I'm mentally defaulting to a scenario of Trump winning. It seems as unlikely as it did in 2016. When it happened. I don't think I could handle 4 more years of a Trump world (this is sending enough...
I'm mentally defaulting to a scenario of Trump winning. It seems as unlikely as it did in 2016. When it happened.
I don't think I could handle 4 more years of a Trump world (this is sending enough international ripples to not just be a US thing) if I didn't prepare for it now. If Biden winds, good. I mean, it's the only sane scenario. But it probably won't happen, right? Trump is the only hope a large part of old America has on maintaining its worldview and they're happily take all his bullshit to achieve that. There's good evidence that people are too shy to admit to voting for Trump in telephone interviews but then do at the voting booth. In fact, that's about the only explanation for 2016.
What I don't get is, I know from friends who actually get to vote in the US election and seriously considered not doing so because, basically, "Bernie would have been better". Like, pick your battles. Even if the establishment won't be broken in this election, the difference in direction is just mind-boggling.
Don't remember where I read it but it's basically the only explanation for why the polls were so wrong in 2016 I ever heard that made sense. Could be bullshit?
Don't remember where I read it but it's basically the only explanation for why the polls were so wrong in 2016 I ever heard that made sense. Could be bullshit?
I'd highly recommend you read that 538 article or this one they link to inside of it. TLDR: Shy Trump voters aren't statistically significant if they even exist at all, and the polls weren't any...
I'd highly recommend you read that 538 article or this one they link to inside of it.
TLDR: Shy Trump voters aren't statistically significant if they even exist at all, and the polls weren't any more inaccurate than usual in 2016.
I personally think people think of 2016 as such a "huge upset" because of just how outlandish it is for someone like Trump to win. They weren't mentally prepared for the possibility that he could actually win, so they interpreted a moderate lead for Clinton as a foregone conclusion.
As a general policy you should never believe the salesperson. Nate Silver/538 is the salesperson and is doing everything they can to say "please ignore how wrong we were by letting us explain that...
...calculated that some of the vote models could be off by 15 to 20 percent.
Virtually all the major vote forecasters, including Nate Silver’s FiveThirtyEight site, The New York Times Upshot and the Princeton Election Consortium, put Mrs. Clinton’s chances of winning in the 70 to 99 percent range.
As a general policy you should never believe the salesperson. Nate Silver/538 is the salesperson and is doing everything they can to say "please ignore how wrong we were by letting us explain that all of the models are shit we were right all along".
If the margin of error is that wide, the polls are useless.
538 put clinton at 71.4%, and repeatedly emphasized that these are probabilities that can still easily flip to a Trump win. Other organizations have come to essentially the same conclusion about...
Other organizations have come to essentially the same conclusion about the polls in 16', mentioning the same college graduate bias that 538 has while also being critical of the shy Trump voter theory.
All polls have a generous margin of error because they can't be perfect, that's why we try to calculate probabilities instead of being able to point to a winner by polling alone.
I think your general policy of not trusting the salesperson is a good one to have, but 538 has at least proven to me more than a few times that I can double check their info and they aren't lying.
I think of 538 like this: it's not so much that they are infallible, and more that they go into some pretty serious detail about their sources, their methodology, and how they try to quantify and...
I think of 538 like this: it's not so much that they are infallible, and more that they go into some pretty serious detail about their sources, their methodology, and how they try to quantify and estimate sources of error, e.g. Often their code and data is open-source as well, which is a plus. I have yet to find any other forecaster that is so open about their work.
That's not how margins of error work. The model isn't forecasting a score, it's assigning a probability of an event occurring. In this case, it was saying Trump's odds of winning the election...
That's not how margins of error work. The model isn't forecasting a score, it's assigning a probability of an event occurring. In this case, it was saying Trump's odds of winning the election (28%) were slightly better than the odds of a person winning two coin-flips in a row (25%).
On any individual poll yes. The point of a metaanalysis that aggregates polls is to increase the size of the sample and to average out many errors introduced by methodological differences. That's...
On any individual poll yes. The point of a metaanalysis that aggregates polls is to increase the size of the sample and to average out many errors introduced by methodological differences.
That's also not applicable to the model's projection of who is winning or losing. Again, it's not a score it's a probability of an event happening.
That doesn't fit into the probability they're trying to give there though, that statistic is trying to get a definite number taking into account the margins of error from various polls. If you...
That doesn't fit into the probability they're trying to give there though, that statistic is trying to get a definite number taking into account the margins of error from various polls. If you click on electoral or popular votes below it or go further down to the electoral votes bar graph they all show the full spread of possibilities alongside the averages. They tend to save any explaining the probabilities beyond that to their text posts.
