Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Lord Shonus]

Polls were always going to tighten as we head into the final stretch (they always do), much of the claimed GOP rebound is driven by a single poll, that uses a different methodolgy than all the previous ones. Which means that it cannot be directly compared to what we've been seeing for weeks. Not to mention that even the most reputable poll is still a single data point. Not to mention that even revised projections are only disappointing by the standards of "we were hoping to hold one house and make gains in the other", which is incredibly difficult for the incumbent party in a midterm. As the polls look now, the Dems still have about a two out of three chance of holding the Senate, which is incredibly unusual in a midterm. Even the GOP taking a narrow win in both chambers wouldn't be a "red wave" - the "expected" result based on historical trends is far more drastic than that.

Also, most analysis organizations are building historical trends into their predictions, and who knows if that's even meaningful at this point.

[Vector]

@Shonus: Thank you, that actually helps a lot in terms of context. The NYT commentary here is basically "hue hue hue you're all gonna die, red wave red wave, let's sell a buncha papers," which, not helpful.

[JoshuaFH]

What the hell is happening over there Vector? Is there a Tsunami of dinoflagellates inbound to your location?

[Loud Whispers]

I thought the voting patterns in the USA was that people no longer changed who they voted for in significant numbers, and had calcified into war camps. What could the repubs possibly do to persuade demvoters, or mobilise repubs?

[Vector]

The New York Times, I have noticed, has some excellent reportage on certain subjects due to unprecedented access. But it seems to be taking the position that "fascism is fun and fear sells papers" and is permitting its comment sections to be bombarded with anti-trans bigotry.

I made the red wave comment based on the NYT article I read, which went light on the analysis and heavy on the fear.

[Lord Shonus]

That's always a factor (the effect that the New York Times is postulating based on a poll of around 800 people is that Republican women were driven away from the party due to Dobbs but are sticking with tribalism now), but there is always a contingent of voters that vote on the fringes. Smaller than it used to be, but still there.

The more important factor is who votes in the first place. Pennsylvania Republicans who aren't enthusiastic about Dr. Oz aren't going to switch votes and go for Fetterman, but they're quite likely to simply stay home. That's the main thing that's been hammering the GOP in polls so far, a lot of women who were politics apathetic were outraged by Dobbs and are mobilized. "Likely Voter" polls are tending to undervalue this group because they don't have a strong voting history.

A big problem this administration is that the media absolutely got addicted to Trump. His constant insanity meant that reporters could do their jobs with relatively little work, and there was always something to bump up a slow news day. They're also, unfortunately, noting that their biggest markets are older conservatives (even for the traditionally liberal papers) and have been adjusting accordingly.

[McTraveller]

One of these days I'll read up on polling science. I know it's not like say statistics for engineering, because humans are most definitely not politically homogenous or randomly distributed; I'm very curious to know how you don't end up with massive sampling bias in any kind of poll of a sub-population compared to the "actual" elections.  I feel like you'd have to actually know the true distribution to be able to remove the bias - which means you'd basically have to have conducted the actual election.

[nenjin]

Well, sampling bias tends to come from not having enough money, time, integrity, w/e.

And the science and math around statistics is pretty interesting, and a little terrifying for what it says about our ability to predict human behavior.

I barely passed college statistics so I'm by no means an expert. Vector could tell you what an actual mathematician would say. But what I came away with is that, for a fairly designed question and a large enough sample, you can get to about 95% surety on outcomes. I can't remember where the magic number is, I think it's around 20,0000 samples....? to get there. Takes increasingly larger sample sizes to get up to 97, 98% confidence. Most polls are not that though. And the news will report just about anything that calls itself a poll because they need something to talk about.

Also, because how much money Trump has been making in donations in 2020 is in the news cycle, learn about one of the billionaires who thinks we should tear down democracy and is backing candidates to that end.

[None]

A reminder, of course, that it's very important you choose your sample carefully. 30 may be statistically significant, and 20,000 may seem like an enormous sample size, but if your sample population is a facebook group for 'Moms who knit tiny sweaters for dogs' and your sample itself is self-reported/opt in from this group, it's still a very narrow representation of any greater population as a whole.

It's like every time Fox purports that Biden is the most unpopular president yet- yeah, according to whom, Fox viewers? Perhaps the Pew Research Center (originating from conservative family and founders of the Sunoco oil company), which is funded in parts by the Templeton Foundation (seeking to increase the intersection of religion and science, supporting christian bias in religious studies, promoting intelligent design, and funding conservative think tanks) and the Pew Charitable Trusts, which funds the Texas Public Policy Foundation, which supports school vouchers, promote fossil fuel use, and reject climate change?

Now, I don't know the metrics nor the article from the New York Times, so I won't say that the poll results shouldn't be alarming, but it's generally very important to scrutinize the process behind a poll, who they're asking, and why. And also to take a good hard look at who claims to be nonpartisan and what their motivations are (or where their money comes from).

[The_Explorer]

polls are like aiming for the toilet while trying to impress your friends with how good your aim is by standing a few feet away from it. Some of it may go in, but mostly not

[Loud Whispers]

polls are like aiming for the toilet while trying to impress your friends with how good your aim is by standing a few feet away from it. Some of it may go in, but mostly not

Not to a champion pollster

[scriver]

Or a champion pisster!

[Loud Whispers]

Or a champion pisster!

That lends a disturbing spin to the old phrase taking the piss

[Max™]

Please don't spin while taking the piss, but if you do I'm not cleaning it off the walls for you.

[Loud Whispers]

Initiating spin

[interstellar theme starts playing]