Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
Trafalgar got lucky in 2016 they were being paid by the Trump campaign to manufacture polls to make it look like Trump was going to win. It's likely they will fade away in history like many others who were paid to create a poll.
Status:
"everybody getting reported now.."
(set 21 days ago)
Location: Pine Grove,AL
29,549 posts, read 16,536,658 times
Reputation: 6032
Quote:
Originally Posted by tribecavsbrowns
If they were to include Ohio, which I agree is not a swing state, would there be someone else more accurate across those eight states plus Ohio? I'm fine if there is! I hold zero stock, literally or figuratively, in AtlasIntel. But those eight states seem like pretty important states if you're doing state polling in 2020. And it seems like if you were to add Ohio, based on what I saw from other polls of Ohio, they would still be most accurate.
Depends on if they recalculated the data all together for their averages or pulled it directly from 538( the site seems to say they pulled it without changing it, but I could be wrong)
This matters because 538's averages include (all States technically)Ohio, NC, Iowa, New Hampshire, Virginia ,Colorado, Texas and Nevada, but Atlas does not when comparing themselves to those pollsters.
Specifically in the average on that spread sheet/slideshow, they show Nevada as blank on their line, but have numbers for everyone else, but include Nevada in their average(basically as 0.0).
Atlas only had 7 polls , where as say PPP was polling states Like Alaska, and Data for Progress was polling Kansas, and Morning Consult was polling Missouri.
Some of these pollsters were being judged by polls from 20+ States and only lost the average because they were way off in Wisconsin, or because they only had Biden up by 7 in Colorado or something.
They manufactured polls in 2016 and 2020 they got lucky when Trump won even Trump was surprised he won and in 2020 he lost after Trafalgar made claims that Trump was way ahead in MI, MN only to find they were wildly incorrect in some of these midwest states.
Depends on if they recalculated the data all together for their averages or pulled it directly from 538( the site seems to say they pulled it without changing it, but I could be wrong)
This matters because 538's averages include (all States technically)Ohio, NC, Iowa, New Hampshire, Virginia ,Colorado, Texas and Nevada, but Atlas does not when comparing themselves to those pollsters.
Specifically in the average on that spread sheet/slideshow, they show Nevada as blank on their line, but have numbers for everyone else, but include Nevada in their average(basically as 0.0).
Atlas only had 7 polls , where as say PPP was polling states Like Alaska, and Data for Progress was polling Kansas, and Morning Consult was polling Missouri.
Some of these pollsters were being judged by polls from 20+ States and only lost the average because they were way off in Wisconsin, or because they only had Biden up by 7 in Colorado or something.
But again, does this really matter if, based on polls anyone can pull up from 538's website, AtlasIntel is the most accurate across WI/MI/PA/OH/NC/GA/FL/TX/AZ, and Trafalgar is the second most accurate? It just seems that if people feel so strongly that AtlasIntel and Trafalgar weren't the two most accurate, it shouldn't be that hard for them to find one pollster that disproves this.
Otherwise, this is just a critique of AtlasIntel's PR messaging masquerading as a critique of their actual premise.
And regarding some of these other pollsters, maybe Morning Consult wouldn't have missed by 12 points in Wisconsin if they spent less time and resources polling in Missouri (a state that was never in play). Just food for thought.
They manufactured polls in 2016 and 2020 they got lucky when Trump won even Trump was surprised he won and in 2020 he lost after Trafalgar made claims that Trump was way ahead in MI, MN only to find they were wildly incorrect in some of these midwest states.
Trafalgar's final MN poll was Biden +3. Did they make claims somewhere else implying they thought Trump was way ahead in MN? Or was this an error by you?
Status:
"everybody getting reported now.."
(set 21 days ago)
Location: Pine Grove,AL
29,549 posts, read 16,536,658 times
Reputation: 6032
Quote:
Originally Posted by tribecavsbrowns
But again, does this really matter if, based on polls anyone can pull up from 538's website, AtlasIntel is the most accurate across WI/MI/PA/OH/NC/GA/FL/TX/AZ, and Trafalgar is the second most accurate? It just seems that if people feel so strongly that AtlasIntel and Trafalgar weren't the two most accurate, it shouldn't be that hard for them to find one pollster that disproves this.