I see this as hominem shit about Nate Silver regularly, but thankfully I don't actually see it in the saner corners of the internet such as tildes. Sad that changes today. Probabilities are...
I see this as hominem shit about Nate Silver regularly, but thankfully I don't actually see it in the saner corners of the internet such as tildes. Sad that changes today.
Probabilities are probabilities, 538 was by far the most correct model of all the ones available before the election, and just because you say "there's a higher probability X wins than Y" doesn't make you wrong it Y wins. In retrospect, their 70ish percent prediction is very reasonable. It's the people who went over 90 you shouldn't listen to.
I see this as hominem shit about Nate Silver regularly, but thankfully I don't actually see it in the saner corners of the internet such as tildes. Sad that changes today.
Did I say Nate Silver is good? How is your article relevant? Notice the difference between our comments: I talk about the data, you talk about the guy. "Guy bad, so data bad."
Did I say Nate Silver is good?
How is your article relevant?
Notice the difference between our comments: I talk about the data, you talk about the guy. "Guy bad, so data bad."
argumentum ad hominem: refers to a rhetorical strategy where the speaker attacks the character, motive, or some other attribute of the person making an argument rather than attacking the substance of the argument itself
I've said it before and I'll say it again. Early in-person voting starts here in NC on Thursday October 15th and I've already got plans to be in line when they open at 8am on Saturday October...
I've said it before and I'll say it again.
Early in-person voting starts here in NC on Thursday October 15th and I've already got plans to be in line when they open at 8am on Saturday October 17th.
Whether a vote is counted or not has nothing to do with the federal government, and everything to do with the state. The only votes that count on the federal level are the electoral votes -...
The legitimacy of mail in ballots will be absolutely trashed in the conservative media, the case will be brought up to the Supreme Court, where, guess what, the newly minted 5-3 conservative super majority will favour throwing out the mail in ballots.
Whether a vote is counted or not has nothing to do with the federal government, and everything to do with the state. The only votes that count on the federal level are the electoral votes - whether a mail in ballot is valid or not has nothing to do with the federal government.
Ohhh, you up in Maine? Jealous. I never got to use RCV before I moved, but dear god am I glad LePage is out. The 2010 and 2014 Maine gubernatorial elections are pretty much the perfect example...
Ohhh, you up in Maine? Jealous. I never got to use RCV before I moved, but dear god am I glad LePage is out. The 2010 and 2014 Maine gubernatorial elections are pretty much the perfect example case for why we need Ranked Choice Voting.
I'll be right there with you on the same day but in neighboring Tennessee. Even if there hadn't been a court ruling that limited vote by mail here, I'd still be there in person the first day of...
I'll be right there with you on the same day but in neighboring Tennessee. Even if there hadn't been a court ruling that limited vote by mail here, I'd still be there in person the first day of early voting. Not gonna leave it up to chance given the situation with USPS.
This is... just not true. Polls (almost) all come with a confidence interval or margin of error reported. For instance: If the poll doesn't list that, it's probably not worth reading. Now what...
We do not know anything. Polls mean nothing
This is... just not true. Polls (almost) all come with a confidence interval or margin of error reported. For instance:
1,302 likely voters nationwide were surveyed from September 17 - 21 with a margin of error of +/- 2.7 percentage points.
If the poll doesn't list that, it's probably not worth reading.
Now what does this mean practically? It’s related to the width of the bell curve of possible outcomes (and inversely related to the number of respondents). This is simplifying a bit, as in the real world the curves are not as nice and smooth, but it serves to get the point across.
For those of us not as familiar with bell curves out there, the way you get a probability out of one is by taking the area underneath region corresponding to the scenario (e.g., Biden wins by over 100 electoral college votes). So what they’re saying here is that:
the median (most likely outcome) is Biden wins, but not in a blow out
less likely, but still possible: Biden wins in a landslide, OR, Trump wins legitimately (as possible given his... shit)
Both of the latter scenarios result from the polls being off in one direction or another, typically fairly close to the bounds of the published error margin (you know, exactly what happened in 2016)
It’s not that polls are some awful method of forecasting the chances of something happening. It’s mostly that people (and in particular the news media) have little to no idea what probabilities mean. Trump is currently running about 1 in 5 chance in winning so like... slightly lower than picking a card from a deck and getting a heart. But that means that Biden has **roughly the same chance of blowing his ass out of the water **. It's also slightly lower than the same people forecast his chances of winning in 2016 were, FWIW (which were ~1 in 3).
I understand that statistics exist. I get that it's a very, very, very complex field that most of us understand absolutely nothing about. I will be the first to admit that I know nothing about it....
