|
|
On November 04 2012 08:07 jdseemoreglass wrote:
I feel kind of bad for the conservatives who are doing psychological jumping jacks trying to avoid what has been obvious for a while now, and I'm glad I'm not emotionally invested in this election at all. The first time Obama came center stage over 4 years ago I knew he'd be a 2 term president immediately. He did some things I really disagree with, but he's not the anti-Christ. I dare say he was better than W at least.
Given the political atmosphere, I'm pretty sure this post makes you a RINO.
|
On November 04 2012 07:52 paralleluniverse wrote:Show nested quote +On November 04 2012 07:48 sc2superfan101 wrote:On November 04 2012 07:42 Feartheguru wrote:On November 04 2012 07:40 sc2superfan101 wrote:I said earlier that I think RCP is inaccurate. I think Rasmussen is much closer to the truth. they've got Romney tied in OH, for example. SC2superfan: 538 cannot be trusted because it disagrees with RCP, oh wait, that's inaccurate too. is it really so painful to you that I happen to mistrust Nate Silver's polling data? most people agree with you, that Obama is likely to win. I happen to hold a different opinion, in part based on RCP and Rasmussen, and in part based on a gut feeling. I wonder why it is that you are SO positive that 538 is accurate? is it based on any real-time data that you posses, or is it largely the same as with me, which is mainly gut feeling and simple bias? it seems a bit hypocritical for you to cite 538 as absolute evidence, and I can't even cite RCP as a possible alternative. Because RCP uses an unweighted average and that's statistically wrong. Weighting by sample size reduces the standard error of an estimator. Nate Silver weights by sample size, time elapsed, and reliability. What you're doing is just denialism and anti-intellectualism.
it's not denial-ism or anti-intellectualism to suggest that Nate Silver's model might be inaccurate, no matter how much you like him and his blog.
|
On November 04 2012 09:39 DoubleReed wrote:Show nested quote +On November 04 2012 08:07 jdseemoreglass wrote:
I feel kind of bad for the conservatives who are doing psychological jumping jacks trying to avoid what has been obvious for a while now, and I'm glad I'm not emotionally invested in this election at all. The first time Obama came center stage over 4 years ago I knew he'd be a 2 term president immediately. He did some things I really disagree with, but he's not the anti-Christ. I dare say he was better than W at least. Given the political atmosphere, I'm pretty sure this post makes you a RINO.
nah, it just makes him not a conservative. the RINO title is for politicians (and pundits).
|
CNN has an interesting Romney revealed thing on right now, quite a fascinating man.
|
On November 04 2012 10:11 sc2superfan101 wrote:Show nested quote +On November 04 2012 07:52 paralleluniverse wrote:On November 04 2012 07:48 sc2superfan101 wrote:On November 04 2012 07:42 Feartheguru wrote:On November 04 2012 07:40 sc2superfan101 wrote:I said earlier that I think RCP is inaccurate. I think Rasmussen is much closer to the truth. they've got Romney tied in OH, for example. SC2superfan: 538 cannot be trusted because it disagrees with RCP, oh wait, that's inaccurate too. is it really so painful to you that I happen to mistrust Nate Silver's polling data? most people agree with you, that Obama is likely to win. I happen to hold a different opinion, in part based on RCP and Rasmussen, and in part based on a gut feeling. I wonder why it is that you are SO positive that 538 is accurate? is it based on any real-time data that you posses, or is it largely the same as with me, which is mainly gut feeling and simple bias? it seems a bit hypocritical for you to cite 538 as absolute evidence, and I can't even cite RCP as a possible alternative. Because RCP uses an unweighted average and that's statistically wrong. Weighting by sample size reduces the standard error of an estimator. Nate Silver weights by sample size, time elapsed, and reliability. What you're doing is just denialism and anti-intellectualism. it's not denial-ism or anti-intellectualism to suggest that Nate Silver's model might be inaccurate, no matter how much you like him and his blog.
