The definitive work on the oversampling of Ds in the polls.

[meazza]

I feel like you are trying to negotiate with the *fact* that Obama is winning in this election. This goes along perfectly with my favorite saying on this board : Republicans and facts are like oil and water.

 

Instead of fighting with facts, another way to deal with them is to just accept them.

 

How can you say that a random sample of polls is "fact"?

 

The odds are that he could win this election based on the random samples performed ad nauseum but that isn't "fact".

Edited November 1, 2012 by meazza

[OCinBuffalo]

http://www.cjr.org/s...obabilities.php

 

I guess everyone is wrong...

His predictions in 2010 were great as well.

 

http://en.wikipedia....-term_elections

2 things:

 

1. Read your own wiki link. Ron Klain is a democrat. In fact he's the hero in the "Republicans stole the 2000 election movie" on HBO. Why would a Democrat, who knows this, find fault with Silver...if he "just want's to hear what he wants to hear"? The fundamental problem with Silver: he is using bad demographic assumptions. We can talk about models all day, but if you don't have a sound electorate represented in your averages(raw data), all you are doing is averaging bad.

 

Also, and I have been waiting patiently with this: Silver uses an "ideology score" or rating, does he not? Read your own info. No. Let me do it for you: http://fivethirtyeig...choice-of-ryan/ Yeah...that's right, I've had this queued up since Ryan became VP.

 

IF Silver hasn't modified his "ideology" scores and is still rating Romney as arch-conservative...then isn't he badly misrepresenting voter's views of Romney? But, Romney's ideology hasn't changed...or, wait...has it? Well? Has it? Your side keeps saying it has. Yeah. Warning: you have now stepped into it ...let's see try to get out of it. :lol:

 

I've been waiting to use this one, and now seems like the right time. It's going to be fun seeing you try to argue both for and against Romney being a moderate, at the same time. For, to try and preserve what Obama has been saying about "you can't trust Romney", and against, to try and keep your Nate Silver argument going.

 

This is going to be fun! Wriggle..wriggle...wriggle. EDIT: The simple fact is that the "ideology score" itself, as a concept, that heavily informs Silver's work here, is flawed. This is perhaps the single biggest factor that breaks Silver's model. He is trying to assign a single number...to a thing nobody agrees on this time, and certainly not Democrats. Last time, the ideology score may have had less of a factor, or it may have been congruent with the rest of the data. This time...we'll see.

 

2. I wonder, what are you going to say on Monday, when Nate Silver does his last minute shift towards Romney, with Obama still winning, but just enough to be "margin of error" territory? Are you going to tell me that it's just because the new raw data he received moved his models that way?

 

I was watching him in 2010. That's precisely what he did. Notice that the entire wiki never talks about where he was 6 months out from 2010, 3 months, etc. No. It only talks about the "final prediction" on "election eve".

 

Coincidence?

Edited November 1, 2012 by OCinBuffalo

[fjl2nd]

2 things:

 

1. Read your own wiki link. Ron Klain is a democrat. In fact he's the hero in the "Republicans stole the 2000 election movie" on HBO. Why would a Democrat, who knows this, find fault with Silver...if he "just want's to hear what he wants to hear"? The fundamental problem with Silver: he is using bad demographic assumptions. We can talk about models all day, but if you don't have a sound electorate represented in your averages(raw data), all you are doing is averaging bad.

 

Also, and I have been waiting patiently with this: Silver uses an "ideology score" or rating, does he not? Read your own info. No. Let me do it for you: http://fivethirtyeig...choice-of-ryan/ Yeah...that's right, I've had this queued up since Ryan became VP.

 

IF Silver hasn't modified his "ideology" scores and is still rating Romney as arch-conservative...then isn't he badly misrepresenting voter's views of Romney? But, Romney's ideology hasn't changed...or, wait...has it? Well? Has it? Your side keeps saying it has. Yeah. Warning: you have now stepped into it ...let's see try to get out of it. :lol:

 

I've been waiting to use this one, and now seems like the right time. It's going to be fun seeing you try to argue both for and against Romney being a moderate, at the same time. For, to try and preserve what Obama has been saying about "you can't trust Romney", and against, to try and keep your Nate Silver argument going.

 

This is going to be fun! Wriggle..wriggle...wriggle. EDIT: The simple fact is that the "ideology score" itself, as a concept, that heavily informs Silver's work here, is flawed. This is perhaps the single biggest factor that breaks Silver's model. He is trying to assign a single number...to a thing nobody agrees on this time, and certainly not Democrats. Last time, the ideology score may have had less of a factor, or it may have been congruent with the rest of the data. This time...we'll see.