I understand that statistics exist. I get that it's a very, very, very complex field that most of us understand absolutely nothing about. I will be the first to admit that I know nothing about it.
To my unscientific mind, they don't mean much. Especially when every article like this always ends with the quote I pulled out in my previous post.
I thought Trump would win in 2016. I vividly remember 2016 teaching me what a "smug liberal" was with all the overconfident online Hillary supporters thinking they had it in the bag while bashing...
I thought Trump would win in 2016. I vividly remember 2016 teaching me what a "smug liberal" was with all the overconfident online Hillary supporters thinking they had it in the bag while bashing the progressive side of the party. Top that off with the fervor over those dumb leaked emails and the Republicans leaking Comey's letter, I had the gut feeling of her impending loss on election day.
This time around, I'm very confident in a Joe Biden win. It may or may not be a landslide, but any outcome where Trump is declared the winner will be extremely suspect to me. The only doubt that I have in the integrity of the election is purely from Trump's side. Everything he "competes" in, he cheats. I think the evidence of the high turnout of the special elections and the 2018 mid-terms have me convinced of the outcome this time around. As weird as it sounds, but the pessimism of people I have talked to about the sureness of Trump's re-election gives me even more reason to believe there will be an overwhelming turnout against him. As long as this energy doesn't dip to learned helplessness levels where voters end up not turning out due to dread, I don't see the majority of Americans voting for him. That's not to say he can't win, because if he does I guess I will know what others felt the day after election day in 2016, I just think the odds are incredibly against him.
Of course, this is all based on feeling. I could be wrong. In fact, I have contingency plans if I am wrong. But I have hope.
I'm mentally defaulting to a scenario of Trump winning. It seems as unlikely as it did in 2016. When it happened.
I don't think I could handle 4 more years of a Trump world (this is sending enough international ripples to not just be a US thing) if I didn't prepare for it now. If Biden winds, good. I mean, it's the only sane scenario. But it probably won't happen, right? Trump is the only hope a large part of old America has on maintaining its worldview and they're happily take all his bullshit to achieve that. There's good evidence that people are too shy to admit to voting for Trump in telephone interviews but then do at the voting booth. In fact, that's about the only explanation for 2016.
What I don't get is, I know from friends who actually get to vote in the US election and seriously considered not doing so because, basically, "Bernie would have been better". Like, pick your battles. Even if the establishment won't be broken in this election, the difference in direction is just mind-boggling.
Source? FiveThirtyEight on the contrary has found very little evidence for shy Trump voters.
Don't remember where I read it but it's basically the only explanation for why the polls were so wrong in 2016 I ever heard that made sense. Could be bullshit?
I'd highly recommend you read that 538 article or this one they link to inside of it.
TLDR: Shy Trump voters aren't statistically significant if they even exist at all, and the polls weren't any more inaccurate than usual in 2016.
I personally think people think of 2016 as such a "huge upset" because of just how outlandish it is for someone like Trump to win. They weren't mentally prepared for the possibility that he could actually win, so they interpreted a moderate lead for Clinton as a foregone conclusion.
As a general policy you should never believe the salesperson. Nate Silver/538 is the salesperson and is doing everything they can to say "please ignore how wrong we were by letting us explain that
all of the models are shitwe were right all along".If the margin of error is that wide, the polls are useless.
538 put clinton at 71.4%, and repeatedly emphasized that these are probabilities that can still easily flip to a Trump win.
Other organizations have come to essentially the same conclusion about the polls in 16', mentioning the same college graduate bias that 538 has while also being critical of the shy Trump voter theory.
All polls have a generous margin of error because they can't be perfect, that's why we try to calculate probabilities instead of being able to point to a winner by polling alone.
I think your general policy of not trusting the salesperson is a good one to have, but 538 has at least proven to me more than a few times that I can double check their info and they aren't lying.
I think of 538 like this: it's not so much that they are infallible, and more that they go into some pretty serious detail about their sources, their methodology, and how they try to quantify and estimate sources of error, e.g. Often their code and data is open-source as well, which is a plus. I have yet to find any other forecaster that is so open about their work.
That's not how margins of error work. The model isn't forecasting a score, it's assigning a probability of an event occurring. In this case, it was saying Trump's odds of winning the election (28%) were slightly better than the odds of a person winning two coin-flips in a row (25%).
On any individual poll yes. The point of a metaanalysis that aggregates polls is to increase the size of the sample and to average out many errors introduced by methodological differences.
That's also not applicable to the model's projection of who is winning or losing. Again, it's not a score it's a probability of an event happening.
