While the sentiment is solid that polls are not a very good predictor, what’s even more unreliable is leaning into anecdotes of seeing “excitement” in a social media post, which is what this article is doing. So your comment comes off as ‘discard the polls, someone on social media says they see lots of Harris for president signs in Florida’.
So it seems reasonable to say the polls indicate a less rosy picture than some social media post expressing feel good about seeing signs of Harris enthusiasm, but ultimately either way don’t feel defeated nor complacent and get out and vote your preference.
Do you remember 2016? Polls were saying Clinton would beat Trump by a significant margin.
If you’re approaching this logically, you’d notice the trend on data being unreliable when Trump is on the ballot.
It’s mostly attributed to inaccuracies in putting appropriate weight on likely voters vs. unlikely voters. People considered unlikely to vote by pollsters went out and voted, and they voted for Trump.
Measuring racism is also something that polling is bad at. People simply don’t like to admit to being racist. Is this related to the reason why polling on Trump is inaccurate? We don’t know because there’s no data on this. Some things polling just fails at. Can’t do much when people won’t provide you with data that may be relevant.
We do know that Trump’s primary numbers were lower than polling indicated it would be. Does that mean his numbers in the general will be lower than the polls we’re seeing right now? We don’t know.
What effect did January 6 have on people’s decisions? Some people may not want to talk about it. But the week before election they’ll probably be seeing political ads showing video about Jan. 6 and ask people straight up “do you want this to happen again?” which might people who might say Jan. 6 wasn’t a big deal to privately think otherwise just stay home on Election Day. Polling is based on past trends, so isn’t going to be good a predicting anything after unprecedented events.
After this election pollsters have a baseline for how likely people will vote for a candidate lost the previous election, tried to overthrow the government, was convicted of felonies, had an assassination attempt vs. a candidate that suddenly became prominent after the sitting President and presumptive nominee dropped out the race 3.5 months before the election. But right now there’s not a lot of data there on this particular scenario.
The data is simply too unreliable to make any prediction on anything. So… vote!
A fair assessment as to why polling may be unreliable. However keep in mind this thread started as a rebuttal of blind anecdotal enthusiasm in a social media post. The story is ‘someone posted on social media that Florida looks like Harris country’, and they posted that polls suggest that post is too optimistic. Polls may be imperfect, but the methodology is far closer to informative than “I saw some Harris campaign signs around”.
Even if that were the case, we would need some data to indicate which poll is accurate so then we’ll know some number that won’t actually determine who will end up as president.
Might be better to just ignore the simplistic number that doesn’t indicate anything useful and instead focus on numbers indicating what issues people care about and try to convince as many people as possible to vote for Harris.
Good write up, but you’re doing the thing you said not to do (approaching this logically).
My half baked opinion on this is that people are lying to pollsters. I think it’s people of all political walks and for varied reasons, but it’s the only thing that keeps making sense.
Even exit polls are getting it wrong. Like, that can only happen if people are lying.
Recognize that the data may be flawed. Polling is incredibly accurate, but only if you survey a simple random sample. And that is very difficult to do. It introduces a lot of difficulty in getting right answers. Some polling methodologies will try to manipulate the raw data and weight it to try and make it representative, but that introduces a whole host of problems.
2016 and 2020 under predicted Trump’s popularity for instance, while 2022 under predicted Democrats’ popularity. We don’t know what the situation now.
Polls are still useful, but you have to treat them with a grain of salt. What tends to be more accurate is changes within the same polling group over time.
Problem is that polling would have to have all the exact same behaviors as an actual election
The ballot boxes don’t come to the people, the people opt to go to the ballot boxes. So cold calling/mailing people means you’ve changed the engagement to include people that wouldn’t actually go out to vote. Some try to measure likelihood to vote, but if the reason is ‘laziness’, a lot of people are unlikely to admit they won’t vote.
Some population sees the polls as a strategic tool, and may modify their participation to advance what they think their outcome needs. Declare support for the opposing candidate to put the fear of losing into like-minded voters, for example.
People know the polls don’t actually decide anything, so even if they will vote, they may dismiss polls as a waste of their time. Or even being distrustful of the agenda behind the poll and decline to participate thinking that works best to undermine potentially malicious polling
People have more confidence in the ballot being secret than polling. If someone thinks their answer will be seen/overheard by a spouse, that may change their tune. If someone thinks something vile would actually be in their benefit, they may be reluctant to admit that, but happy to act on it at the ballot box.
Now polls are better than “gut feelings” or “this person posted to social media their gut feelings”, but the ultimate answer is we have no way of accurate prediction, so don’t be encouraged or discouraged too much and just go vote.
You need to look at the actual statistical science. If you find 45% support for something, but there’s a 3% margin of error with a 95% conference interval, then there’s a 95% chance that the true value is anywhere from 42-48%. And that’s with a perfect, simple random sample.
It has its uses, but you have to be aware of its limitations and caveats.
