• IAmTheZeke@lemmy.world
    link
    fedilink
    arrow-up
    40
    ·
    3 months ago

    Polling guru Nate Silver and his election prediction model gave Donald Trump a 63.8% chance of winning the electoral college in an update to his latest election forecast on Sunday, after a NYT-Siena College poll found Donald Trump leading Vice President Kamala Harris by 1 percentage point.

    He’s just a guy analizing the polls. The source is Fox News. He mentions in the article that tomorrow’s debate could make that poll not matter.

    Should you trust Nate or polls? They’re fun but… Who is answering these polls? Who wants to answer them before even October?

    So yeah take it seriously that a poll found that a lot of support for Trump exists. But it’s just a moment of time for whoever they polled. Tomorrow’s response will be a much better indication of any momentum.

    • ⓝⓞ🅞🅝🅔@lemmy.ca
      link
      fedilink
      arrow-up
      14
      ·
      3 months ago

      It just seems strange because I don’t think that many people are on the fence. Perhaps I’m crazy, but I feel most people know exactly who they’re voting for already. Makes me wonder how valid this cross-section was that was used as the sample set. If it accurately represents the US, including undecided voters, then… 😮

      • randon31415@lemmy.world
        link
        fedilink
        arrow-up
        17
        ·
        3 months ago

        but I feel most people know exactly who they’re voting for already

        The cross-section of people you know are more politically off the fence than the entire nation. Those that aren’t online at all are also more undecided and less likely to interact with you.

      • someguy3@lemmy.world
        link
        fedilink
        arrow-up
        14
        ·
        3 months ago

        I listen to those news things that interview people on the street and I’m amazed at how many are uninformed and can go either way.

        • zabadoh@ani.social
          link
          fedilink
          arrow-up
          6
          ·
          3 months ago

          There’s a Trump undercount in polling: Trump voters don’t trust “MSM” and therefore don’t answer calls from pollsters, or are embarrassed to admit they will vote for him.

          Same goes for asking random people on the street.

            • actually@lemmy.world
              link
              fedilink
              arrow-up
              4
              ·
              3 months ago

              I don’t know many people (boomers and younger) who answer the phone from numbers they do not recognize. I would like to imagine that the people who do answer strange numbers tend to be out of touch. Bias in the polls to fools or the lucky who are not spammed ?

            • TehWorld@lemmy.world
              link
              fedilink
              English
              arrow-up
              2
              ·
              3 months ago

              And an undercount of women who are telling their husbands and anyone else who asks that they’ll be voting Trump, but will actually vote for Harris when the time comes. And an undercount of bro-ski-s who claim to support Harris, but secretly hate the fact that they can’t get a ‘female’ that will cater to their every whim and will vote Trump because he’ll increase oppression of women. And an undercount of cat ladies… etc. Most “high quality” models at least attempt to mitigate these over and undercounts, which definitely skews results, and why poll aggregators are important. It helps to eliminate biases in polling types. There’s really only ONE poll that matters. VOTE! BRING YOUR FRIENDS!

          • SpaceCowboy@lemmy.ca
            link
            fedilink
            arrow-up
            6
            ·
            3 months ago

            Pollsters are compensating for that undercount of unlikely voters. 2016 they were low, 2020 still low but pretty close. They will have scaled it up to be more accurate this go around.

            Except there’s a few snags there. In between the 2020 election and now, there was an insurrection, Roe v. Wade was overturned, Trump was convicted of crimes and indicted for many more. These are things that a statistical process can’t really account for when putting weight on how likely a respondant is to actually vote.

            Trump lost in 2020. Do all of these events incentivize more people will turn out for him this time than in the last election? Or will less people turn out for him?

            Every time something unprecedented happens it negatively impacts the ability for a scientific statistical process to predict the outcome. Science can’t predict things there’s no model for, and how do you can’t have a model for something you haven’t seen before. And a hell of a lot of unprecedented shit has happened. Maybe next time a convicted felon that tried to overthrow democracy runs in an election there can be accurate polling, but it’s not going to be the case in this election.

            There really is no way to know what will happen on election day. So there’s else to do other than maximum effort until election day.

      • bamboo@lemmy.blahaj.zone
        link
        fedilink
        English
        arrow-up
        12
        ·
        3 months ago

        The issue isn’t really people on the fence for Trump or Harris but mainly with generating turnout. After Biden’s poor debate performance, people didn’t change their mind and decide to vote for Trump, they became apathetic and maybe wouldn’t show up to vote.

        Harris doesn’t need to persuade people to abandon Trump, she needs to get people excited to show up to vote.

    • MonkRome@lemmy.world
      link
      fedilink
      English
      arrow-up
      6
      ·
      3 months ago

      He’s not polling, he is aggregating all of the polls into a prediction model. Either way it is just a snapshot in time.

    • Rhaedas@fedia.io
      link
      fedilink
      arrow-up
      5
      ·
      3 months ago

      The key to doing statistics well is to make sure you aren’t changing the results with any bias. This means enough samples, a good selection of samples, and weighing the outcome correctly. Even honest polling in pre-election is hard to get right, and because of that it’s easy to make things lean towards results if you want to get certain results, or or getting paid to get those results.

      There’s only one poll that matters, and that poll should include as large of a sample as possible, and be counted correctly. Even though some will try to prevent that from happening.