It is if you can't offer up any better reason for it than your gut feeling.
|
On November 04 2012 10:20 Jumbled wrote:Show nested quote +On November 04 2012 10:11 sc2superfan101 wrote:On November 04 2012 07:52 paralleluniverse wrote:On November 04 2012 07:48 sc2superfan101 wrote:On November 04 2012 07:42 Feartheguru wrote:On November 04 2012 07:40 sc2superfan101 wrote:I said earlier that I think RCP is inaccurate. I think Rasmussen is much closer to the truth. they've got Romney tied in OH, for example. SC2superfan: 538 cannot be trusted because it disagrees with RCP, oh wait, that's inaccurate too. is it really so painful to you that I happen to mistrust Nate Silver's polling data? most people agree with you, that Obama is likely to win. I happen to hold a different opinion, in part based on RCP and Rasmussen, and in part based on a gut feeling. I wonder why it is that you are SO positive that 538 is accurate? is it based on any real-time data that you posses, or is it largely the same as with me, which is mainly gut feeling and simple bias? it seems a bit hypocritical for you to cite 538 as absolute evidence, and I can't even cite RCP as a possible alternative. Because RCP uses an unweighted average and that's statistically wrong. Weighting by sample size reduces the standard error of an estimator. Nate Silver weights by sample size, time elapsed, and reliability. What you're doing is just denialism and anti-intellectualism. it's not denial-ism or anti-intellectualism to suggest that Nate Silver's model might be inaccurate, no matter how much you like him and his blog. It is if you can't offer up any better reason for it than your gut feeling.
not really, especially since I can offer up better reasons:
Undecideds usually break for the challenger
Romney has a lot of momentum and Barack isn't polling very well (below 50% in a lot of areas, within the MoE in a lot of states)
Most experienced pundits on either side are calling a close election and some have expressed doubt about Silver's predictions.
Silver has yet to prove himself (one presidential and one mid-term is... well... not enough).
I don't know how Silver decides to weigh his polls or what criteria he uses to determine which polls to use.
and even if I didn't have these reasons, it still wouldn't be denial-ism because I'm not denying any fact (unless you think Nate Silver's predictions are factual, which would not surprise me in the least if you do), and it's not anti-intellectualism because I'm not saying that Silver is wrong or shouldn't use his model, I'm just mildly questioning it's accuracy. skepticism until the facts are out (election day) should be the norm, not the exception.
|
Nate Silver calculates a probability based on an aggregate of polls, weighing them based on size and reliability in the past. The weighting far as "reliability" goes can be slightly subjective, but overall it's pretty transparent statistics-- no funny stuff going on. Anyone with pretty basic stats knowledge and a copy of excel (or some other statistical analysis tool) could do what he does. Unfortunately, he thought of it first and got himself a cushy niche.
Thing like "momentum" and "not polling over 50%" are very vague in statistical terms ad hard to quantify, so generally they aren't. The Now-cast and forecast account for this to some degree, but as election day draws nearer they also converge.
EDIT: Also, Nate Silver had a post about the "undecided voters breaking for the challenger". I think he concluded that based on previous elections that its false.
http://fivethirtyeight.blogs.nytimes.com/2012/07/22/do-presidential-polls-break-toward-challengers/
I think PU mentioned this, but it seems that a significant amount of skepticism towards 538 is because it kind of makes the pundits obsolete. This is one thing I like about sports-- they have the stats and commentators both.
----------
For the sake of not spamming my postcount:
At this point, the election is looking pretty over. Romney really shot himself in the foot with the comments about auto companies moving jobs to China. Now not only the unions but the companies are mad at him.
|
On November 04 2012 10:11 sc2superfan101 wrote:Show nested quote +On November 04 2012 07:52 paralleluniverse wrote:On November 04 2012 07:48 sc2superfan101 wrote:On November 04 2012 07:42 Feartheguru wrote:On November 04 2012 07:40 sc2superfan101 wrote:I said earlier that I think RCP is inaccurate. I think Rasmussen is much closer to the truth. they've got Romney tied in OH, for example. SC2superfan: 538 cannot be trusted because it disagrees with RCP, oh wait, that's inaccurate too. is it really so painful to you that I happen to mistrust Nate Silver's polling data? most people agree with you, that Obama is likely to win. I happen to hold a different opinion, in part based on RCP and Rasmussen, and in part based on a gut feeling. I wonder why it is that you are SO positive that 538 is accurate? is it based on any real-time data that you posses, or is it largely the same as with me, which is mainly gut feeling and simple bias? it seems a bit hypocritical for you to cite 538 as absolute evidence, and I can't even cite RCP as a possible alternative. Because RCP uses an unweighted average and that's statistically wrong. Weighting by sample size reduces the standard error of an estimator. Nate Silver weights by sample size, time elapsed, and reliability. What you're doing is just denialism and anti-intellectualism. it's not denial-ism or anti-intellectualism to suggest that Nate Silver's model might be inaccurate, no matter how much you like him and his blog.