Are you just arguing for the sake of arguing? I’m saying that even a perfect sample will not necessarily lead to an accurate conclusion, and having a perfect sample is incredibly difficult on top of that.
Now factor in a major event occurring, and people’s opinions and thoughts being in flux. To properly gauge mood, you need to give people time to process – hence why immediate polling is not helpful.
You do realize that the person you originally responded to was saying that polls probably aren’t helpful right now, not that polls are universally useless?
Polls become more and more unreliable in the modern age. We have the least accurate polling in 40 years according to pew research. Pollsters report a 3% margin of error when it’s more like 6-7%. There is every reason to be skeptical of polling and not take them too seriously.
Even if Polling is less accurate than it was, and I haven’t seen any such claims by an authority on this matter least of all Pew Research, it is still a lot more accurate than your thoughts and feelings, mate.
Take a look at THIS LINK. It’s FiveThirtyEight’s composite polling for the state with individual polls listed down below, one by Redfield & Wilton Strategies sponsored by The Telegraph with Trump +8 and another by InsiderAdvantage sponsored by WTVT (Tampa, Fla.) with Trump +10.
He said don’t rely on polls “too much”, not “not at all”. Those with reading comprehension would recognize what he meant was that there is real possibility that there is a smaller gap to bridge than you might think.
You’re on some weak ego tangent that has nothing to do with anything, quoting an expired poll aggregate of Biden v Trump.
Now here is what I am saying: Rely on the polls. Use data to back your beliefs. Reject emotional responses which fuel your personal biases, be objective and make the best choices based on verifiably true information. If you have a better source than a poll that is great, if not then the poll is better than you.
I think his point is not that polling is supremely reliable, just that it is more reliable than the article here, which is entirely based on one person’s gut feeling about what they randomly see (and want to see). Florida may not be a lost cause , but it’s also not something to get your hopes up too much over. The polling is at least a decent relative indicator that FL is a much more uphill battle than other states with a closer polling margin.
I understand the point but people are trying to change “more accurate than a gut feeling” to “the best predictive tools we have”, which betrays how accurate they are.
I’m not sure anyone here would defend the methodology of these polls but they keep referencing them constantly.
I understand we have nothing else, but maybe we just can’t predict the future as well as we think we can.
You can see the largest sample size polls say about Clinton +2, which is close to the national popular vote, but there are several in the list that predicted a trump advantage. Trump won due to single digit wins in swing states.
don’t rely on polls too much especially right now
While the sentiment is solid that polls are not a very good predictor, what’s even more unreliable is leaning into anecdotes of seeing “excitement” in a social media post, which is what this article is doing. So your comment comes off as ‘discard the polls, someone on social media says they see lots of Harris for president signs in Florida’.
So it seems reasonable to say the polls indicate a less rosy picture than some social media post expressing feel good about seeing signs of Harris enthusiasm, but ultimately either way don’t feel defeated nor complacent and get out and vote your preference.
Don’t rely on data or logic? The fuq?
fuck off, mate
It’s not very difficult to use logic to see why the data isn’t as useful as you seem to think it is.
Then what exactly are we deriving these claims of surge of support for Harris on, if not quantifiable recorded support for Harris?
deleted by creator
Do you remember 2016? Polls were saying Clinton would beat Trump by a significant margin.
If you’re approaching this logically, you’d notice the trend on data being unreliable when Trump is on the ballot.
It’s mostly attributed to inaccuracies in putting appropriate weight on likely voters vs. unlikely voters. People considered unlikely to vote by pollsters went out and voted, and they voted for Trump.
Measuring racism is also something that polling is bad at. People simply don’t like to admit to being racist. Is this related to the reason why polling on Trump is inaccurate? We don’t know because there’s no data on this. Some things polling just fails at. Can’t do much when people won’t provide you with data that may be relevant.
We do know that Trump’s primary numbers were lower than polling indicated it would be. Does that mean his numbers in the general will be lower than the polls we’re seeing right now? We don’t know.
What effect did January 6 have on people’s decisions? Some people may not want to talk about it. But the week before election they’ll probably be seeing political ads showing video about Jan. 6 and ask people straight up “do you want this to happen again?” which might people who might say Jan. 6 wasn’t a big deal to privately think otherwise just stay home on Election Day. Polling is based on past trends, so isn’t going to be good a predicting anything after unprecedented events.
After this election pollsters have a baseline for how likely people will vote for a candidate lost the previous election, tried to overthrow the government, was convicted of felonies, had an assassination attempt vs. a candidate that suddenly became prominent after the sitting President and presumptive nominee dropped out the race 3.5 months before the election. But right now there’s not a lot of data there on this particular scenario.
The data is simply too unreliable to make any prediction on anything. So… vote!
A fair assessment as to why polling may be unreliable. However keep in mind this thread started as a rebuttal of blind anecdotal enthusiasm in a social media post. The story is ‘someone posted on social media that Florida looks like Harris country’, and they posted that polls suggest that post is too optimistic. Polls may be imperfect, but the methodology is far closer to informative than “I saw some Harris campaign signs around”.
