Post Comment Preview Comment
To reply to a Comment: Click "Reply" at the bottom of the comment; after being approved your comment will appear directly underneath the comment you replied to.
View All
Favorites
Highlights
Recency  | 
Popularity
Page:  « First  ‹ Previous  1 2 3 4 5  Next ›  Last »  (15 total)
photo
HUFFPOST SUPER USER
lapdogs
Avid News Reader
01:37 PM on 11/09/2012
Nate Silver (five thirty eight) was by far was the best !!!

He nailed every state and DC - and when the counting gets finished in Florida, Obama will win that state. (Leaning for Obama right now) like Nate Silver predicted.

As for Karl Rove, we'll, um, um, ....
photo
HUFFPOST SUPER USER
John Nemesh
12:54 PM on 11/09/2012
Now that we can see just how badly the polls were biased in favor of Romney and the GOP. Hopefully, these guys take the hint and fix whatever problems they have. We aren't interested in polls that reinforce ideology, we are interested in polling that reflects the truth!
photo
talking point squasher
We've got the trickle down frown...
12:51 PM on 11/09/2012
So what does this mean? Well, it could mean that our data set does not represent the population (bingo!), it can alos mean that there are special causes for the data points in question that are signaling these data points may not be representative of the population (bingo2!). It would suggest a non-normal data set for which most traditional statistics cannot be applied. This is why I content Nate Silver is using a lot of dark magic in his calculations. My guess is that he unbiases the data somehow, probably through historical anaylsis of these polls, and factors an offset of some sort, or some form of coefficient or algorythmn to somehow nromalize the data. The problem here is that you'd need to know your population mean ahead of time, which of course, he does not. My assumption is that he has figured out, statistically, some sort of bias factor in the data sets, probably by analyzing and characterizing the bias of each poll individually (because bias could be different poll to poll) over time (previous elections) and applying those coefficients to the new data sets.
And that's why someone like Silver can be accurate. People read these polls and look at the absolute data nad rush to judgment. Silver instead does statistical analysis of the data to find out what it really means, and looks beyond bias from special causes (not calling cell phones, for instance) to arrive at a statistically relevant conclusion.
This user has chosen to opt out of the Badges program
photo
05:07 PM on 11/09/2012
All you need to do is look at Gallup and Rasmussen and reverse the numbers.
photo
talking point squasher
We've got the trickle down frown...
12:51 PM on 11/09/2012
Listen, this polling nonsense is junk science. As a six sigma guy fairly well versed in statistics, I can tell you that there is little statistically speaking, that is correct about how this polling is done, and how the data is represented. Use common sense, in a nation as diverse as this, does anyone believe a survey of 1000 people is enough to be accurate on these elections within ± 3-5%?
And for further proof, consider the first table in the article. Only 6 of the 28 polls detailed in the table have Obama = or > 50%. So, assuming we consider the sum of the polls as our data set, what you would have is data where the actual result isn't centered within the polling data set. So, if we took these polling result and plotted them on a bell curve (let's stick with the Obama %), we should expect that we would see samples on both side of the mean. If we round up Obama's actual % to 51%, what we have is only 1 data set that resides on the actual population mean, and EVERY OTHER datapoint is biased to the low side of the bell curve.
This user has chosen to opt out of the Badges program
photo
05:15 PM on 11/09/2012
Don't blame statistics, blame the underlying assumptions that lead to using poor methods: over sampling republicans, misweighing cell phone responses, and using a more stringent definition of a likely voter. The there's the robocalls,etc.

