• mozz
    link
    fedilink
    206 months ago

    One fascinating statement hidden in the article:

    The Bloomberg News/Morning Consult poll of registered voters does not take into account how likely respondents are to vote in an election still more than six months away.

    I thought yesterday about posting the story Joe Biden Has Stunning 9-Point Lead Over Donald Trump Among Actual Voters, but I decided that reporting on the polls as an indication of the quality of the candidate is just as misleading when it’s pro Biden as anti Biden.

    But, if we are going to treat the polls as things as newsworthy as the news likes to treat them, that little statement is a pretty fuckin significant oversight.

    • @[email protected]OP
      link
      fedilink
      46 months ago

      And then your poll…

      However, the poll also found that 46 percent of overall respondents would vote for Trump in the presidential election, while 44 percent would vote for Biden when only offered the choice of these two candidates.

      • mozz
        link
        fedilink
        136 months ago

        Do you not think it’s relevant to attempt to figure out, when you’re polling voters, which ones of them are likely to actually vote in the upcoming election?

        I mean I know that Trump is at this point leading by some small handful of percentage points of overall respondents in most head-to-head matchups. My point was that limiting it to likely voters seems like it makes a pretty dramatic difference. No?

    • @[email protected]
      link
      fedilink
      English
      36 months ago

      From your article:

      However, the poll also found that 46 percent of overall respondents would vote for Trump in the presidential election, while 44 percent would vote for Biden when only offered the choice of these two candidates.

      You should read articles before posting them…

      You should also believe in science, and whether you agree with the polls or not, statistical analysis is a valid science.

      • mozz
        link
        fedilink
        106 months ago

        Yep, I read it. What was the question that led to +9 for Biden versus the question that led to +2 for Trump, again? It’s two different questions to analyze statistically, and knowing which is which is pretty valid, yes.

        • @[email protected]
          link
          fedilink
          English
          26 months ago

          It’s…

          It’s right there in what I quoted…

          However, the poll also found that 46 percent of overall respondents would vote for Trump in the presidential election, while 44 percent would vote for Biden when only offered the choice of these two candidates.

          What are you confused about?

          If the only two options are trump or Biden, more people will pick trump than Biden.

          • mozz
            link
            fedilink
            12
            edit-2
            6 months ago

            What are you confused about?

            Not confused. I’m asking you a question to see if you know the answer. Your first try (“when only offered the choice of these two candidates”) wasn’t the answer – when other candidates are included, Biden wins by 2% among all poll respondents. Want to try again or should I tell you? What’s the question that leads to +9% and how is it different from the one you quoted?

            • @[email protected]
              link
              fedilink
              English
              1
              edit-2
              6 months ago

              What was the question that led to +9 for Biden versus the question that led to +2 for Trump, again?

              I assumed you knew that since that’s why you were linking it…

              According to a Public Opinion Strategies poll for NBC News, the Democratic incumbent is ahead of his Republican challenger by 9 points among people who voted in the 2020 general election and 2022 mid-term elections.

              So it excluded everyone younger than 22 and people who didn’t vote in both 2020 and 2022…

              Do you not understand how big of a demographic that is?

              Non presidential years almost always see a dip in voters.

              But your article doesn’t link the to the poll, so I can’t tell you if that was head to head or with other candidates.

              • mozz
                link
                fedilink
                76 months ago

                I assumed you knew that since that’s why you were linking it…

                Obviously I knew that yes; as I already explained, that’s why I asked. Have you not seen this before, someone asking a question they know the answer to as part of a debate to see the other’s person’s response before taking the next step in the conversation?

                IDK, maybe I should change the way I talk to people on Lemmy. You seemed to be genuinely for-real confused by it and I’ve seen that before more than once (where people assume that I’m asking questions because I must not know anything about the topic).

                So it excluded everyone younger than 22 and people who didn’t vote in both 2020 and 2022…

                Do you not understand how big of a demographic that is?

                I do, yes. But I think that including it (including one factor that introduces, maybe imperfectly, an impact into the poll to account for different people having different probabilities of voting, instead of treating them all as the same) is better than treating all people as equally likely to vote, when clearly they are not. You wouldn’t agree with that?

                There’s a difference between discounting a whole demographic (we polled only whites and not blacks) and selecting particular people to poll based on criteria which make them statistically more likely to impact the election.

                No?

                • @[email protected]
                  link
                  fedilink
                  English
                  16 months ago

                  based on criteria which make them statistically more likely to impact the election.

                  No?

                  If it was just 2020, yeah, I could see that arguement.

                  But a non presidential election is always going to have lower turnout.

                  So I don’t see any worth in only counting people who voted in 2020 and 2022.

                  You know a big thing my graduate level statistical analyst prof told me the first day of class?

                  Anyone can find a weird sample to validate preconceived notions

                  Without seeing the poll (your article doesn’t link it) it seems safe to assume Newsweek looked for the highest pro Biden result, and presented as something they intentionally checked for.

                  It’s really really not uncommon.

                  And to be clear, this isn’t a problem with the data or polling practices, just in how sometimes the media picks their result first then hunts for the data to rationalize it.

                  • mozz
                    link
                    fedilink
                    4
                    edit-2
                    6 months ago

                    it seems safe to assume Newsweek looked for the highest pro Biden result, and presented as something they intentionally checked for.

                    Yeah, I pretty much agree with this. That’s the other reason I didn’t post the Newsweek article as a story. I’ve absolutely seen this from “the other side,” but that doesn’t mean that the answer is dueling cherry-picked samples. I only brought it up as a way of making the argument that failing to limit to only likely voters is a very significant flaw in OP’s poll.

                    To me, the factual analysis of which candidate people should be supporting based on how they’re performing is the main thing to look at, with how the polls are looking as sort of a distant tactical afterthought because it’s obviously relevant on some level to how the election is shaping up.