Not all polls predicted that
The polls pretty accurately predicted the popular vote, but Trump won in 6 highly contested swing states which at the time included Florida
Even if that were the case, we would need some data to indicate which poll is accurate so then we’ll know some number that won’t actually determine who will end up as president.
Might be better to just ignore the simplistic number that doesn’t indicate anything useful and instead focus on numbers indicating what issues people care about and try to convince as many people as possible to vote for Harris.
deleted by creator
That is completely off topic. The conversation here is about whether no polling is more trustworthy than some polling.
Good write up, but you’re doing the thing you said not to do (approaching this logically).
My half baked opinion on this is that people are lying to pollsters. I think it’s people of all political walks and for varied reasons, but it’s the only thing that keeps making sense.
Even exit polls are getting it wrong. Like, that can only happen if people are lying.
Exit polls getting it wrong didn’t mean people are lying. People may be refusing to answer in a way that skews one way or another.
Recognize that the data may be flawed. Polling is incredibly accurate, but only if you survey a simple random sample. And that is very difficult to do. It introduces a lot of difficulty in getting right answers. Some polling methodologies will try to manipulate the raw data and weight it to try and make it representative, but that introduces a whole host of problems.
2016 and 2020 under predicted Trump’s popularity for instance, while 2022 under predicted Democrats’ popularity. We don’t know what the situation now.
Polls are still useful, but you have to treat them with a grain of salt. What tends to be more accurate is changes within the same polling group over time.
Problem is that polling would have to have all the exact same behaviors as an actual election
Now polls are better than “gut feelings” or “this person posted to social media their gut feelings”, but the ultimate answer is we have no way of accurate prediction, so don’t be encouraged or discouraged too much and just go vote.
Polling is not an inferior source to your gut feelings.
You need to look at the actual statistical science. If you find 45% support for something, but there’s a 3% margin of error with a 95% conference interval, then there’s a 95% chance that the true value is anywhere from 42-48%. And that’s with a perfect, simple random sample.
It has its uses, but you have to be aware of its limitations and caveats.
But whats the interval on shit you just make up? Probably not as good a source as the polling.
Are you just arguing for the sake of arguing? I’m saying that even a perfect sample will not necessarily lead to an accurate conclusion, and having a perfect sample is incredibly difficult on top of that.
Now factor in a major event occurring, and people’s opinions and thoughts being in flux. To properly gauge mood, you need to give people time to process – hence why immediate polling is not helpful.
You do realize that the person you originally responded to was saying that polls probably aren’t helpful right now, not that polls are universally useless?
Check the context of this thread. Then my words will make more sense and your point reveals itself to be coping to reaffirm unscientific bias.
Maybe not your gut feelings.
Polls become more and more unreliable in the modern age. We have the least accurate polling in 40 years according to pew research. Pollsters report a 3% margin of error when it’s more like 6-7%. There is every reason to be skeptical of polling and not take them too seriously.
It’s just personally annoying to me that Trump is even within 20 points.
Even if Polling is less accurate than it was, and I haven’t seen any such claims by an authority on this matter least of all Pew Research, it is still a lot more accurate than your thoughts and feelings, mate.
Take a look at THIS LINK. It’s FiveThirtyEight’s composite polling for the state with individual polls listed down below, one by Redfield & Wilton Strategies sponsored by The Telegraph with Trump +8 and another by InsiderAdvantage sponsored by WTVT (Tampa, Fla.) with Trump +10.
He said don’t rely on polls “too much”, not “not at all”. Those with reading comprehension would recognize what he meant was that there is real possibility that there is a smaller gap to bridge than you might think.
You’re on some weak ego tangent that has nothing to do with anything, quoting an expired poll aggregate of Biden v Trump.
Now here is what I am saying: Rely on the polls. Use data to back your beliefs. Reject emotional responses which fuel your personal biases, be objective and make the best choices based on verifiably true information. If you have a better source than a poll that is great, if not then the poll is better than you.
If you think polling is that reliable be my guest. Noones trying to force you to be reasonable.
You have selective vision, I wonder if optometrists can do anything about that?
I think his point is not that polling is supremely reliable, just that it is more reliable than the article here, which is entirely based on one person’s gut feeling about what they randomly see (and want to see). Florida may not be a lost cause , but it’s also not something to get your hopes up too much over. The polling is at least a decent relative indicator that FL is a much more uphill battle than other states with a closer polling margin.
I understand the point but people are trying to change “more accurate than a gut feeling” to “the best predictive tools we have”, which betrays how accurate they are.
I’m not sure anyone here would defend the methodology of these polls but they keep referencing them constantly.
I understand we have nothing else, but maybe we just can’t predict the future as well as we think we can.
Every poll in 2016 showed Trump losing, until he didnt
Look at this list of polls for 2016: LINK
You can see the largest sample size polls say about Clinton +2, which is close to the national popular vote, but there are several in the list that predicted a trump advantage. Trump won due to single digit wins in swing states.