That doesn't fit into the probability they're trying to give there though, that statistic is trying to get a definite number taking into account the margins of error from various polls. If you click on electoral or popular votes below it or go further down to the electoral votes bar graph they all show the full spread of possibilities alongside the averages. They tend to save any explaining the probabilities beyond that to their text posts.
I see this as hominem shit about Nate Silver regularly, but thankfully I don't actually see it in the saner corners of the internet such as tildes. Sad that changes today.
Probabilities are probabilities, 538 was by far the most correct model of all the ones available before the election, and just because you say "there's a higher probability X wins than Y" doesn't make you wrong it Y wins. In retrospect, their 70ish percent prediction is very reasonable. It's the people who went over 90 you shouldn't listen to.
He isn't as good as you think he is.
Notice the difference between our comments: I talk about the data, you talk about the guy. "Guy bad, so data bad."
I'm not sure I could even fathom the concept of a "shy Trump voter". Every Trump voter I've ever met was quite proud to share that fact.
But the polls weren't wrong in 2016. People just don't understand how probabilities work.
I've said it before and I'll say it again.
Early in-person voting starts here in NC on Thursday October 15th and I've already got plans to be in line when they open at 8am on Saturday October 17th.
God I can only dream of an absolute blow out.
Whether a vote is counted or not has nothing to do with the federal government, and everything to do with the state. The only votes that count on the federal level are the electoral votes - whether a mail in ballot is valid or not has nothing to do with the federal government.
Bush v gore, anyone?
Ohhh, you up in Maine? Jealous. I never got to use RCV before I moved, but dear god am I glad LePage is out. The 2010 and 2014 Maine gubernatorial elections are pretty much the perfect example case for why we need Ranked Choice Voting.
I'll be right there with you on the same day but in neighboring Tennessee. Even if there hadn't been a court ruling that limited vote by mail here, I'd still be there in person the first day of early voting. Not gonna leave it up to chance given the situation with USPS.
This hits the nail on the head. We do not know anything. Polls mean nothing. Just look at how the 2016 election went.
This is... just not true. Polls (almost) all come with a confidence interval or margin of error reported. For instance:
If the poll doesn't list that, it's probably not worth reading.
Now what does this mean practically? It’s related to the width of the bell curve of possible outcomes (and inversely related to the number of respondents). This is simplifying a bit, as in the real world the curves are not as nice and smooth, but it serves to get the point across.
For those of us not as familiar with bell curves out there, the way you get a probability out of one is by taking the area underneath region corresponding to the scenario (e.g., Biden wins by over 100 electoral college votes). So what they’re saying here is that:
Both of the latter scenarios result from the polls being off in one direction or another, typically fairly close to the bounds of the published error margin (you know, exactly what happened in 2016)
It’s not that polls are some awful method of forecasting the chances of something happening. It’s mostly that people (and in particular the news media) have little to no idea what probabilities mean. Trump is currently running about 1 in 5 chance in winning so like... slightly lower than picking a card from a deck and getting a heart. But that means that Biden has **roughly the same chance of blowing his ass out of the water **. It's also slightly lower than the same people forecast his chances of winning in 2016 were, FWIW (which were ~1 in 3).
I understand that statistics exist. I get that it's a very, very, very complex field that most of us understand absolutely nothing about. I will be the first to admit that I know nothing about it.
To my unscientific mind, they don't mean much. Especially when every article like this always ends with the quote I pulled out in my previous post.
This came off as extremely snarky to me. It's not a big deal, that was just my reaction to it.
Fair enough, I'll delete that bit. Apologies.
I thought Trump would win in 2016. I vividly remember 2016 teaching me what a "smug liberal" was with all the overconfident online Hillary supporters thinking they had it in the bag while bashing the progressive side of the party. Top that off with the fervor over those dumb leaked emails and the Republicans leaking Comey's letter, I had the gut feeling of her impending loss on election day.
This time around, I'm very confident in a Joe Biden win. It may or may not be a landslide, but any outcome where Trump is declared the winner will be extremely suspect to me. The only doubt that I have in the integrity of the election is purely from Trump's side. Everything he "competes" in, he cheats. I think the evidence of the high turnout of the special elections and the 2018 mid-terms have me convinced of the outcome this time around. As weird as it sounds, but the pessimism of people I have talked to about the sureness of Trump's re-election gives me even more reason to believe there will be an overwhelming turnout against him. As long as this energy doesn't dip to learned helplessness levels where voters end up not turning out due to dread, I don't see the majority of Americans voting for him. That's not to say he can't win, because if he does I guess I will know what others felt the day after election day in 2016, I just think the odds are incredibly against him.
Of course, this is all based on feeling. I could be wrong. In fact, I have contingency plans if I am wrong. But I have hope.