 

2. I wonder, what are you going to say on Monday, when Nate Silver does his last minute shift towards Romney, with Obama still winning, but just enough to be "margin of error" territory? Are you going to tell me that it's just because the new raw data he received moved his models that way?

 

I was watching him in 2010. That's precisely what he did. Notice that the entire wiki never talks about where he was 6 months out from 2010, 3 months, etc. No. It only talks about the "final prediction" on "election eve".

 

Coincidence?

 

I don't see anywhere where Silver says he incorporates the ideology score into his model. That chart is based off of another statistical model he links to.

 

Maybe I missed it, but I'm sure you have links from his with his up to date scores? Probably not. Romney is a moderate....maybe. He is nothing. He says what people want to hear.

 

Silver doesn't really use bad demographic assumptions. He uses the polls that are available. Why are the polls getting these demographic problems if you say the electorate is completely different? Are they purposely rigged?

[OCinBuffalo]

I don't see anywhere where Silver says he incorporates the ideology score into his model. That chart is based off of another statistical model he links to.

 

Maybe I missed it, but I'm sure you have links from his with his up to date scores? Probably not. Romney is a moderate....maybe. He is nothing. He says what people want to hear.

 

Silver doesn't really use bad demographic assumptions. He uses the polls that are available. Why are the polls getting these demographic problems if you say the electorate is completely different? Are they purposely rigged?

I thought you "knew" all about Silver's methodology, to the point that you could not only defend it, but come after me for not "knowing" it. Well? Instead, it ends with me: teaching.

 

Nate Silver has determined, wrongly, that baseball-->politics. He believes that his data patterning methodology can use an "ideology score" no different than he used On Base %. Understand: his patterning methods are formidable. However they are dependent on statistics, that perform in baseball, but don't perform, the way his patterning methodology REQUIRES them to, in politics. They are also dependent on polls(more on that below).

 

Why does he need "ideology score"? Same reason he needs OBP. His models won't work otherwise. He needs "batter performance". He needs a way to identify how a candidate will perform against the "pitching" = the economic factors, voters aggregate views on issues, etc.

 

The fundamental flaw of "ideology score": in baseball, a single is a single, and both the Yankees and Red Sox see it as such. OBP is a composite statistic(or indicator) of different %s and #s, and the relationship between the "children" to the "parent" is always he same.

 

Ideology score is also a composite statistic. But it's a crappy one, compared to OBP. The relationship of the parent to the children will change. To account for that you must redefine the scoring system, re-run it, and redefine everything it touches in Silver's model, practically on a weekly basis. Silver uses this stat as though it is OBP. It simply isn't. It is unlikely to be measured objectively. You'd have to have somebody besides Silver, or ideally, a whole lot of somebodies, and do this weekly for it to even approach being right.

 

So what is this score already?

 

He uses this: http://www.voteview.com/dwnomin.htm, this http://ideologicalcartography.com/ and polls of what voters think the ideology of a candidate is (yeah...as if the average voter is familiar enough to make consistent, accurate calls like that. There is a reason Wall Street analysts have a job.) to determine the "ideology score".

 

Look at those links. See anything that looks like Single, Double, Bunt? Neither do I. Is there anything here that will behave the same way over time, or that people will perceive the same way as a single? Is this even sorta consistent? Or, simply stated: was Mitt Romney ever a US Congressman?

 

Ideology score is not OBP. You don't get to rate a single in baseball. Or, call a single not a single, because you think it was a lame blooper that the batter got lucky on.

 

But, if you are doing DW-Nominate and you are a Democrat, the difference between Ron Paul, and Mitt Romney is what? 5 pts? You say that because "Paul is such an EXTREMIST!" This has massive effect in this model. However if you were to rate Howard Dean, and Barack Obama...the difference would be 1-2 pts, because you don't see as much difference between them.

 

See the inherent bias? A 1000 other voters may rate this 1000 other ways.

 

The funny part: what we are really looking at via all of this? Nate Silver's partisan, subconscious perception of what the odds -->should be. We still don't know who he rated how, why, and whether this has changed. It's a unintentional view into Silver's mind.

 

So, again, Nate Silver is a moron. He's brilliant, but, he's a moron for not seeing the flaws here.

 

--------------------------------------

 

As far as the polls: I think it's safe to say we've established that there is a serious out of whack, partisan divide here, when it comes to demographic models. Somebody is dead wrong. We can't know that until election day.

