Why does the GOP keep losing? Because polls are broken and Trump is "electoral poison" (video)

I think Galen’s talking about using the results of primaries, and various state and by-elections to make predictions about the big one. Eg. Trump seriously underperformed in the primaries, and Democrats are winning local elections in deep-Red areas, so he’ll probably make a similarly poor showing in the main election etc


Yes. And no-fault divorce. I’m sure taking away a woman’s right to have her own bank account or credit card is in their plans too. None of this is hyperbole. They’re very determined on these matters.

There’s no appeal in these policies to the vast majority of younger voters, so this election is their last chance to get into office and wipe out or neuter liberal-democratic institutions so that they continue what they started with Dobbs.

I wish he was required to handle fresh sheep entrails as part of his job.


And that sums up why they’re not accurate, like any push-poll. The polling org only gets replies from motivated respondents, and not an actual random sample of the population.

If I got a text message asking to participate in a poll where the question was “do you enjoy participating in text-based polling?”, I’d ignore it, as would most people. And the poll results would show that over 90% of people enjoy participating in text-based polling.

The poll result will depend entirely and disproportionately on whether the question pushes a person’s buttons. Maybe a useful metric is what percentage of people respond at all to a given question. But on its own, the percentage yes/no is pretty meaningless in polls like this.


Technology changes have definitely undermined the accuracy and usefulness of polls. I think generational shifts are a second-level cause; that is, if landlines and fax were still the only means for people to reach people, younger voters would still be as likely to receive telephone survey calls as they were 50 years ago.

My personal opinion is that polls march on because of the 24-hour news cycle. American news programming is first and foremost entertainment, and polls, no matter how flawed, provide a quick “Breaking News!” headline and a number-loaded circus poster graphic they can repeat endlessly while breathless announcers tell us how significant it all is.

The 24-hour news model is built upon the notion of large, more or less homogeneous audiences, which online audiences are not. I suspect that as TV and cable news die out polls will probably fade with them.


I’ve mentioned this before, but I read Nate Silver’s book he published around that time (I think it came out in 2012) and he relates a story in that book from when he got into online poker. He talks about having great initial success and thinking he was a much better poker player than he was, and then got brought down to Earth with a series of losses. He used that story to then talk about the importance of statistics, ironically completely ignoring the possibility that his early success in predicting the results of Presidential elections might be subject to the same bias. And, obviously, it was. And you’re right, to this day, he has refused to acknowledge that he got 2016 really wrong.


Thing is, as wrong as Silver has been about many, many things, 538 wasn’t wrong about 2016 and the presidential election result. They showed Hillary up significantly the week before the election - then accurately showed the race shifting to a draw after Comey’s Friday Surprise.


In 2016, I attended a talk given by a representative of the Australian Bureau of Statistics (ABS), who described the ABS as “beautifully optimized for a world that no longer exists”. It was an interesting walk-through of what kinds of polls and data were clearly gone, which were on their way out, and what they were trying instead. It was still very much an open question for social / technology use reasons, and I believe it remains so - nearly a decade later.


There are idealized elections, and then there are human elections with butterfly ballots, and handging chads, elections held in the middle of bad weather, and elections held during epidemics, elections in which weapons are brandished, and elections which don’t have enough ballots or polling places.

In that sense, the elections we have now are approximations, and sometimes the quality of those approximations is not very reassuring.


I think it’s always been bullshit, though? I mean, taking a random sampling of a much, much larger population is only going to tell you so much about that large population, period. All your really learning is what the people you’re polling believe. That’s it. Which is useful for… something. And how one frames questions can be leading, contradictory, confusing, etc. It’s attempt to make a science out of something that is inherently messy.

No audience is - that’s the biggest lie of the mass mediated era in general, that audiences are homogeneous and they’ve never been and never will be.


The year 2020 was one of the longest decades of my life, including the reagan-ridden 80s, FFS.


Justice Thomas, concurring in Dobbs, suggested the Court use its newfound disdain for stare decisis to reconsider Griswold v. Connecticut (which established the right of adults to obtain contraceptives) and Obergefell v. Hodges (which established the right of same-sex couples to marry). He was basically inviting challenges to these cases based on Dobbs, which of course he would gleefully approve.


You left out Loving v Virginia, legalizing interracial marriage. Wonder how Thomas would view that one?


I’d also accept being forced to breathe toxic fumes from a volcanic vent…


I’m sure he believes the leopards won’t eat his face dissolve his marriage.


He thinks he’s in the club and there’d be an exception for him if it were repealed.


This topic was automatically closed after 5 days. New replies are no longer allowed.