Before the last elections recede too far into the background, I think it is very important that there be a comprehensive analysis of why and how the polling was so far off the actual results. It was wrong in 2016, though after James Comey violated his responsibility of neutrality before an election, and pretended that there were “new emails,” Hillary Clinton dropped three points or so in the polls, so that right before election day, she was only a point or two ahead; and she did win the popular vote by 2.8%. Comey’s unconscionable actions will live in infamy, as he managed to get Trump elected; and virtually, and still possibly, destroy American democracy.’
In 2018, the polls somewhat underestimated the Democratic victory in House races In 2020, polls had shown Biden up around eight points, but his popular vote margin was about 4.3%. This caused some analysts to believe that the polls were systematically underrating the Trump vote; that a cohort of his supporters chose not to admit that they were going to vote for him. Another theory was that the models were wrong.
But that was far less glaring an error compared to what happened in 2022. Polls showed, or were interpreted to show, the coming ‘red wave,” which would overwhelmingly give control of Congress to Republicans. Various estimates had Republicans winning 30-40 House seats, and 3-4 Senate seats. Some predicted even greater margins, plus big Republican wins in key gubernatorial races such as Wisconsin and Arizona.
What actually happened, was that Republicans won eight Congressional seats; will lose one Senate seat; and lost virtually every important race for governor except Nevada. They even lost Kansas again.
Those results were gratifying, even if we greatly regret the loss of the House, which is essentially due to the fact that Republicans gerrymander their states relentlessly, while Democrats, always tying to do the virtuous thing, have essentially given over districting power to neutral commissions and state courts. I wrote extensively about that, and it has to stop if Democrats actually want to save the country. But even with that, the Congressional wins for Republicans mostly came from extreme gerrymandering in Florida, Texas, and Ohio; and then New York utterly botching the districting process, and forcing out a fairly popular governor to be replaced by a rather weak one, with no coattails. And then California’s neutral districting process actually helping Republicans I would greatly hope that Democratic leadership wakes up to this, and stops choosing “virtue” over winning elections and saving America and the planet.
Even with that, the extent to which the polling was wrong, was enormous. I tried to stay away from polls for a while, but then I found myself looking at them in the two weeks or so before the election; and they got worse, although they were consistently overstated by the media, and that is another big story which cannot be swept under the rug, though the media is trying to.
I did watch the news, specifically MSNBC, which is by far the least right-wing biased of the three cable news stations. And even they were almost consistently running stories which foretold disaster for Democrats. “Inflation is soaring, which is very bad news for Biden and Democrats.” “Abortion is receding as a major issue; crime is now second in importance for voters, after inflation.” “Democratic voters are not enthusiastic, while Republicans are.” Actually this last theme was mostly purveyed by the written media. Put together, there was virtually no place where anyone could get news, which did not have an overwhelming “news bias” in favor of a Republican big or even landslide victory.
There were a few people who were fighting against this tide. notably Simon Rosenberg, who has been a Democratic strategist, and Tom Bonier, who is more of a statistician. As soon as the early vote started, they would look at the statistics which showed a strong majority of early turnout favoring Democrats in key states. They believed that this portended better results for Democrats than the polls or the media forecast. This approach was discounted and even mocked by the likes of Nate Silver of “538,” and Dave Wasserman of “Redistrict” and The Cook Report, who said that early vote totals have no predictive value; and who along with others in written media, used the scornful term “hopium” to describe any argument which sought to suggest that there would not be the red wave.
Rosenberg started to decry some of the polls; he believed that Republicans were “flooding the zone” with polls, some very skewed toward them, in order to both distort the polling averages, and create despair among Democrats. We know that “Real Clear Politics,” which the eminent political writer Ronald Brownstein recently said was “a propaganda arm of the Republican Party, no more, no less,” always uses polling averages.If a sizable percentage of those polls are biased; that is, with skewed samples, those polls will affect the averages, which very unfortunately keep getting quoted by the television media.
