Originally published at: Biden surges ahead of Trump with RFK Jr. in the mix, according to new poll | Boing Boing
…
5 Likes
What happens if the Biden loses his seat?
2 Likes
Even if this woo peddler and hatemonger with a famous name seems to be peeling away votes from the other woo peddler and hatemonger with a famous name, the Dems still need to fight as if they’re in a tight race with the latter.
28 Likes
Thanks but I’ll hold my applause until after the election.
Criminy, we don’t even know who T-bag’s running mate is yet. And here that’s an important consideration. Of course if Trump voters had any sense they wouldn’t be Trump voters.
Here’s hoping I can vote for a Millennial in 2028! Err, I ought to back up and just say “here’s hoping I can vote on 2028” 
32 Likes
Trump should offer to make Kennedy his VP, and then they could run on a straight anti-science, anti-reality ticket. “MAIA: Make America Ignorant Again”.
11 Likes
No poll is worth paying attention to. Not even one that purports to show our favored candidate is ahead. And ESPECIALLY not one taken more than a year before the election in question.
18 Likes
No individual poll is worth paying attention to. However, many polls from various sources taken in aggregate can give a good picture of where things are headed. It will be interesting to see how this shakes out when other pollsters ask similar questions. They all have different methodologies and survey different people.
IIRC, many polls tend to show that people are open to supporting a third party ticket far out from elections, but election day gets closer people end up abandoning them for the GOP or Dems.
Polling isn’t meant to breed complacency, but in a functioning democracy it can serve as a check against electoral shenanigans.
9 Likes
I question all of the polls’ basic premise: that surveying around a thousand people can accurately predict how 150 million people are going to vote. I know statisticians study this stuff and claim that’s a good sample size to make that prediction, but I just don’t understand how, and I think the variability from poll to poll shows that. I just think people’s opinions and actions based on those opinions are harder to predict than that. Plus, I don’t trust that people are giving honest responses. Not everyone, anyway. Also, I think the publication of the poll results can impact peoples’ opinions, so while the measurement itself may not affect the results, the reporting of the measurement may affect future measurements. I think polling is broken right now, and I think there have been enough elections going differently than polling predicted to prove that.
I cannot for the life of me figure out how 16% of voters would pick RFK, Jr. right now. Anyone who has listened to what he says and likes it would probably vote for Trump because it’s a similar message. Anyone who has listened to it and doesn’t like it is voting for Biden. And the last group would be people who haven’t heard anything he has to say, and I don’t understand why those people wouldn’t otherwise be Biden voters because they’re likely choosing RFK, Jr. because of his name, which would seem to indicate they’re traditional Democratic voters. I don’t know. I haven’t trusted polls since 2016, and I’ve seen nothing here to change my mind on that.
14 Likes
The math only works in theory if those thousand people are chosen completely at random. But that’s not really possible. And they don’t really have a way to compensate for the lack of randomness.
10 Likes
Yep. And people are also capable of not being honest. They could be lying intentionally, and they could be lying to themselves (saying they’re undecided when really they aren’t…they’re just hoping a better option presents itself at some point), and the math can’t compensate for that, either. I mean, I guess the margin of error allegedly accounts for both of those, and the margin of error here is 3.9%, but even that seems sus to me.
12 Likes
In addition to everything you mention, there’s also the fact that a lot of surveys still rely on landline calls. That means that the respondents are going to skew in certain directions in terms of age, income, and other characteristics.
12 Likes
This one says it was a combination of phone, text, and internet, but I didn’t see a breakdown of how much of each there was.
9 Likes
Well, so this is an individual poll…
I’ve seen several and they are all WILDLY different. Which, once again, lends itself to the view that polls taken a year before an election are inherently useless drivel.
Polling when done well maybe. When done in a timely fashion just before the vote maybe. Polling done slapdash a year before voting happens? Totally useless and I absolutely stand by that.
11 Likes
If you run some polls of your own, not even on politics but on things like “hard or soft G in GIF?” Or “worst Star Wars Movie (list here)”, or individual issues “should we forgive college debit?” you will see at first the percentages bounce around a lot, but pretty quickly they stabilize and with a shockingly small number of responses the numbers stay stable multiple orders of magnitude out.
There are some exceptions, like for example BB leans toward people that are socially similar so starting a poll on debit forgiveness on BB will swing hard towards “forgive”, but if someone who sees it in BB circulated it on a more fiscally “conservative” platform you will get an influx of a new opinion. (Which speaks towards the importance of getting a statistically significant random sample, not “people who tend to have similar opinions”)
So you can have a pretty accurate prediction of millions of voters with only thousands of sampled people (it probably needs tens of thousands though). Changes over time probably make a bigger difference then sample sizes though, things happening between now and a year from now can be very significant, a good economy favoring the incumbent bad economy favoring the challengers. A domestic or foreign emergency handled well or poorly.
8 Likes
Or done in good faith.
Rasmussen Reports uses research techniques that make its polls favor Republicans, Ipsos Public Affairs research director Mallory Newall said
[…]
Newall singled out Rasmussen’s practice of adjusting results by party identification in arguing that the pollster, which has been touted by President Trump and often has shown him with higher approval ratings than in other polls, favors the GOP.
12 Likes
Can you imagine the kind of person that picks up an unknown number, and is then happy to spend some time discussing who they’d vote for? Or replies to an unknown text about politics?
Thats the kind of person who is waaay overrepresented in these polls.
15 Likes
Even worse, I think, than that. I volunteered to make phone calls for Clinton in 2016. For years afterwards, I’ve gotten phone calls and texts from the DNC, from various Democratic candidates, and from pollsters. I got my name on lists because I volunteered and proved I’m politically active. So yeah, these aren’t exactly random samplings of people.
8 Likes
I call those types of people “Nyborgs”, from Glengarry Glen Ross.
Levene: I’ll cut you in on the $82,000 sale I just closed.
Williamson: Bruce and Harriet Nyborg? You wanna see the memos? They’re nuts. They used to call in every week when I was with Webb and we were selling Arizona. They’re nuts. Did you see how they were living? How can you delude yourself?
Levene: I got their check.
Williamson: Yeah? Well, forget it, frame it, it’s worthless.
Levene: The check is no good?
Williamson: Yeah. If you wanna wait around, I’ll pull the memo. I’m busy right now.
Levene: Wait a minute! The check is no good? They’re nuts?
Williamson: You wanna call the bank, Shelley? I called them. I call them four months ago when we first got the lead. The people are insane. They just like talking to salesmen.
10 Likes
Yeah, when you hear people complain about stats here and elsewhere its a mix of issues. Some of them, like how sample sizes don’t have to be large, are pretty well dealt with in undergrad stats courses. (Maybe not the very first course, if it is an applied one, but for sure by the second.) Others, like non-representative samples, can be dealt with decently well by more advanced techniques like poststratifaction. See e.g. 28 Poststratification | Stan User’s Guide (Whether a given poll actually does, or just reports the outcome that they wanted, is another matter.)
Other issues are pretty hard, like the question of what all this means over a year from the election.
6 Likes
