• Tips gratefully accepted here. Thanks!:

  • Recent Comments

    Propertius on Imagining a Fifteen-Minute Cha…
    jmac on Imagining a Fifteen-Minute Cha…
    Beata on Imagining a Fifteen-Minute Cha…
    Ivory Bill Woodpecke… on Imagining a Fifteen-Minute Cha…
    Ivory Bill Woodpecke… on The Year of the Scapegoat: Fre…
    Ivory Bill Woodpecke… on The Year of the Scapegoat: Fre…
    William on Imagining a Fifteen-Minute Cha…
    Propertius on Imagining a Fifteen-Minute Cha…
    William on Imagining a Fifteen-Minute Cha…
    Beata on Imagining a Fifteen-Minute Cha…
    jmac on Imagining a Fifteen-Minute Cha…
    Beata on Imagining a Fifteen-Minute Cha…
    Beata on The Year of the Scapegoat: Fre…
    Propertius on The Year of the Scapegoat: Fre…
    Propertius on The Year of the Scapegoat: Fre…
  • Categories

  • Tags

    abortion Add new tag Afghanistan Al Franken Anglachel Atrios bankers Barack Obama Bernie Sanders big pharma Bill Clinton cocktails Conflucians Say Dailykos Democratic Party Democrats Digby DNC Donald Trump Donna Brazile Economy Elizabeth Warren feminism Florida Fox News General Glenn Beck Glenn Greenwald Goldman Sachs health care Health Care Reform Hillary Clinton Howard Dean John Edwards John McCain Jon Corzine Karl Rove Matt Taibbi Media medicare Michelle Obama Michigan misogyny Mitt Romney Morning Edition Morning News Links Nancy Pelosi New Jersey news NO WE WON'T Obama Obamacare OccupyWallStreet occupy wall street Open thread Paul Krugman Politics Presidential Election 2008 PUMA racism Republicans research Sarah Palin sexism Single Payer snark Social Security Supreme Court Terry Gross Texas Tim Geithner unemployment Wall Street WikiLeaks women
  • Archives

  • History

    September 2021
    S M T W T F S
  • RSS Paul Krugman: Conscience of a Liberal

    • An error has occurred; the feed is probably down. Try again later.
  • The Confluence

    The Confluence

  • RSS Suburban Guerrilla

  • RSS Ian Welsh

    • Open Thread
      Use to discuss topics unrelated to recent posts.
  • Top Posts

Do Political Polls Have Value? Or Do They Not? Choose One

It was very good news that Governor Newsom of California defeated the Recall attempt, and also great news that he did it in overwhelming fashion. I was very worried about this vote, at least five weeks ago, when the first poll in a while, conducted by Survey USA, showed a turnaround from earlier polls by other pollsters, which had Newsom about ten points ahead. SUSA’s poll had the Recall winning by about six percent, something like 51-45%

Much was made of this poll by other media. I certainly reacted to it. I had seen Grey Davis lose his lead against the Recall, and lose rather substantially to Arnold Schwarzenegger. I knew that it could happen again. I knew that Republicans would likely have a greater voting enthusiasm than Democrats. This worried me a great deal. Not only would it be a tragedy for the state and nation, I would have to move out of California. I was not going to live in the West Coast version of Florida, and see Larry Elder appoint a Far Right zealot to replace Dianne Feinstein if she had to retire before the next election. And trends are so important in elections, particularly in an expected low turnout election. And if the SUSA poll was accurate, things looked very bad for Newsom, and I wrote that it looked as if he would lose.

Well, he did not lose, he won in a tremendous landslide. He had a lot of campaign funds to spend, and he used them well. He started to lead in polls done a couple of weeks before the election. The last poll I saw about a week before the election had him up about 14%. My worries were allayed, and I was confident, though wary. And as we saw, Newsom is winning by something like 25%, a resounding defeat for Trump, Trumpism, Trumpists, and Republicans in general, who are virtually synonymous with the other three.

But why was the SUSA poll absolutely wrong? Well, it was done a month or so before the election. And there was another one about the same time, showing Newsom ahead by only a few points. So maybe things changed drastically. But there was another poll on Tuesday, an exit poll which showed that only about 8-10% had made up their mind in the last week, the rest already had. Well, if all of those who just decided, voted against the Recall, that would help account for the margin. But even so, it looks like the SUSA poll was not close to accurate. Mark Murphy, not my favorite, but at least not a Trumpist, was a former campaign manager and strategist for Republican Presidents. He said on the night of the election count, that this race was over three month ago. Well, how was Yes on Recall winning in a poll one month ago, unless the poll was very flawed, or Murphy was very wrong?

Now, there were so many polls during the last couple of presidential elections, that they all seem to blur. But I thought that SUSA was a decent polling organization. How could they have Newsom behind by six percent, when a month later he wins by twenty five percent?

There is a mathematical theory behind the validity of polling. I actually learned more about it in my Statistics courses at UCLA Graduate School of Management. I have forgotten some of the details, but the working out of the validity of polls is an elegant mathematical proof. It has to do with what is called the Normal Distribution, and how it predicts with relative accuracy, depending upon the number polled, and the integrity of the sample, the results of the much larger voting numbers. Underlying the entire theory of the accuracy of political polling, is the mathematics which shows that an accurate representative small sample can predict the actual voting outcomes.