As it turned out, Rosenberg and Bonier, and a computer person by the name of Christopher Bouzy, were far more accurate in their analysis than the vast majority of the “mainstream political sites,” and the mainstream media. This is to their immense credit, but it also calls for the very serious question as to whether Right-Wing Republican media bias is so entrenched in the media which they essentially own, that we cannot get an accurate sense not only of what is going on in the country, but how people are going to vote.
And of course it is very likely that we are looking at propaganda, the kind that totalitarian countries use to beat down the populace, and to brainwash them into believing their lies. We will not delve into the terrible power of endless propaganda, such as is seen on Fox News, but it is horrifying; and it is largely the result of Ronald Reagan carrying out the wishes of his very rich and very biased backers, and getting rid of the “Fairness Doctrine,”which essentially required news networks to provide balanced coverage. So Fox came up, and they were awful from the start, and are now even worse; a streaming faucet of lies and slanders, all intended to poison and control the minds of their slavish viewers.
I never, ever watch Fox, but it is there. And when CNN has turned hard Right, that only leaves MSNBC, which generally tries, but even they fell prey to the right-wing polling bias and propaganda. NPR, which I guess used to be good, is now apparently also a mouthpiece for a kind of pseudo-scholarly Republican narrative. One can turn off most of this, and watch selectively, but the effect of the propaganda is still there.
Still, the question remains: how were the polls so wrong, and what were the polling algorithms which led to such wrong predictions? We know that the science of poling is based on a statistically proven theory that effectively sampling a subset of a population can give a fairly accurate view of the entire population. The larger the subset, the less the “margin of error.” But of course this presumes that subset is accurately chosen in terms of demographics: age, ethnicity, gender, and other factors.
It was somewhat amusing, albeit upsetting, to look at the various pre-election polls, and see that Trafalgar, for example, systematically showed a Republican bias. All you had to do was to look at one of their daily polls showing a very low approval rating for Biden, or a Republican lead of four points or so on the generic Congressional poll, and know that this same sample would yield leads for the Republican in any race they polled. The sample was skewed, almost certainly deliberately. And there were several of these, and they were all inaccurate.
Polling cannot be expected to always be accurate. As is always said, a poll is an instant snapshot. But either some of these pollsters have an agenda intended to influence and dispirit voters, or they are simply bad at polling. The inaccuracy of these polls cannot be just tossed off with a shrug.
I would hope that there would be some very scholarly and in-depth pieces from those who are expert in statistical analysis, about what led to these consistent errors. Maybe some of it was due to not taking account of younger and even first-time voters. Maybe inaccurate demographic analysis. But it was not something to be ignored, because bad polling, deliberate or simply flawed, has consequences. Could Democrats have done even better, if some people had not been discouraged from standing in line for four hours, by thinking that their candidates would lose? Just because we did better than anticipated, does not mean that these polls did not have some of their intended effects.
Not too many people read statistical analyses. But I, not being an expert in this field, would want to see the following: 1) The media should stop quoting Real Clear Politics polling averages. 2) All polls but the most respected ones should be treated with skepticism. 3) The Republican control of the media should be explored by those few independent media sources,so that some people might at least become more aware of it. 4) Put people like Rosenberg and Bonie and Bouzy on television to give a different view of what the polls are saying. 5) Stop talking about polls so much. Biden’s “Unfavorable” numbers, reported on every single day, did not seem to predict the election results at all. And the “enthusiasm gap” seemed to disappear in the last week before the election, but only Rosenberg pointed it out, as far as I know.
Polls can be fun, but like everything else in our society, Republicans have shown that they can be used as a propaganda tool toward their fascist ends. Did you see that Elon Musk apparently ran a poll as to whether those who voted, which probably largely consisted of his group of followers, and bots, favored Donald Trump being reinstated to Twitter. After that, he said “The people have spoken!” That kind of poll, the kind they do in autocracies, is no more than a self-perpetuating device to rationalize anything that you want to pretend the general public favors. And far too many polls have become like that, as we have seen.
Filed under: General | 10 Comments »