electoral college—but Richard Baris didn’t poll nationally (however privately he told me he had Hillary winning the popular vote). In other words, NONE of the national polls were right, and all but one of them were wrong in the same direction.
This requires us to ask if they were wrong because they were incompetent, or if they were wrong because they were biased? I think most of us know the answer. The major pollsters almost all oversampled Democrats in every poll; they usually oversampled women, and often oversampled minorities. In some specific cases that I found, Suffolk (among others) occasionally asked for “the youngest voter in the home,” which is heavily biased toward Democrats. In short, the pollsters never had a chance at accuarcy, because of GIGO: Garbage In, Garbage Out. Rush Limbaugh, among others, has frequently argued that polling exists today to shape public opinion, not to test it.
But, historically, has anything really changed? We all remember the “Dewey Defeats Truman” headline in 1948. But was that really based on polls or what is called “exit polling?” More the latter.
I would wager, however, you never heard of this: in 1999, historian Thomas Mahl published a book called Desperate Deception in which he explored British attempts to bring the United States into the European war in 1940. As part of their efforts, the British Government infiltrated both Harris and Gallup, the two major American polling organizations of the day, with scores of British agents, often with the complicity of the Roosevelt administration. British agents were instructed to move American public opinion toward supporting the proposition that the U.S. would enter the war in Europe. To accomplish that, the polls were manipulated to show higher levels of support for intervention, and to ignore polls (or answers) that showed the public was against intervention. One poll, for example, was taken at a military veterans’ meeting.
The point is, if it was that easy for the British—a foreign power—to infiltrate and manipulate our domestic polling organizatios, how easy would it be for our own elites and or “law enforcement” agencies (the FBI or the CIA) to do so? How easy would it be to subtly shift all polls by 2-3 points (about the average error of all battleground state polls in 2016). For example, the Ohio polls were off by about three or four points; the Wisconsin average error was six points! (Pollster Richard Baris, who “called” Wisconsin for Hillary, said his own polls showed Donald Trump would win, but that since the margin was so big in all his competitors’ polls, “I didn’t believe my own polling,” he said.)
Understanding a poll requires you to get inside the “internals,” something very few people have the time or energy to do. But if you want to truly understand it, find out what the breakdown of Democrat/Republican/Independent voters is in the poll. Gallup (for what it’s worth) has it fluctuating between even Democrat/Republican and Democrat +2 or +3 over the GOP. Any poll that has seven or eight percent more Democrats than Republicans, or which has more than one or two percent more women than men, is utterly worthless. Whenever you see a poll that makes you doubt your common sense, just remember the British and Desperate Deception.