BROKEN: Can you understand what you read in the Times?

WEDNESDAY, FEBRUARY 28, 2018

Nate Cohn strikes again:
Another day, another hours-long struggle with a Nate Cohn "Upshot" piece!

To review yesterday's struggle, click here.
Seriously, can anyone paraphrase Cohn's heavily jargonized prose?

We were able to glean one clear (and surprising) factual claim from this morning's report. Here it is, in our own words, as we understand it:
According to Cohn, 45 percent of the voters in the November 2016 election were "whites without a college degree." But the official exit poll, conducted by Edison, mistakenly reported the figure as 34 percent.
That's a very large difference. We can't say if Cohn's claim is true or false. Nor do we feel sure we could paraphrase anything else he wrote.

As was true yesterday, so too today. Cohn's piece seemed to be written in highly jargonized prose. Consider this passage, which follows his claim about the large exit poll error:
COHN (2/28/18): How could the exit poll be off by so much? In fact, most polls are represented by too many well-educated respondents; most high-quality surveys weight their samples to adequately represent less educated voters. But the exit polls aren’t weighted by education, and so they start and end with far too many well-educated voters. (The exit polls are weighted only by demographic characteristics that the exit poll interviewer can visually identify, like gender, whether someone is African-American, and a rough guess of age; they can’t guess education.)
"Most polls are represented by too many well-educated respondents?" That lingo strikes us as odd.

How can a poll be "represented by" a group of respondents? Does that highlighted statement maybe mean something like this?
In fact, most polls sample, or interview, too many well-educated respondents.

(Perhaps because "better-educated" people are more likely to respond to a poll, while less-educated people are more likely to refuse participation?)
Is that what that clumsy locution meant? That would be our assumption, but as Cohn continues, we find ourselves even more bollixed:
COHN (continuing directly): This winds up biasing the rest of the survey because the exit polls are weighted to match the actual result of a far less educated country. In general, the exit polls underestimate Republican support, probably in no small part because they overrepresent young, nonwhite and well-educated voters. But this process leaves the underlying educational bias of the sample intact, and the result is that Republican-leaning voters are given more weight to compensate for an electorate that represents Democratic-leaning voting groups.
Do you feel sure you know what's been said? The terms "biasing" and "weighted" add a bit of instant fuzz to the brew, and then we encounter this:
"In general, the exit polls underestimate Republican support, probably in no small part because they overrepresent young, nonwhite and well-educated voters."
In general, the exit polls "underestimate Republican support?" They underestimate Republican support for whom? Every time you think you know what he's saying, he throws in a puzzler like that.

Staff at Starbucks have started to worry as they watch us struggle with Cohn each morning. We'll ask the same question we asked yesterday:

Cohn is discussing significant topics. Does anyone edit his work?

Another confounding problem: We read Cohn's piece this very morning in our hard-copy Times, right there on A17.

Online, in the Today's Paper listing, Cohn's report doesn't appear!

16 comments:

  1. Hordes of deplorable Morlocks are lurking in the dark, confusing reputable liberal scientists.

    Something's gotta be done about that.

    ReplyDelete
  2. Mr Somerby, Why don't you just drop this ignorant series on Cohn's article? Plenty of people understood it. Obviously, you ultimately think the fight over re-districting is just liberals whining again because they are such losers. You could spare the 1000's of words and just cut to the chase.

    ReplyDelete
  3. Cohn actually writes a fairly intelligent and thoughful column. IF Somerby was genuine interested in thoughtful newspaper articles (as he claims to be), he would actually try and understand this article, even if he disagreed with it.

    But Somerby
    1) Doesn't know and can't be bothered to find out even basic stats.
    2) Nitpicks over Cohn's not mentioning something that should be obvious to anyone with a double digit IQ.


    Somerby's goal is to attack liberals and defend Trump. He is a Trumptard to his very core - an ignorant liar.

    ReplyDelete
    Replies
    1. This is, I believe, is your second recent comment taking as a given that Somerby (and by implication, similarly educated readers) should be willing to review the material covered in the early weeks of a specific college survey course in order to make his way through a newspaper article. If you want to add something useful to the discussion 1:58 PM best you hold off on doing so until after you've reoriented yourself with the reality down here during one of you're infrequent visits to planet Earth.

      Delete
    2. Anyone who has any remote interest in politics (or claims to have remote interest in politics) should know enough stats to understand what Cohn said. In fact, high school math should be sufficient.

      But even those who don't know that much can find out the jargon through the wonderful world of Google. No one is asking them to figure out the derivation of mathematical formula.

      In particular anyone who possesses some humility should avoid taking potshots at a column just because they are innumerate. Clearly, Somerby doesn't think so, and apaprently you don't either. Have fun deep underground on Earth with the other Know Nothings !

      Delete
    3. So @CMike, every article must be dumbed down so that Somerby can understand it without hurting his little noggin? Regardless of how complicated the topic may be. Otherwise, it is the WRITER's fault, never the reader's.

      Besides, I don't completely buy Somerby's feigned ignorance. Did you yourself try to read and understand Cohn's article, or did you just immediately dismiss it because Somerby claimed it was opaque?

      Delete
    4. And to add to my message of 5:16, the basic point that Cohn makes is easy to understand, and reasonably apolitical. If Bob disagrees with Cohns basic point, he should dig into the data and make a coherent point. But Bob's sole interest is in complaining about how liberal. newspapers don't cover serious topics, then bitching because he's too stupid to understand a serious topic.

      Delete
    5. @5:16 Bob’s right, the article was extremely fuzzy.

      “…most high-quality surveys weight their samples to adequately represent less educated voters.”

