Posted on 10/29/2012 2:04:08 AM PDT by MacMattico
For those DU lurkers reading this they'll probably squeal with delight at this thread, but is Nate Silver's record of political predictions really as great as is posted all over the Internet? I know he's with the NY slimes now, but he did predict the house going R in 2010, albeit only with 45 seats changing sides. (low prediction R) and 2008 he was really on, although that was much easier to call. It's killing me because I'm on pins and needles. I like to think with Gallup and Rasmussen actually in agreement Romney is up, but what am I to think when I hear of Silver's track record and that of some statistician named Wang? I like it that Kasich says we're taking Ohio, but I'm dying here!
This is NOT an ordinary election year!
LLS
I’ve been reading Nate Silver’s stuff for quite a while, so here are a couple of things Nate himself would tell you:
1. There’s a 25% chance of a Romney win. That’s not 0%. The predictions are based on polling and this 25% represents the “known unknowns” in the polling data.
2. There are also “unknown unknowns” in the polling data. There’s no reason why Romney couldn’t out-perform his polling.
3. Romney is ahead in national polls. State polls show Obama winning. Nate Silver doesn’t have a good explanation for this, and is willing to admit that this suggests that poll data is weird this year.
As far as what other people have claimed:
1. Nate Silver has never worked for Daily Kos.
2. The 2008 Obama campaign did not give him internal polls.
I don’t know where people got any of that. In any case, it wouldn’t matter. The numbers are what they are.
He’s only as good as the polls he’s using. He is just running joint probabilities based on the odds in the state-level opinion polls. If the polls, in the aggregate, are right, then he’ll be right. He also weights the polls on a ranking system based on how good he thinks they are, and he does not give Rasmussen top marks. If he’s right about that, then fine.
He also rejects the idea that the pro-Dem weighting of so many of the polls is from anything other than Dem turnout being high this year, which I have a feeling is going to be his downfall.
You got it exactly right. He is a Kos radical who got picked up off their boards by the axelrod team in 2008. H looked good in 2008 because it was an easy election to call. It is easy to predict a great outcome when it matches up with your ideology. In 2010 he was WAY off. Everyone with a brain knew we would win the House. He tried to lower expectations to keep his base enthused. Same in 2012. Lots of wishful thinking.
he doesn’t even conduct his own pols- he cherry picks other polls to come to his conclusion....there’s a word for that- using the work of others- its called lazy....
Are Rasmussen and Gallup always right? No. But which polling firms get it right most of the time for the past 8 years? People here need to pick their sources and stick with them. For me, it’s Rasmussen. When his polls indicate we’re in trouble I’ll start fretting. All other polls are noise to me.
Silver is a mediocre analyst who has benefitted from applying some basic math to fields filled with people who don't understand math: baseball & politics.
Your claims are suspect:
#1 is false according to Wikipedia.
#2 has been widely reported, and to my knowledge, never refuted.
Please post links to support your denial of these claims.
Although meteorology is getting better, good polling is getting harder and more expensive due to: changing technology (cell phones, caller ID, and privacy restrictions); low response rates; rising public distrust of telephone calls from strangers due to the high volume of calls that are scams or unwanted cold call solicitations; the use of cheap computer polling, bad phone lines, and poorly qualified staff; the piggybacking of public polls with commercial polls; the high rate of absentee and early voting that distorts sampling; the difficulty of capturing last minute shifts in public sentiment; and uncertainty over who will actually vote in any given election.
Good polling -- the kind that Presidential campaigns rely on for decision making -- is horrifically expensive because it corrects for those and other issues with trained and reliable staff, pricy databases, and demographic and sampling adjustments designed by top line analysts and data modelers with doctorates and better.
Most published polls are cheaply done by second and third tier pollsters. In contrast, Presidential campaign polls are expensive and the results are tightly guarded. There is a way to tell though what the campaigns think is happening: be alert to the fundamentals of the election and watch the ad buys and candidate travel.
For example, Obama's cancellation of his central Florida trip set for today tells us that he and his campaign think his time is better spent preening on national TV so as to exploit the news value of Hurricane Sandy. It also tells us that Florida is tilting to Romney and is no longer worth Presidential campaign time.