I guess what irks me about the pundit backlash against Silver is that unlike pundits, he's not in the prediction business. He's in the probability business.
What he's doing is no different from what other stats gurus and odd-makers have been doing for ages. Saying that Romney has a 25% of winning is no different than saying a team that is two possessions behind with 45 secs left has a 25% of winning. It's not impossible to come from behind (25% is actually a good chance!) but if you're an outsider betting your reputation or career on it, its a risky bet. I think that's entirely fair to point out.
People shouldn't judge the guy based on whether Obama wins (Romney can always make his hand on the river), judge him based on his math and methodology.
|
on undecideds breaking toward the challenger, I can't say much. Silver's analysis is a bit out of my league with the math, so I don't know how to respond to his findings, whatever they are:
http://www.pollingreport.com/incumbent.htm
|
My main beef with that study is this:
The 155 polls we collected and analyzed were the final polls conducted in each particular race; most were completed within two weeks of election day. They cover both general and primary elections, and Democratic and Republican incumbents. They are predominantly from statewide races, with a few U.S. House, mayoral and countywide contests thrown in. Most are from the 1986 and 1988 elections, although a few stretch back to the 1970s.
It uses data from local/state elections mostly, as opposed to Nate Silver who used past presidential elections. That might be part of why they got different results.
|
On November 04 2012 10:12 sc2superfan101 wrote:Show nested quote +On November 04 2012 09:39 DoubleReed wrote:On November 04 2012 08:07 jdseemoreglass wrote:
I feel kind of bad for the conservatives who are doing psychological jumping jacks trying to avoid what has been obvious for a while now, and I'm glad I'm not emotionally invested in this election at all. The first time Obama came center stage over 4 years ago I knew he'd be a 2 term president immediately. He did some things I really disagree with, but he's not the anti-Christ. I dare say he was better than W at least. Given the political atmosphere, I'm pretty sure this post makes you a RINO. nah, it just makes him not a conservative. the RINO title is for politicians (and pundits).
Thank you for clarifying this for those reading. Conservatism are a political movement that has primarily existed as a faction within the Republican party but is not the Republican party. The ideals of it exist independently of the Republican platform and many times there was never widespread conservative support for the Republican presidential nominee. (Sometime after JFK in the 1960s conservatives by and large departed from the Democratic party, to show that it was not always this way.)
RINO is when you get to elected office and vote with the opposition party. It encompasses those seen as being too willing to compromise too much in congressional bills and those seen by the conservative faction as ideological enemies or not fully supportive. It is a pejorative. Tea party ideals are huge campaigning against incumbents who are seen as RINOs, and 2010 saw a few famous ones being ousted.