      Really? How do they do that? Am I supposed to research how research firms “weight” their polling? Seems to me it would take little time on Cohn’s part to explain.

      Continuing directly:

      “But the exit polls aren’t weighted by education, and so they start and end with far too many well-educated voters. (The exit polls are weighted only by demographic characteristics that the exit poll interviewer can visually identify, like gender, whether someone is African-American, and a rough guess of age; they can’t guess education.)”

      If exit polls aren’t “weighted” by education, then how do they end up with “far too many well-educated voters?” If they’re only represented by “demographic characteristics,” why does that happen?

      “This winds up biasing the rest of the survey because the exit polls are weighted to match the actual result of a far less educated country.”

      Wait, the exit polls are “weighted” as well? To match the actual result of a far less educated country? Cohn had just got finished saying that the exit polls over-represent educated voters.

      “In general, the exit polls underestimate Republican support, probably in no small part because they over-represent young, nonwhite and well-educated voters. But this process leaves the underlying educational bias of the sample intact, and the result is that Republican-leaning voters are given more weight to compensate for an electorate that represents Democratic-leaning voting groups.”

      Look, if that’s all clear to you, hats-off, but I’m with Bob on this one, media critic that he is. This is a confusing mishmash. If all polls are “weighted,” then a statistical nudging has been achieved, nothing more. I question the use of polls generally, because generally the pols don’t give a damn about what the electorate actually wants from them. They only care about their own electoral prospects.

      Leroy

      Delete
    6. Compared to the average Times reader Bob Somerby is innumerate... sure, go with that.

      (By the way, in thirty years I don't remember ever seeing an elementary explanation of the meaning or implication of the term "margin of error" for a poll anywhere in any MainStreamMedia report. Maybe everybody should Google that- but don't follow the link to Wikipedia unless you have a little bit of background in statistics, in which case you would all ready have an adequate understanding of what it implies.)

      Delete
    7. '“…most high-quality surveys weight their samples to adequately represent less educated voters.”

      Really? How do they do that? Am I supposed to research how research firms “weight” their polling? Seems to me it would take little time on Cohn’s part to explain.'

      Seems to me it's not particularly relevant to the main point that Cohn was making, so he didn't go out of his way to explain it. In any case, I think different polling firms use different methodlogies, and some may not even reveal how they decide what the weighting should be (as a trade secret). Perhaps if Cohn were writing a research paper, he might have gone out of his way to go into this in detail.

      Your other comment about weighting is actually far more reasonable tham Somerby's mindless bitching. To simplify greatly, exit polls oversample educated people.

      Top line results are finally adjusted based on actual results. So if Rs are undersampled initially during exit polling, but educated people are oversampled, and then when Rs are scaled up to match actual results, the educational bias makes it appear that more educated whites voted R.

      Cohn may or may not be correct, but it's a reasonable argument, and he is using the Pew study to backup his argument.

      In short, it's the sort of intelligent thoughtful article that anyone interested in thoughtful articles should applaud. Not Somerby, whose main goal is to defend Trump and attack liberals.

      Delete
    8. Discrepancies such as these can be partially explained by the fact that the data come from the responses of humans and are gathered by other humans. Don’t discount the fact that people that are being surveyed don’t always tell the truth. Consider that when Americans are polled whether they regularly attend church—and regular attendance is carefully defined—the results are usually around 40%. When actual church attendance is directly measured, the measurement is generally half of that or less.

      https://churchleaders.com/pastors/pastor-articles/139575-7-startling-facts-an-up-close-look-at-church-attendance-in-america.html

      If people are willing to fudge whether they go to church or not, they certainly might do the same regarding whether they have a degree or not. They might even fudge answers about their own beliefs.

      I once asked a very close, very conservative cousin whether he believed Obama was born in Hawaii and he said, “Of course.” When I asked if he was surveyed what would he say, he said that he would give the opposite answer (he used more colorful language).

      Regarding the pollsters, sometimes they can be part of the problem. For example, some might ask specifically whether you have a bachelor’s degree or above, while some might ask a little less precisely if you are college educated. Many people with associate’s degrees or even people that just took a couple of courses would likely answer the latter question in the affirmative.

      Also, as has been pointed out by others, people belonging to some demographic groups might be more willing to answer a pollster's questions, over-representing those groups.

      Delete
  4. Trump is a liar and a thief.

    ReplyDelete
  5. "The terms "biasing" and "weighted" add a bit of instant fuzz to the brew."

    Only to someone surrounds himself in writing with wailing imaginary friends because he has done no work with real people in real campaigns.

    "In general, the exit polls "underestimate Republican support?" They underestimate Republican support for whom? Every time you think you know what he's saying, he throws in a puzzler like that."

    This shows more ignorance that Bob's boneheaded complaint yesterday about five statewide races in Pennsylvania.

    ReplyDelete
    Replies
    1. Let's try to interpret this sentence:
      "In general, the exit polls underestimate Republican support."

      Perhaps "Republican support" doesn't mean support for someone by Republicans. It might mean support for Republicans by voters.

      So the sentence could be equivalent to:
      "In general, the exit polls underestimate support for Republicans."

      In fact, that's the interpretation that occurred to me the first time I read it. I don't agree with Bob, who calls it a puzzler.

      Delete
  6. From Joe Lenski of Edison Research.
    ““Since we are at the polling place, our interviewers can record some characteristics of voters who decline to take part in the survey – approximate age, gender and race – so we can adjust our results for those factors,” Lenski said. “But there are other items we can’t account for visually. People don’t have a D or an R stamped on their foreheads, and you don’t know as they walk past you whether or not they have a college degree.”
    Representing reticent Republicans?

    ReplyDelete