When it comes to Atlas actually picking the winner, their percentage is 42%, there are literally 100 pollster who got a better win percentage.
Quote:
Otherwise, this is just a critique of AtlasIntel's PR messaging masquerading as a critique of their actual premise.
refer to previous comment.
Quote:
And regarding some of these other pollsters, maybe Morning Consult wouldn't have missed by 12 points in Wisconsin if they spent less time and resources polling in Missouri (a state that was never in play). Just food for thought.
When it comes to Atlas actually picking the winner, their percentage is 42%, there are literally 100 pollster who got a better win percentage.
No, to say there are 100 pollsters who got a better win percentage on battleground states in 2020 than AtlasIntel would be a false statement. And that's not what you literally said, but it's obvious you were implying it.
I'm not sure if there were other pollsters with a better win percentage (a metric which you introduced) than AtlasIntel. I thought we were all generally in agreement that polled margin versus actual margin was how we were measuring accuracy. If there was a pollster with a better win percentage than AtlasIntel, now might be the time to share who it was. I can tell you with near certainty how many pollsters had more accurate margins than AtlasIntel, and it wasn't 100. It starts with a "Z" and rhymes with "hero."
Status:
"everybody getting reported now.."
(set 21 days ago)
Location: Pine Grove,AL
29,549 posts, read 16,536,658 times
Reputation: 6032
Quote:
Originally Posted by tribecavsbrowns
No, to say there are 100 pollsters who got a better win percentage on battleground states in 2020 than AtlasIntel would be a false statement. And that's not what you literally said, but it's obvious you were implying it.
There are 100+ pollsters who have a perfect 100% accuracy rate on 538(number of polls administered is between 1 and 28) , there are over 300 pollsters who have better than 43% accuracy.
There are at least 32 with a error rate lower than 1.34(atlas' rate), but again Atlas didnt poll Nevada. But counts Nevada in everyone else's average
Quote:
I'm not sure if there were other pollsters with a better win percentage (a metric which you introduced) than AtlasIntel. I thought we were all generally in agreement that polled margin versus actual margin was how we were measuring accuracy.
Again, the fact that Atlas doesnt poll some state makes that a hard argument to make. They purposefully choose not to release data from some states so they can make a PR argument of being more accurate. No different than Trafalgar who decided to stop polling close states mid August.
Quote:
If there was a pollster with a better win percentage than AtlasIntel, now might be the time to share who it was.
I can tell you with near certainty how many pollsters had more accurate margins than AtlasIntel, and it wasn't 100. It starts with a "Z" and rhymes with "hero."
Nope, its like 30. Same link , you can change which category why want to focus on so you can see error in polling as well.
Why can't you name one of the 100 or one of the 30? Didn't any of them poll the 2020 battleground states and have a better win percentage in those battleground states OR smaller margins in those battleground states than AtlasIntel?
Status:
"everybody getting reported now.."
(set 21 days ago)
Location: Pine Grove,AL
29,549 posts, read 16,536,658 times
Reputation: 6032
Quote:
Originally Posted by tribecavsbrowns
Why can't you name one of the 100 or one of the 30? Didn't any of them poll the 2020 battleground states and have a better win percentage in those battleground states OR smaller margins in those battleground states than AtlasIntel?
I literally gave you a link to 538 that lists them and gives you percentages and data, the same website you are using for your claims.
Are you pretending you dont see the link ?
Did you honestly overlook it in your hast to respond ?
Quote:
Again, talking about MI/WI/PA/OH/TX/FL/GA/NC/AZ.
Again, Atlas doesnt use the states you are listing, specifically because using Ohio would bump them to 3rd by their own metric on their website.
They also only compare themselves to bigger firms who do 30+ polls, if you compare them to other smaller firms who do the same number of polls (5,6,7,8) they arent even in the top 20, hence the link to 538 that shows all polling firms, not just big ones.
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