Similarly, the expanded Romney campaign effort in the upper Midwest suggests that his campaign sees potential gains there, especially in Wisconsin. In addition, independents and undecideds seem to be going toward Romney, generic GOP identification is at an historic high, and the GOP base is far more fired up than it was in 2008, with Obama's base not fully engaged.
My guess is that Silver's final take on the election will be that it tilts Romney but is too close to call for certain. Meanwhile, as for what a journalist and former pollster thinks, take comfort from Michael Barone's belief that Romney will win.
As for published polls, the bipartisan Battleground is usually the best, with the raw data publicly available and competing partisan analyses by the Tarrance Group and Lake Research. And Tarrance is not just the best GOP pollster, they are the best in the business.
The Weekly Standard reports that the Battleground poll now projects Romney winning by 5 per cent: New Projection: Romney 52, Obama 47.
If that is accurate -- and I am convinced that it is -- Romney wins, with major GOP victories at every level of the ballot.
Although meteorology is getting better, good polling is getting harder and more expensive due to: changing technology (cell phones, caller ID, and privacy restrictions); low response rates; rising public distrust of telephone calls from strangers due to the high volume of calls that are scams or unwanted cold call solicitations; the use of cheap computer polling, bad phone lines, and poorly qualified staff; the piggybacking of public polls with commercial polls; the high rate of absentee and early voting that distorts sampling; the difficulty of capturing last minute shifts in public sentiment; and uncertainty over who will actually vote in any given election.
Good polling -- the kind that Presidential campaigns rely on for decision making -- is horrifically expensive because it corrects for those and other issues with trained and reliable staff, pricy databases, and demographic and sampling adjustments designed by top line analysts and data modelers with doctorates and better.
Most published polls are cheaply done by second and third tier pollsters. In contrast, Presidential campaign polls are expensive and the results are tightly guarded. There is a way to tell though what the campaigns think is happening: be alert to the fundamentals of the election and watch the ad buys and candidate travel.
For example, Obama's cancellation of his central Florida trip set for today tells us that he and his campaign think his time is better spent preening on national TV so as to exploit the news value of Hurricane Sandy. It also tells us that Florida is tilting to Romney and is no longer worth Presidential campaign time.
Similarly, the expanded Romney campaign effort in the upper Midwest suggests that his campaign sees potential gains there, especially in Wisconsin. In addition, independents and undecideds seem to be going toward Romney, generic GOP identification is at an historic high, and the GOP base is far more fired up than it was in 2008, with Obama's base not fully engaged.
My guess is that Silver's final take on the election will be that it tilts Romney but is too close to call for certain. Meanwhile, as for what a journalist and former pollster thinks, take comfort from Michael Barone's belief that Romney will win.
As for published polls, the bipartisan Battleground is usually the best, with the raw data publicly available and competing partisan analyses by the Tarrance Group and Lake Research. And Tarrance is not just the best GOP pollster, they are the best in the business.
The Weekly Standard reports that the Battleground poll now projects Romney winning by 5 per cent: New Projection: Romney 52, Obama 47.
If that is accurate -- and I am convinced that it is -- Romney wins, with major GOP victories at every level of the ballot.
Wikipedia says that Nate Silver POSTED on Daily Kos as “poblano”. This is true. He was not, however, employed by Daily Kos. If I wrote a Wikipedia article about you, it might say, “so and so posted as ‘Skilllspitter’ on Free Republic,” but you don’t work for Free Republic, do you?
“#2 has been widely reported, and to my knowledge, never refuted.”
1. Nate Silver has always maintained that he does not include campaign polls in his averages because campaigns “cherry pick” what they release so you can’t get an accurate history of their polling.
2. Nate Silver has always listed each and every poll that he DOES include and the numbers add up.
Something being “widely reported” does not make it true. Do you have any other support for this claim other than “widely reported”? An actual source?
You have set up a straw man to fight. No one suggested the Obama internal polls were included in Silver's averages.
However, it has been widely reported that the Obama campaign provided Silver with their internal polling in '08 .
Do you continue to deny this, and on what evidence?
Wikipedia says that Nate Silver POSTED on Daily Kos as “poblano”. This is true. He was not, however, employed by Daily Kos.
"Employed by" is an unfortunate turn of phrase by whoever used it, but clearly he was a Kos contributor, which you have now admitted after denying he "worked for" the site. I don't think many blog contributors, in general, are paid a salary, but they are still working for and contributing to the site.
“Do you continue to deny this, and on what evidence?”