Also, politics has creeped into the polling business. Gallup was one of the best until the family sold it. It's now a GOTP mouthpiece.
photo
talking point squasher
We've got the trickle down frown...
03:55 PM on 11/12/2012
Absolutely correct.  That is largely my point.  Statistics are garbage in, garbage out.  The hardest thing to do in statistical analysis, in my view, is to generate a truly unbiased, random sample of the population.
photo
HUFFPOST SUPER USER
JohnSawyer
arglebargy
05:22 PM on 11/09/2012
Check election.princeton.edu for a look at how even disparate data can be analyzed properly, as long as you have enough of it to cancel out poll biases.
photo
talking point squasher
We've got the trickle down frown...
03:57 PM on 11/12/2012
We do similar running thermal models predicting free convection, I assume it is a similar principle.  Calculate an offest based on a comparsion of the models vs actual data would be my guess...
photo
HUFFPOST SUPER USER
rgeze
get schooled by the facts
12:49 PM on 11/09/2012
Nate Silver or bust. Whoop! Go Nate!
photo
HUFFPOST SUPER USER
tepeters
12:48 PM on 11/09/2012
This is why you need someone like Nate Silver-he got it right. He knew how to aggregate and weight these widespread polls. Thank God for the nerds!!!
12:41 PM on 11/09/2012
I watched the polls closely and always felt the "Romney Surge" was parts of the media's effort to keep the polling close in order to keep the public fully engaged, it worked on me, I was nervous up until they announced Ohio for Obama.
photo
talking point squasher
We've got the trickle down frown...
12:41 PM on 11/09/2012
Guys, I was saying this all through the election season. Anyone trying to read this election on national polls was going to have it wrong. They underpredicted Obama and I PREDICTED thi would happen, I can provide links to many of my comments over the last month.
The reality is that these pollsters have no way to assure that they are taking a purely random sample of their populations being surveyed. Furthermore, they are largely relying on landline phones. I would also be curious as to how they ensure that they get folks who work nights and weekends, meanin, do they make the calls during the same time at night? If so, they fail to sample working class Americans, many of whom broke for Obama.
Bottom line, it is my sincere hope that these polls largely go away. They did not add any value to this political season, other than providing fodder for talking heads to trumpet the good news or twist themselves into pretzels to refute the bad news.
One thing that should be readily apparent, is unless polls are more heavily weighting cell phone and internet users, there polls will bias GOP, because who are the ones who still use landline phones and are waiting to take calls at 8PM at night? That's right, the elderly, who largely broke Robme.
12:27 PM on 11/09/2012
I ignored all of the polls and just monitored Nate Silver's 538 blog. He nailed it. Like he always does. If you were following Nate you knew that Romney never came close to winning this thing. He never came to close to being in the lead. So I was not surprised at the outcome.
javagirl023
NRA, making a list and checking it twice!
12:30 PM on 11/09/2012
Aggregating the polls. Called every state.
This user has chosen to opt out of the Badges program
photo
01:09 PM on 11/09/2012
Bingo!
12:27 PM on 11/09/2012
Would everyone here PLEASE stop referring to Nate Silver in this context?! NATE SILVER IS NOT NOT NOT A POLLSTER. He uses the poll results of others as data points for his mathematical models and algorithms that produce those remarkably accurate predictions. But he depends on the pollsters for the numbers he crunches.
This user has chosen to opt out of the Badges program
01:37 PM on 11/09/2012
Nevertheless, he did get the numbers right.
01:56 PM on 11/09/2012
Indeed. I believe his only miss was the North Dakota U.S. Senate race.
This user has chosen to opt out of the Badges program
photo
05:18 PM on 11/09/2012
He also depends on the history of each polls past predictions. This allows him to reweigh the current findings.
photo
HUFFPOST SUPER USER
ejay579
MURKA! Numba one 4 EVA!
12:26 PM on 11/09/2012
Gallup and Rasmussen, the most biased polls. No surprise there. These are the two the wingnuts I know almost invariable cited in the weeks and days leading up to the election.
photo
HUFFPOST COMMUNITY MODERATOR
Safire
Vote! Vote as if your life depends on it.
12:23 PM on 11/09/2012
Gallup lost ALL credibility! What the heck happened to them this season?
photo
Brush with Death
Retired First Sgt. tree hugger.
12:24 PM on 11/09/2012
They sold out. If you look you will probably see that they were bought by a company that is owned by Mitt.
photo
HUFFPOST COMMUNITY MODERATOR
Safire
Vote! Vote as if your life depends on it.
12:31 PM on 11/09/2012
Aaah! That explains it!
HUFFPOST SUPER USER
VonMarco
Common Sense is not so Common
12:20 PM on 11/09/2012
Nate Silver and Intrade...........accurate!
Republican crybabies
Enemy of plutocrats
12:19 PM on 11/09/2012
Here's a thought - maybe landline polling will be less and less accurate going forward.
HUFFPOST SUPER USER
patdamenace
12:12 PM on 11/09/2012
Nate Silver is the only one who got it right. Anyone else was just a big joke.
photo
HUFFPOST SUPER USER
CPAwADD
Always look on the bright side of life.
12:18 PM on 11/09/2012
Sam Wang did very well too, but Nate was a little better. Interesting all of the aggregators had small Republican biases probably caused by Gallop and Rasmussen.
12:19 PM on 11/09/2012
(irritated sigh). Nate Silver is a very admirable fellow, but as he would quickly tell you he is NOT a pollster. He aggregates, averages and extrapolates the polls of other. He therefore is as relevant to the above discussion as Sammy Davis Jr.
HUFFPOST SUPER USER
patdamenace
12:45 PM on 11/09/2012
Calm down, Dunraven. I know Nate is not a pollster. Still, to me, he was the one I trusted the most. He might not be relevant to the discussion but nothing wrong with mentioning him. Like I said during the campaign, he was the only one I trusted with who was going to win the election. Not Gallup, Rasmussen or any other polls.
01:39 PM on 11/09/2012
No one is referring to him as a pollster. Nevertheless his approach is very relevant to the point of this article which is that given that all polls have bias, and Nate Silvers method, like others minimizes that bias.