Trafalgar Is A Garbage Poll -- They Believe In the Bradley Effect (Missouri, compare)

[dsjj251]

Originally Posted by tribecavsbrowns

Have we talked before? Sorry if we have, and I've forgotten.

No, But people have been attacking 538 for years because they seem to not understand what they do, LOL

[Elliott_CA]

Originally Posted by hooligan

OK, I have no interest in who might have lucked into a "good year" or not.

This is the correct answer. Any poll that systematically fudges their numbers to account for "hidden voters" that's based on supposition and not fact is a garbage poll. Just because their wrong approach got lucky in some states means nothing.

[tribecavsbrowns]

Originally Posted by Elliott_CA

This is the correct answer. Any poll that systematically fudges their numbers to account for "hidden voters" that's based on supposition and not fact is a garbage poll. Just because their wrong approach got lucky in some states means nothing.

You say these polls got "lucky" -- do you think ABC/WaPo got "unlucky" overestimating Biden by 16 points in Wisconsin or Quinnipiac got "unlucky" overestimating Biden by 12 points in Ohio?

[dsjj251]

Originally Posted by tribecavsbrowns

You say these polls got "lucky" -- do you think ABC/WaPo got "unlucky" overestimating Biden by 16 points in Wisconsin or Quinnipiac got "unlucky" overestimating Biden by 12 points in Ohio?

Yes over estimating margins is bad. Id even argue that I trust a pollster who is wrong on the winner, but only misses the margin by 2%, rather than someone who is right and misses by 16.

The problem for you is that Atlas chose not to poll Ohio, North Carolina, Iowa or Nevada probably because they believed their data was too far off, same as Trafalgar .

[tribecavsbrowns]

Originally Posted by dsjj251

Yes over estimating margins is bad. Id even argue that I trust a pollster who is wrong on the winner, but only misses the margin by 2%, rather than someone who is right and misses by 16.

The problem for you is that Atlas chose not to poll Ohio, North Carolina, Iowa or Nevada probably because they believed their data was too far off, same as Trafalgar .

Atlas did poll Ohio and North Carolina: https://www.atlasintel.org/polls/general-release-polls

[dsjj251]

Originally Posted by tribecavsbrowns

Atlas did poll Ohio and North Carolina: https://www.atlasintel.org/polls/general-release-polls

They did not include Ohio In there analysis of them being the best pollster in 2020, probably because they were off by 4 points.

They only list, Arizona, FLorida, Georgia, Michigan, North Carolina, Pennsylvania, Texas, and Wisconsin.

Their Ohio margin was the one they were most off, maybe they purposefully didnt include it, or dont see it as a swing state.

The problem then becomes that all of their analysis of other groups include Ohio except the composite for hand picked states.

Either way, its purposefully fudging the numbers.

[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.

[Elliott_CA]

Originally Posted by tribecavsbrowns

You say these polls got "lucky" -- do you think ABC/WaPo got "unlucky" overestimating Biden by 16 points in Wisconsin or Quinnipiac got "unlucky" overestimating Biden by 12 points in Ohio?

ABC/WaPo and Quinnipiac probably have systematic flaws that need to be reviewed and corrected. But that is independent of and is unrelated to Trafalgar's flawed "shy voter" methodology. It's like a novice gambler who plays slots with a Martingale system -- if they come home a winner, they make the erroneous conclusion that Martingale works.

[tribecavsbrowns]

Originally Posted by Elliott_CA

ABC/WaPo and Quinnipiac probably have systematic flaws that need to be reviewed and corrected. But that is independent of and is unrelated to Trafalgar's flawed "shy voter" methodology. It's like a novice gambler who plays slots with a Martingale system -- if they come home a winner, they make the erroneous conclusion that Martingale works.

I agree that they're independent. I also agree that ABC/WaPo, Quinnipiac, et al., must have systematic flaws (in order to produce such poor results, in one direction, for two straight presidential elections).

But as for Trafalgar, you seem to still think it was luck; hence, the Martingale simile. But you also mention an assumption made by Trafalgar that seems like it was a big reason for them getting closer to the actual margins than most other pollsters.

So, did this assumption (the "shy voter") make Trafalgar's polls more accurate -- in terms of polled margin versus actual margin, for a given state -- or less?

[kell490]

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.