If the claim is that Silver HAD Obama’s internal polling, but didn’t put it in his average, then I guess I don’t deny that. I don’t believe it — and you haven’t provided any source — but maybe it’s true. If Silver didn’t include the polls in his analysis, then I don’t see why it matters.
“I don’t think many blog contributors, in general, are paid a salary, but they are still working for and contributing to the site.”
OK, so we agree on the basics here. Again, I don’t see why it matters. When he started in 2007, he could have posted his articles here on Free Republic, but he would have been zotted — not because he was wrong, but because no one wanted to hear it. That’s not necessarily a bad thing; FR is an activist site and should be focused on firing up the troops, not bumming them out. If we are going to bring up the topic of Nate Silver, however, we should focus on factual arguments about polling data and not innuendo (ZOMG he posted on Daily Kos! Freak out!). That’s why I brought up these side issues.
“However, it has been widely reported that the Obama campaign provided Silver with their internal polling in ‘08 .”
I finally found the source of this (it was mentioned in another place today, this time with a link). It was reported by Buzzfeed that the 2008 Obama campaign asked Silver to run their internal polls through his model. I don’t see what bearing this has on the validity of his publicly-available results.
Romney is consistently ahead by 4-5 points in national polls and he has him at a 25% chance to win? Come on.
The guy is a partisan hack.
You are very quick to defend this Silver fraud. Is he a personal friend of yours?
“Romney is consistently ahead by 4-5 points in national polls and he has him at a 25% chance to win? Come on.”
Here’s the RCP average page:
http://www.realclearpolitics.com/epolls/2012/president/us/general_election_romney_vs_obama-1171.html
Perhaps you could find a time this year where Romney was ahead 4-5 points in national polls. Individual POLL, yes, but not when you average all the various polls together.
In any case, national polls don’t matter. If we had a national vote, that might be significant. Instead, we have an electoral college where 48 of the 51 contests award all their electors to the winner. What matters is state polling, and not just one poll, but the consensus among polls.
“You are very quick to defend this Silver fraud. Is he a personal friend of yours?”
No, I just think he’s got a lot of solid analysis of the state of the polls. His methodology is not without flaws, but I’d rather focus on logical arguments about what he says rather than resort to guilt-by-association attacks.
Look, the poster was just freaking out about Nate Silver’s predictions and I was just pointing out that ever Nate Silver admits that Mitt Romney still has a shot at winning the election. I don’t know why anyone has a problem with that. It seems to me that people are acting scared and defensive. Don’t fall into that trap.
Which is like mixing ice cream with dog sh*t. Only an idiot or someone with an agenda would program garbage into their model.
Nate Silver admits that Mitt Romney still has a shot at winning the election.
You're trolling.
“It was reported by Buzzfeed that the 2008 Obama campaign asked Silver to run their internal polls through his model. I don’t see what bearing this has on the validity of his publicly-available results.”
Because his “mystique” doesn’t come from his present “results,” but from his nailing the election exactly in 2008.
Internal polls are far superior than the public polls published by the newspapers and news networks, and even the big polling firms. They have to be: a lot is running on them. The public polls don’t show PA all that close, for instance, but you can tell that it is (or at least both campaigns’ internal polls say it is) because with precious little time and every dollar counting now, both are going to PA this week.
The accuracy, once you know what you are aiming at (which the internal poll will give you), is easy to achieve by manipulating public polls in a system which is about as transparent as the Cook County assessors calculation on your real estate tax (btw, Nate was at the U of Chicago while Obama-or as I call him, Hussain-was there, as was I. The U of C, the Chicago Machine, etc. small little interconnecting worlds)
Of course, that the newspapers and media are propagandizing for Obama, not only in their reporting, but their polling as well (oversampling Demcraps, focusing on registered rather than likely voters etc.), makes Silver’s job all that more easy. Tinkering with the average of lopsided polls to the result your want, child’s play.
So why do it? Well, look at the fright it caused here. LOTS of the left call it all over, and cite 538 as their authority. Massage the numbers and mass distribute to demoralize your opposition, and the polls can not only reflect reality, they can be made to effect it. Or try to.
Btw, Rasmussen got it exactly right in 2004 and 2008, and without internal help. If their poll showed Romney in trouble, it would be cause for panic. As for Silver, he’s just shilling for Hussain.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.