Theoretically, the larger the sample, the less is the so-called margin of error, the chance that the poll will not reflect the actual results, were the election held on the days of the polling. As we always hear, a poll is just a snapshot at that particular moment, things can change. And sometimes they do, but that does not give the pollsters an easy out, if their polls ultimately prove inaccurate on election day.

It does seem as if national political polls have become less accurate. I recall that the Gallup Polls used to be very accurate when done a couple of days before a Presidential election. Sometimes they missed a late surge, such as Reagan in 1980, or to a lesser extent, Obama in 2008, but other times they were very accurate, as I believe was the case in 1968, and 1976, 1988, 1992, and a few others.

Now, in 2016, the general belief is that the polls vastly underrated Trump’s vote. But actually, to my great concern, those polls got very tight two days or so before the election, mostly due to James Comey’s unconscionable letter to the Republicans in the House about finding new emails, which was completely untrue, and should relegate him to infamy forever, for effectively putting someone in the Presidency who was a a literal nightmare for the world.

Back to the polls, they showed Hillary up by about 3-4%, and she won by about 2.8% of the popular vote, not enough to win the Electoral College, skewed toward Republicans,, because even the smallest states are given three electoral votes, and California, which should have something like 77 times the electoral votes of Wyoming, has only about 18 times the EVs. That is the true disgrace to our democracy, an inherently biased Electoral College which is biased toward the small, and almost inevitably Republican, states.

In 2020, the presidential polls were further off. Biden was leading by about 8-10% in almost every national poll right up to the election. There might have been one where he was winning by 6-7%.. But he only won by about 4.9%, and once again the bias of the Electoral College almost lost him the Presidency, in a race he won by seven million popular votes. That is terrifying in itself, but it also highlights something wrong with national polling.

The general “explanation” is that the pollsters are systematically underrating the Republican vote. I am not close enough to the data to know why this should be, and why the pollsters have not adjusted. Obviously, a poll is only accurate if it mirrors the actual demographic makeup. If it oversamples Democrats, or does not reflect heretofore non-voters who have become galvanized into voting, the polls are not reliable. At this point, I cannot be confident that they are accurate. I do not need polls, none of us not in the actual campaigns, do. We might be better without hearing them., particularly if they lead to assumptions and expectations which may be misleading or just inaccurate.

Of course, now the pollsters are going to have to weigh the vote suppression tactics of Republicans. We could see polls showing a Democrat in a Senate race be five points ahead, and lose by five points, because many Democrats had their votes not allowed or counted. What a horror that would be. Actually, the Senate polling was poor in 2020, as several races which looked good for Democrats, went four or more percent in the other direction, including North Carolina, and Maine. Why were Democratic voters overpolled? Is it possible that in some cases, there was cheating, or throwing out of votes on the Republican side.? Always accuse your enemies of doing what you are doing,” was said by the figure of pure evil, Joseph Goebbels.

I think we are all sick of internet polls, and also the ones we see which are just there for headlines or propaganda. “Sixty percent of people say they would choose job success over free time.” Those kind of things are so amorphous and subjective (How much success? How little free time?) as to mean nothing. They are a sort of pornography, if one wants to use that term. They draw clicks and viewers. But they lead to assumptions and conclusions which are not valid or warranted, at least by those vague or leading questions.

When I would tell my father about a particular poll in an important election, he would usually say that polls were not accurate. I would try to explain my perception that yes, the polls on subjective matters were flawed and sometimes deliberately intended to obtain a desired result, but that the science behind polling made simple “Which candidate are you voting for?” polls, done by respected pollsters, mostly reliable.. But maybe not even those are accurate now, and he was more right than I had thought.

Polling is not going away,though, and we will hear about new polls every day. Favorability ratings, preferences, most important issues, etc., etc. Who knows how accurate they are? Who can tell if extrapolations and conclusions made from them are warranted? Maybe we should have a moratorium on polling? How many are in favor of that, what are the numbers?

3 Responses

  1. To me, polling is just another tool those well versed in the science of marketing use to manipulate the public. Of course this is amplified by social media. As someone who had a long career in what is now commonly called IT, I find it sad that the technology was perverted.

    • Yes, it can be very unsettling to contemplate. I think back to a couple of weeks ago when the very odd Sam Stein touted a poll from Politico, the very unsettling organization which he now runs, where people were asked about whether they would be in favor of withdrawal from Afghanistan, if it meant that the Taliban would completely take over, and terrorism would increase. That is almost beyond the definition of a “push poll,” set up to elicit a certain response. And so much of the media is set up to accept such bogus polls as meaningful. As you say, it has been corrupted to become a propaganda device. We all learned to ignore the old “90% of toothpaste users prefer…,” but these deliberately skewed polls on policy issues or popularity of leaders, are very dangerous.

  2. I don’t know about you guys, but I *always* hang up on pollsters. I suspect a lot of other people do, too.

Comments are closed.

%d bloggers like this: