I’d like to remind everyone that 8 years ago, the polls showed Hillary was going to trounce Trump pretty handedly. There was tons of discussion after the election about how the polls could be so wrong.
I think Fivethityeight’s explanation went something like…
If a candidate is only polling 40% to their opponents’s 60%, and you were to run the election 10 times with a different sampling of voters each time, it doesn’t mean that the candidate will lose by 60% every time. It means they’re going to win four times out of ten.
Don’t let polls lull you into either complacency or despair. The only thing polls are really good for is giving pundits something to talk about in the 24 hour news cycle. Polls don’t decide the election. Only actual votes on actual ballots that are actually submitted in time decide the election.
And this messaging is a large part of what led to low democrat turnout when Hillary was running for office. Her early campaign had basically been “lol don’t worry about this, he’s an idiot who doesn’t gave a chance of winning.” It wasn’t until about a month before the actual election that someone in her campaign team realized this would lull voters into a false sense of security. Suddenly, their entire tone changed from “he has no chance of winning” to “oh for fucks sake please go vote”. But it was too little, too late. Democrat voters stayed home, and handed the win to Trump.
There was an acknowledged gap in polling in 2016 that excluded likely trump voters. It has since been resolved and polling in 2020 and 2022 was highly accurate.
With that said, I am not advocating for taking polls as gospel, 2016 showed us there can be flaws and mistakes. At the end of the day, I don’t give a fuck what the polls say, we all need to show up and vote. If Kamala had a 50 point lead in my state, I am still showing up and voting for her.
“There was an acknowledged gap in polling in 2016 that excluded likely trump voters. It has since been resolved and polling in 2020 and 2022 was highly accurate.”
I would respectfully dispute that statement. In the last dozen or so special elections as well as the last mid term election the Republicans have under performed with respect to their polling.
There is a large swath of the population that doesn’t participate in polls because they don’t answer the phone for strange numbers and don’t answer questions online or in person.
This “silent” population segment has favored Dems over the last few years but they could just as easily go for Republicans (as we saw with Trump in 2016) as we simply do not have good polling. I think they will swing Democrat again this election but we should take nothing for granted.
Sounds like you’re thinking of 538’s election needle, not polling data. If a candidate has 60% of the votes in a poll, assuming the poll is accurate, they win 100% of the time. The standard deviation on a population this big is practically 0
Very very true, I put time and money into Bernie Sanders second bid, the polling made it look like he was going to win the primary in a devistating landslide. It never materialized, his base, if they ever were serious weren’t serious enough to actually make it to a polling place on the day of. Very disappointing. Never think the polling will match the voting, they can be very different animals.
I sure wish people would actually vote in the primaries and the general. The charts of which states “no vote” would win if it were a candidate are all insane.
That’s a lot of things put in place to make voting hard to impossible, some states just reduce polling places to 10-20 from 100s, surprise kick people off the voting rolls, and the classic just start legislating and criminally prosecuting any organization that attempts to get people to vote.
538 put Trump’s 2016 chances at about 75%. That means Trump needed to flip two coins and have them both come up heads. It wasn’t a ridiculous outside chance at all.
People have also let Comey and his last minute letter off the hook. Polls were really close, but favored Hillary. That letter came too late for any poll to absorb the new information, but it very likely tipped the scales. There were a lot of things that went wrong in that election–it never should have been so close in the first place–but that very likely shifted the outcome.
the polls showed Hillary was going to trounce Trump pretty handedly.
Not true. She was within the margin of error in the swing states.
I think Fivethityeight’s explanation went something like…
Don’t confuse 538’s model with polls. 538 takes polling data as an input, and then runs simulations that output the odds which side will win.
Polls don’t measure the odds a candidate will win, they measure how many people would vote a certain way if the election were held today. Predictive models take that data and do a lot more than simply average the results.
538s model was a good estimator that year too, they leaned towards Hillary (and to be fair, she did win the popular vote) but certainly kept a trump win in the swing states within margin of error.