|
On November 04 2012 10:27 sc2superfan101 wrote:Show nested quote +On November 04 2012 10:20 Jumbled wrote:On November 04 2012 10:11 sc2superfan101 wrote:On November 04 2012 07:52 paralleluniverse wrote:On November 04 2012 07:48 sc2superfan101 wrote:On November 04 2012 07:42 Feartheguru wrote:On November 04 2012 07:40 sc2superfan101 wrote:I said earlier that I think RCP is inaccurate. I think Rasmussen is much closer to the truth. they've got Romney tied in OH, for example. SC2superfan: 538 cannot be trusted because it disagrees with RCP, oh wait, that's inaccurate too. is it really so painful to you that I happen to mistrust Nate Silver's polling data? most people agree with you, that Obama is likely to win. I happen to hold a different opinion, in part based on RCP and Rasmussen, and in part based on a gut feeling. I wonder why it is that you are SO positive that 538 is accurate? is it based on any real-time data that you posses, or is it largely the same as with me, which is mainly gut feeling and simple bias? it seems a bit hypocritical for you to cite 538 as absolute evidence, and I can't even cite RCP as a possible alternative. Because RCP uses an unweighted average and that's statistically wrong. Weighting by sample size reduces the standard error of an estimator. Nate Silver weights by sample size, time elapsed, and reliability. What you're doing is just denialism and anti-intellectualism. it's not denial-ism or anti-intellectualism to suggest that Nate Silver's model might be inaccurate, no matter how much you like him and his blog. It is if you can't offer up any better reason for it than your gut feeling. not really, especially since I can offer up better reasons: Undecideds usually break for the challenger Romney has a lot of momentum and Barack isn't polling very well (below 50% in a lot of areas, within the MoE in a lot of states) Most experienced pundits on either side are calling a close election and some have expressed doubt about Silver's predictions. Silver has yet to prove himself (one presidential and one mid-term is... well... not enough). I don't know how Silver decides to weigh his polls or what criteria he uses to determine which polls to use. and even if I didn't have these reasons, it still wouldn't be denial-ism because I'm not denying any fact (unless you think Nate Silver's predictions are factual, which would not surprise me in the least if you do), and it's not anti-intellectualism because I'm not saying that Silver is wrong or shouldn't use his model, I'm just mildly questioning it's accuracy. skepticism until the facts are out (election day) should be the norm, not the exception.
He weights by sample size, past accuracy (SurveyUSA and Rasmussen have very good track records), and a "house effect" (PPP leans towards Dems and Rasmussen leans towards Reps, for example).
There might be more that he's added since 2008.
e: Oh, and the age of the poll is weighted using a half-life decay model.
|
On November 04 2012 10:27 sc2superfan101 wrote:Show nested quote +On November 04 2012 10:20 Jumbled wrote:On November 04 2012 10:11 sc2superfan101 wrote:On November 04 2012 07:52 paralleluniverse wrote:On November 04 2012 07:48 sc2superfan101 wrote:On November 04 2012 07:42 Feartheguru wrote:On November 04 2012 07:40 sc2superfan101 wrote:I said earlier that I think RCP is inaccurate. I think Rasmussen is much closer to the truth. they've got Romney tied in OH, for example. SC2superfan: 538 cannot be trusted because it disagrees with RCP, oh wait, that's inaccurate too. is it really so painful to you that I happen to mistrust Nate Silver's polling data? most people agree with you, that Obama is likely to win. I happen to hold a different opinion, in part based on RCP and Rasmussen, and in part based on a gut feeling. I wonder why it is that you are SO positive that 538 is accurate? is it based on any real-time data that you posses, or is it largely the same as with me, which is mainly gut feeling and simple bias? it seems a bit hypocritical for you to cite 538 as absolute evidence, and I can't even cite RCP as a possible alternative. Because RCP uses an unweighted average and that's statistically wrong. Weighting by sample size reduces the standard error of an estimator. Nate Silver weights by sample size, time elapsed, and reliability. What you're doing is just denialism and anti-intellectualism. it's not denial-ism or anti-intellectualism to suggest that Nate Silver's model might be inaccurate, no matter how much you like him and his blog. It is if you can't offer up any better reason for it than your gut feeling. not really, especially since I can offer up better reasons: Undecideds usually break for the challenger Romney has a lot of momentum and Barack isn't polling very well (below 50% in a lot of areas, within the MoE in a lot of states) Most experienced pundits on either side are calling a close election and some have expressed doubt about Silver's predictions. Silver has yet to prove himself (one presidential and one mid-term is... well... not enough). I don't know how Silver decides to weigh his polls or what criteria he uses to determine which polls to use. and even if I didn't have these reasons, it still wouldn't be denial-ism because I'm not denying any fact (unless you think Nate Silver's predictions are factual, which would not surprise me in the least if you do), and it's not anti-intellectualism because I'm not saying that Silver is wrong or shouldn't use his model, I'm just mildly questioning it's accuracy. skepticism until the facts are out (election day) should be the norm, not the exception.
The pundits are incredibly invested in the race staying close to call as long as possible. And there's literally no statistical evidence for Romney having momentum anymore, while Obama has been the only name in the news since Sandy hit (save for Romney's FEMA flubs). And if you want to know how Silver weights his polls, you could always read his freely available methodology...