 

We know that Axelrod has been cajoling pollsters for at least a year, to accept his version of the demos of the 2012 electorate.

This model, in my estimation, is fantasy. When the polling firms gets a raw sample, they weight it to fit this model. Example: if I get too many white males, and not enough black females, I weight down my sample to fit the Axelrod demo model. So, some voter's responses get thrown out, because my sample is "too white".

 

There doesn't necessarily have to be anything underhanded going on here. Pollsters have to start somewhere. Many chose to use the demographic model of 2008, and adjust to suit Axelrod's whining. They say that they don't weight by party ID. Fine. But, they do use weight by demos.

 

When your sample is consistently producing fantastical D+6-10 turnout, but also has Romney winning independents by 10+? It's time to reconsider that demographic model, and my main B word here is: they haven't.

 

Nate Silver's main problem here is: he is using all of these polls, and down-weights Rassmussen and Gallup, who have told Axelrod to get F'ed. So, Silver is making a potentially bad problem....worse.

 

IF Axelrod's model is wrong, Silver is hopelessly screwed, and so is RCP for that matter.

[fjl2nd]

Are you Dean Chambers??

 

http://www.slate.com/blogs/weigel/2012/09/24/dean_chambers_meet_the_guy_who_s_re_weighting_polls_to_show_romney_way_ahead_of_obama.html

[OCinBuffalo]

Are you Dean Chambers??

 

http://www.slate.com...d_of_obama.html

No....that guy is an even bigger idiot, but for different reasons. The way he "unskews" is retarded, because it assumes that all of the polls are weighted using the same methods. They aren't. In fact they can't be. Pollster's "claim to fame" is that their methods are both unique and better, which is why they are more accurate.

 

EDIT: Before you even start: no, the demographic model that a pollster uses is not the same thing as their weighting methods. These are two completely different things.

 

So, if you take a poll and "unweight" it, to get back to the raw data, so that you can re-weight it by, in this case, Party ID, you have to use the same method that was used to weight it in the first place. Otherwise, you are no better than the people you are complaining about.

 

This guy doesn't know the method that was used for each poll he "unskews", so he can't do it properly, so he is an even bigger moron than Silver.

 

I can explain in further detail...

 

...but it appears you haven't finished your Nate Silver work yet.

 

Look 4 years ago....I took down Charlie "I'm a football scientist!" Idiot Joyner. I had all these clowns telling me I was wrong...until....I wasn't. Don't be a clown.

Edited November 2, 2012 by OCinBuffalo

[fjl2nd]

OC, I will give you plenty of credit if you are right about all of this. Unlikely, but you would deserve it.

 

The race boils down to this: On Friday, Obama led in 19 battleground polls. Romney led in one.

[OCinBuffalo]

OC, I will give you plenty of credit if you are right about all of this. Unlikely, but you would deserve it.

I am simply applying what I know. I cannot unknow what I know. Asking me to take these polls at face value, since August, is asking me to unknow. Also...I won't necessarily be "right". I am merely offering criticism. I don't have my own model that I am offering as an alternative ( I have a job already). That would make me "right". The thing is: my criticisms can be valid, and, Obama can still win in spite of them. What if Silver is off by 20 pts, but Obama wins the election? He's still not wrong...in the eyes of partisan dopes who don't care about how he got there...but he is wrong.

 

When I was doing this stuff for a job? If I'm off by that much? Fired. Period.

The race boils down to this: On Friday, Obama led in 19 battleground polls. Romney led in one.

Then...this race is not "close". The polls you've identified show that. But, we keep hearing from the Obama people, and the Romney people for that matter: that it is close.

 

If it is "close" we should be seeing ~10-9, not 19-1, right? 19 of something implies...weird results.

 

The fact that 19 polls show the same thing, in a race that is "close" tells me that those 19 polls all have the same issue, or that more than a few of them do.

 

Really, I'm basically done with these polls that continue to show Obama winning the top line, but losing with Is, enthusiasm, racial turnout and have a D+6 or more party ID, and every other internal. It's not even worth talking about. There's simply no way these things are "scientific". Something is F'ed here.

 

It's concerning, because we are supposed to be able to agree on the rules of the game...and then play it. Somebody is cheating. I can't say if they are doing it intentionally. But, regardless of who wins...a lot of people are going to have a lot of explaining to do after this election.

 

It may even cost some pollsters their jobs.

Edited November 3, 2012 by OCinBuffalo

[fjl2nd]

http://www.nytimes.com/interactive/2012/11/02/us/politics/paths-to-the-white-house.html

 

Really cool interactive about paths to the White House for both candidates.