I’d like to remind everyone that 8 years ago, the polls showed Hillary was going to trounce Trump pretty handedly. There was tons of discussion after the election about how the polls could be so wrong.
I think Fivethityeight’s explanation went something like…
If a candidate is only polling 40% to their opponents’s 60%, and you were to run the election 10 times with a different sampling of voters each time, it doesn’t mean that the candidate will lose by 60% every time. It means they’re going to win four times out of ten.
Don’t let polls lull you into either complacency or despair. The only thing polls are really good for is giving pundits something to talk about in the 24 hour news cycle. Polls don’t decide the election. Only actual votes on actual ballots that are actually submitted in time decide the election.
And this messaging is a large part of what led to low democrat turnout when Hillary was running for office. Her early campaign had basically been “lol don’t worry about this, he’s an idiot who doesn’t gave a chance of winning.” It wasn’t until about a month before the actual election that someone in her campaign team realized this would lull voters into a false sense of security. Suddenly, their entire tone changed from “he has no chance of winning” to “oh for fucks sake please go vote”. But it was too little, too late. Democrat voters stayed home, and handed the win to Trump.
Well this time Harris has been telling everyone that Trump is trying to end democracy from the beginning
deleted by creator
Thats absolutely not how polling statistics would work.
There was an acknowledged gap in polling in 2016 that excluded likely trump voters. It has since been resolved and polling in 2020 and 2022 was highly accurate.
With that said, I am not advocating for taking polls as gospel, 2016 showed us there can be flaws and mistakes. At the end of the day, I don’t give a fuck what the polls say, we all need to show up and vote. If Kamala had a 50 point lead in my state, I am still showing up and voting for her.
“There was an acknowledged gap in polling in 2016 that excluded likely trump voters. It has since been resolved and polling in 2020 and 2022 was highly accurate.”
I would respectfully dispute that statement. In the last dozen or so special elections as well as the last mid term election the Republicans have under performed with respect to their polling.
There is a large swath of the population that doesn’t participate in polls because they don’t answer the phone for strange numbers and don’t answer questions online or in person.
This “silent” population segment has favored Dems over the last few years but they could just as easily go for Republicans (as we saw with Trump in 2016) as we simply do not have good polling. I think they will swing Democrat again this election but we should take nothing for granted.
Sounds like you’re thinking of 538’s election needle, not polling data. If a candidate has 60% of the votes in a poll, assuming the poll is accurate, they win 100% of the time. The standard deviation on a population this big is practically 0
That could be. It was 8 years ago and I was very upset. Still haven’t recovered.
Very very true, I put time and money into Bernie Sanders second bid, the polling made it look like he was going to win the primary in a devistating landslide. It never materialized, his base, if they ever were serious weren’t serious enough to actually make it to a polling place on the day of. Very disappointing. Never think the polling will match the voting, they can be very different animals.
I sure wish people would actually vote in the primaries and the general. The charts of which states “no vote” would win if it were a candidate are all insane.
That’s a lot of things put in place to make voting hard to impossible, some states just reduce polling places to 10-20 from 100s, surprise kick people off the voting rolls, and the classic just start legislating and criminally prosecuting any organization that attempts to get people to vote.
538 put Trump’s 2016 chances at about 75%. That means Trump needed to flip two coins and have them both come up heads. It wasn’t a ridiculous outside chance at all.
People have also let Comey and his last minute letter off the hook. Polls were really close, but favored Hillary. That letter came too late for any poll to absorb the new information, but it very likely tipped the scales. There were a lot of things that went wrong in that election–it never should have been so close in the first place–but that very likely shifted the outcome.
Not true. She was within the margin of error in the swing states.
Don’t confuse 538’s model with polls. 538 takes polling data as an input, and then runs simulations that output the odds which side will win.
Polls don’t measure the odds a candidate will win, they measure how many people would vote a certain way if the election were held today. Predictive models take that data and do a lot more than simply average the results.
538s model was a good estimator that year too, they leaned towards Hillary (and to be fair, she did win the popular vote) but certainly kept a trump win in the swing states within margin of error.
270 to win is another good site