“Polls have “called” elections correctly 78 percent of the time” according to that article. Just because they are more accurate than in another time frame does not mean they are accurate overall. This is an incredibly poor rate in the larger picture. Independent groups are notoriously hard to poll and they are the ones that decide elections. If it’s a landslide then of course the poll will be correct. Completely unreliable in close elections. However they make excellent time filters for news networks.
“Polls have “called” elections correctly 78 percent of the time” according to that article. Just
Maybe you should just read their argument as to why this is a garbage metric. Especially if you are arguing they don’t even “relate to reality.”
If always predicting who will win is the requirement for polls, the problem isn’t the polling itself, but your understanding of what a poll means and how statistics work.
Polls only predict well in places where you don’t need polls… hence their 78% success rate. What is their rate in closer elections? Likely right at 50%…useless.
The article talks about this and why it’s a bad metric. If you’re going to ignore their descriptive argument, you’ll just ignore my less than descriptive argument here.
But rest assured that at least part of the problem here is that you don’t understand statistics and probability.
This is a thread where someone made the statement “Trump would win if the election was today.” based on polls. You said yourself, that’s not what polls are for. Take it up with the person who is misusing the poll to make definitive statements like that rather than the person saying you can’t trust the polls for that.
The reason people go to “No relationship with reality” is because many people use the polls to say “will” instead of “favored” or conflate “will” and “favored.” When that’s the standard you are often presented, of course you are going to come to conclusion polling doesn’t have all that much to do with reality. Because it isn’t that predictive. Especially when you’re looking at things where we take this somewhat fuzzy number and turn it into a binary yes or no while the cloud of possibilities comfortably encompasses both outcomes.
So when talking to some making definitive statements about the outcome of an election based on polls, how they are using polls only has a tenuous relationship to reality.
So, like I said, they don’t understand polls and probability? I’m not sure why I have to be pedantic with the other poster, when this poster is just ridiculously wrong.
They’re the exact same mistake. Since the commenter you were responding to wasn’t the one to originally make the mistake, but instead was arguing with someone who’s premise relied on that mistake, it’s weird to only get on them about it.
The Polls Were Historically Accurate In 2022
“Polls have “called” elections correctly 78 percent of the time” according to that article. Just because they are more accurate than in another time frame does not mean they are accurate overall. This is an incredibly poor rate in the larger picture. Independent groups are notoriously hard to poll and they are the ones that decide elections. If it’s a landslide then of course the poll will be correct. Completely unreliable in close elections. However they make excellent time filters for news networks.
Maybe you should just read their argument as to why this is a garbage metric. Especially if you are arguing they don’t even “relate to reality.”
If always predicting who will win is the requirement for polls, the problem isn’t the polling itself, but your understanding of what a poll means and how statistics work.
Polls only predict well in places where you don’t need polls… hence their 78% success rate. What is their rate in closer elections? Likely right at 50%…useless.
The article talks about this and why it’s a bad metric. If you’re going to ignore their descriptive argument, you’ll just ignore my less than descriptive argument here.
But rest assured that at least part of the problem here is that you don’t understand statistics and probability.
This is a thread where someone made the statement “Trump would win if the election was today.” based on polls. You said yourself, that’s not what polls are for. Take it up with the person who is misusing the poll to make definitive statements like that rather than the person saying you can’t trust the polls for that.
Both that poster and this one can be wrong.
The difference is that the other poster is just conflating will with favored and it’s kind of pedantic to argue with that.
This poster is claiming that they are no relationship with reality, which is just blatantly wrong.
The reason people go to “No relationship with reality” is because many people use the polls to say “will” instead of “favored” or conflate “will” and “favored.” When that’s the standard you are often presented, of course you are going to come to conclusion polling doesn’t have all that much to do with reality. Because it isn’t that predictive. Especially when you’re looking at things where we take this somewhat fuzzy number and turn it into a binary yes or no while the cloud of possibilities comfortably encompasses both outcomes.
So when talking to some making definitive statements about the outcome of an election based on polls, how they are using polls only has a tenuous relationship to reality.
So, like I said, they don’t understand polls and probability? I’m not sure why I have to be pedantic with the other poster, when this poster is just ridiculously wrong.
They’re the exact same mistake. Since the commenter you were responding to wasn’t the one to originally make the mistake, but instead was arguing with someone who’s premise relied on that mistake, it’s weird to only get on them about it.