 

Giving NC to Romney and NEV to Obama, it leaves us with: Obama has 103 ways to win, Romney with 24 ways.

[3rdnlng]

http://www.breitbart.com/Big-Government/2012/11/02/CNN-Ohio-Poll-Shows-Romney-Up-13-Among-Election-Day-Voters

 

 

"If you want to understand why voters no longer trust pollsters, look no further than the latest CNN poll of Ohio voters, showing President Barack Obama with a 50%-47% lead over Gov. Mitt Romney--a result that is within the poll’s 3.5% margin of error, but which suggests a slight Obama lead. The internal numbers reveal that Romney is leading among independents by 2%, and winning Election Day voters by a staggering 13%."

[/dev/null]

Here's a shocker poll

http://www.politico.com/blogs/burns-haberman/2012/11/aff-poll-minnesota-a-tossup-148296.html

 

Almost as shockering as two in the pink and one in the stink

[3rdnlng]

http://firstread.nbc...ney-in-fla?lite

 

 

 

"In these surveys, Democrats enjoy a nine-point party-identification advantage in Ohio and a two-point edge in Florida. Republicans have argued that a nine-point advantage is too large in this current political environment; it was eight points in the Buckeye State during Obama’s decisive 2008 victory."

 

 

Are they nucking futs?

Edited November 4, 2012 by 3rdnlng

[/dev/null]

http://triblive.com/politics/politicalheadlines/2878015-74/romney-percent-poll-state-obama-pennsylvania-president-lee-presidential-voters#axzz2BCil7kLj

 

http://www.myfoxdetroit.com/story/19996781/2012/11/04/romney-and-obama-in-dead-heat-in-latest-michigan-poll

[OCinBuffalo]

Here's a shocker poll

http://www.politico....sup-148296.html

 

Almost as shockering as two in the pink and one in the stink

Apparently Bolger is one of the few(3) pollsters in the entire country who picked Harry Reid to be re-elected by 5. I'd rather have that track record, than Nate Silver's.

 

It's like what I say in health care, so often that I'm bored with it: "everything you are doing in the middle, and everything you are doing at the high end, and moving data here and there...is worthless, because, after all these years and after all this money...you still don't collect your raw data properly, and your raw data is what everything else is based on."

 

From what I read, Bolger does raw data, properly, and has for years.

http://firstread.nbc...ney-in-fla?lite

 

"In these surveys, Democrats enjoy a nine-point party-identification advantage in Ohio and a two-point edge in Florida. Republicans have argued that a nine-point advantage is too large in this current political environment; it was eight points in the Buckeye State during Obama’s decisive 2008 victory."

 

Are they nucking futs?

Remember, this has been the CONSISTENT problem since I saw my first WTF? poll in August. It's not like this is some new thing

 

If I didn't see this consistency, I'd be much more worried. I'd rather have consistent, and wrong, because then I know what to expect.

 

It comes down to starting with one model of the electorate's demographics, or, another, and then the weighting methods used. Somebody is using the wrong one....and that's what I'm seeing on all the Sunday shows right now. I've been saying it for months.

 

(Some clown on ABC just stated it: "This is going to be 74, 73, 72 white voters"...and he just said "26% minority in 2008" . Both of those statements simply just aren't supported by the data we have, anywhere near as much as 75% white, 21% black+latino. I find it hysterical that the guy just said 74..and 72...as if just tossing around those #s doesn't mean massive swings, and, losses/wins of multiple states.

 

74...means Romney wins Ohio. 72 means Obama wins close elections all over the place. It's this cavalier approach that makes me think these people are merely repeating talking points, rather than actually knowing WTF they are talking about.)

 

This is the only rational explanation for why we see this divergence in the polls, rather than the convergence we've seen in every other election. One other rational explanation is: the state polls are being done by amateurs with tiny sample sizes, and the national are being done by pros. However, the other way is just as likely to be true: the state polls are being done by new people with new(and better) approaches, and the national are being done by old guys, with old methods that no longer apply.

 

There can be no doubt: The divergence alone tells you that something is wrong here....and somebody needs to start looking for a new job.

Edited November 4, 2012 by OCinBuffalo

[B-Man]

Main Street in Revolt

by Salena Zito

 

FTA:

Two years after suffering a historic shellacking in the 2010 midterm election, Democrats astonishingly have ignored Main Street Americans’ unhappiness.

That 2010 ejection from the U.S. House, and from state legislatures and governors’ offices across the country, didn’t happen inside the Washington Beltway world.