When Rasmussen, RCP, and 538 are presenting different results AND Silver has done a comprehensive analysis why multiple times justifying why his model is more statistically rigorous is when you run into the only "denialism" out there. Especially when his model is the most transparent one out there (I mean, hell, if you want to talk track records Rasmussen shouldn't be trusted this year whatsoever after it's 2010 run).
Edit: It helps that he even built a margin into his model that allows the polls to be systematically wrong (which is where almost all of Romney's chance of winning comes from).
|
On November 04 2012 12:25 jalstar wrote:
He weights by sample size, past accuracy (SurveyUSA and Rasmussen have very good track records), and a "house effect" (PPP leans towards Dems and Rasmussen leans towards Reps, for example).
There might be more that he's added since 2008.
e: Oh, and the age of the poll is weighted using a half-life decay model.
I like Silver's blog. I think he is over-stating Obama's chances by a small margin, but will accurately predict the outcome. That said, it is not unreasonable to question his model by questioning the models of the underlying polls.
Everything you said about weighting by past accuracy, etc is true but there can still be some fundamental doubt about how accurate a pollster is. It may be the case that many pollsters had a model that worked in 2008. But, you can argue that there is a fundamental asymmetry in finding a correct model. When the correct model aligns with your political ideology, you are more likely to find/believe it than when it contradicts your political ideology. To me, the psychological and economic literature support this argument; it is a form of confirmation bias.
So, if pollsters like Quinnipiac, Marist, and PPP(admittedly) really do have Democrat leanings (as some Republicans have suggested) then we have reason to believe that they would have more reliable models in heavily-Democrat years (like 2008) than in years that are close (2012, by some accounts). Each election is kind of a new ballgame as far as modeling the electorate goes, so it seems like we should investigate the models, despite past accuracy.
The other reason we might question the polling models is that they don't seem to line up with what we've seen from early voting so far. Based solely on the number of ballots returned that are affiliated with each party in a number of swing state, it seems pretty clear that Republican turnout is up from 2008 (significantly) and Democrat turnout is down (slightly). Whether this will carry through to election day is obviously uncertain.
Together, the perceived enthusiasm among Republicans (and relative lack among Democrats) and the data from early voting give us a good reason to reexamine the polling models. Personally, I think that the most likely case is that Republicans narrow the turnout gap that was so key in 2008, but not by enough to win. I think that polls that assume an even wider turnout gap than in 2008 (and there are quite a few out there with part ID that favors Democrats more than the 2008 data) are unreliable.
|
I'm mystified by why people love and hate Nate Silver so much. And why people are discussing his methodology so much.
Although I guess similar to controversies with the financial system or tax policy, it's nice to see people learning and talking about statistics.
|
Cayman Islands24199 Posts
interest in silver seemed to balloon after some republicans began bullying him. i mostly know him from baseball lol
|
As a Canadian, I just want a US president that has good international ideas and policies.
If was an American, I would vote Obama - Mitt just seems like a rich scoundrel thats going to feed tax breaks to his fellow millionaires, and let the rest of america carry on the broken backs of the poor and low to middle class, while the upper class eats lobster.
|
i like silver because he's logical and organized, and he does calculations i understand but am too lazy to do. i will admit there's a certain amount of circle jerking though because he confirms what i want. 
|
On November 04 2012 13:34 ticklishmusic wrote:i like silver because he's logical and organized, and he does calculations i understand but am too lazy to do. i will admit there's a certain amount of circle jerking though because he confirms what i want.
Yes I'll say I get a certain sort of self-serving satisfaction out of going down the right column of the 538 blog and thinking, "Yep, makes sense, makes sense." 
|
On November 04 2012 13:19 coverpunch wrote:
I'm mystified by why people love and hate Nate Silver so much. And why people are discussing his methodology so much.
Although I guess similar to controversies with the financial system or tax policy, it's nice to see people learning and talking about statistics.
I love it, I also love seeing statistics in action as I"m currently taking a political statistical analysis course as part of my M.P.A., reading Nate Silver's methodology is very interesting to me.
|
|
|
|
|
|
EPT
