Your Final Projections

A white moon's hot said:
the vast majority of the polls are within the margin of error. while I agree that unabashed confidence in such a tight race is a bit bewildering, the same applies for you. obama does not have this locked up.
Click to expand...

I really hate the term "margin of error". A result within the margin or error is not a "tie", it's just not 95% certain. A candidate who's ahead in a poll *is still more likely to win*. Also, an average of polls has an effectively smaller margin of error than a single poll. While the national polls do seem to mean that Obama's less than 95% likely to win the popular vote, he's leading by about 2.5 points in the average of polls in Ohio, and the margin of error of the average of the polls there is about 1%, so he does have a greater than 95% chance of winning Ohio, and Romney's chance are pretty desperate if that happens.
 
Fluttershy said:
I really hate the term "margin of error". A result within the margin or error is not a "tie", it's just not 95% certain. A candidate who's ahead in a poll *is still more likely to win*. Also, an average of polls has an effectively smaller margin of error than a single poll. While the national polls do seem to mean that Obama's less than 95% likely to win the popular vote, he's leading by about 2.5 points in the average of polls in Ohio, and the margin of error of the average of the polls there is about 1%, so he does have a greater than 95% chance of winning Ohio, and Romney's chance are pretty desperate if that happens.
Click to expand...

margin of error is an apt description of the range of uncertainty. You must not truly understand statistics if that phrase bothers you.


you wrote: Also, an average of polls has an effectively smaller margin of error than a single poll.

Can you explain this? I think you are wrong and you don't know what you are talking about. I've never heard of such a claim.


I googled the phrase and the only link is to the daily KOS site you obviously regurgitated this from
https://www.google.com/search?q=an average of polls smaller margin of error

http://www.dailykos.com/story/2012/11/05/1155699/-A-margin-of-error-question
 
Last edited:
signalmankenneth said:
I predict, Romney will carry the South!

romney_nathan_bedford.jpg


But seriously, Obama wins the electoral college!
Click to expand...

 
tinfoil said:
margin of error is an apt description of the range of uncertainty. You must not truly understand statistics if that phrase bothers you.
Click to expand...

"Margin of error" sound like "Outside this range, there's accuracy, and inside of it, you've just got error and nothing can be gleamed". When in reality, if you lead by something within the MoE, you may not have a 95% certainty of leading by 1%, but you may have a 90% chance, or an 80% chance, which is fairly significant. And ranges outside of the margin of error aren't necessarily certain. Unlikely things happen sometimes.

Now, could you please explain what is wrong with my reasoning that causes me to "not truly understand statistics"? The term both implies too much certainty and too little.

tinfoil said:
you wrote: Also, an average of polls has an effectively smaller margin of error than a single poll.

Can you explain this? I think you are wrong and you don't know what you are talking about. I've never heard of such a claim.


I googled the phrase and the only link is to the daily KOS site you obviously regurgitated this from
https://www.google.com/search?q=an average of polls smaller margin of error

http://www.dailykos.com/story/2012/11/05/1155699/-A-margin-of-error-question
Click to expand...

I don't read dailykos. I got it from fivethirtyeight and, of course, Paul Krugman, who cited them. A poll has an essentially random deviation (the noise), converging on the true results (the signal). When you average multiple measurements of the of the same signal, and only differ in noise, the noise largely cancels itself out, leaving you with a measurement that's closer to the signal. Astronomers also do this when, for instance, they'll average together a bunch of noisy photos of the same object to get a single clear one. Also, the more and more coins you flip and average together, the more you're going to get closer to a result of "50% heads and 50% tails", whereas if you just flip a coin one or a few times, you'll often get a result that differs radically from that.
 
You can even try this at home. Get a noisy ass webcam, take 4 or so photos of your wall, average them together in gimp, and the result will inevitably be clearer than any one photo.
 
So, whose projection was the closest? I know I was off because I gave CO, VA, and FL to Romney.... even though I acknoledged they were too close to call.
 
technically watermark won because he just pasted nate silvers map on here and claimed it for himself as well. so watermark is the clear genius.
 
You must log in or register to reply here.
Back
Top