 

It didn’t reflect the Democrats’ or the media’s conventional wisdom or voter-turnout models. So it just wasn’t part of their reality.

 

In Democrats’ minds, it was never a question of “How did we lose Main Street?” Instead, it was the fault of the “tea party” or of crazy right-wing Republicans.

 

Yet in interview after interview — in Colorado, along Nebraska’s plains, in small Iowa towns or Wisconsin shops, outside closed Ohio steel plants and elsewhere — many Democrats have told me they are furious with the president. Not in a frothing-at-the-mouth or racist way, as many elites suggest. They just have legitimate concerns affecting their lives.

 

These Main Street Democrats in seven battleground states supported Obama in 2008. Now they are disappointed by his broken pledges: Where is the promised bipartisanship? How could health-care reform become such a mess? What direction is the country going in?

Their overriding sentiment is uncertainty over where the president is taking the country. They have no idea but get the feeling it isn’t the direction that traditional Democrats want.

 

They certainly haven’t gotten guidance from the president’s re-election slogans: class warfare, a hyphenated America, spreading the wealth around.

Over and over, these folks expressed unhappiness that fixing the economy doesn’t seem to be Obama’s focus; they have noticed that those in charge have high opinions of themselves but aren’t taking responsibility for the lack of progress.

 

It took Romney just 90 minutes, in a debate hall just a three-hour drive from that Leadville home’s sign, to convince many Americans (including many Democrats) that he passed their threshold test.

 

He came across as a qualified alternative to Obama who believes in their vision of an exceptional America and convinced them he can win.

 

And, just like that, “flyover” America was ready to vote its conscience.

 

What a shame that those from Kael’s “special world” don’t grasp the vicious cycle of their growing disdain for those alienated by their own actions.

They create dangerous narratives through Twitter and on TV that polarize and promote the rigidity of their ideology rather than introspection.

Never once have Main Street Americans heard Washington elites ponder, “What did we Democrats do to lose the confidence of so many voters?”

Plenty of traditional Democrats have voiced such concerns but are not being heard.

 

http://triblive.com/opinion/salena/2883641-74/democrats-main-street-america-obama-romney-americans-election-

 

.

[OCinBuffalo]

http://triblive.com/...icans-election-

 

This lady has been right a lot more than she has been wrong over the last few years.

 

However, Pittsburgh Democrats are the people she is talking about in every detail. I wonder if she's not in a bit of her own echo chamber there.

 

Of course, I like that she's talking about introspection. Perhaps the single biggest problem with Obama winning, if he does? It will prevent a whole lot of reasonable Democrats from doing the introspection they should have done after 2010.

 

No group of people, short of Scientologists, are in bigger need of introspection than Democrats today.

[3rdnlng]

http://www.nationalreview.com/articles/332386/parsing-polls-michael-g-franc#

 

 

"If anything, Gallup understates the case. In 2008, Democrats enjoyed a decisive ten-point advantage in partisan affiliation, 39 percent to 29 percent. When undecided voters were pushed to choose a party, the Democrats’ edge grew by another two points, to 54 percent to 42 percent. Yet in the Gallup polls conducted since October 1, the two parties have pulled even, with Republicans actually ahead by a statistically insignificant percentage point, 36 percent to 35 percent. After being pushed to choose a party, likely voters give the Republicans a further boost, resulting in an overall three-point advantage of 49 percent to 46 percent.

 

If you are keeping score, in slightly less than four years President Obama has presided over an eleven-point decrease in his party’s standing with the American people, 15 points if you include those voters who “lean” one way or the other."

 

 

[OCinBuffalo]

http://www.nationalr...ichael-g-franc#

 

 

"If anything, Gallup understates the case. In 2008, Democrats enjoyed a decisive ten-point advantage in partisan affiliation, 39 percent to 29 percent. When undecided voters were pushed to choose a party, the Democrats’ edge grew by another two points, to 54 percent to 42 percent. Yet in the Gallup polls conducted since October 1, the two parties have pulled even, with Republicans actually ahead by a statistically insignificant percentage point, 36 percent to 35 percent. After being pushed to choose a party, likely voters give the Republicans a further boost, resulting in an overall three-point advantage of 49 percent to 46 percent.

 

If you are keeping score, in slightly less than four years President Obama has presided over an eleven-point decrease in his party’s standing with the American people, 15 points if you include those voters who “lean” one way or the other."

Yep...this election is rapidly becoming more about Axelrod Vs. Gallup, and less about Obama Vs. Romney.

[/dev/null]

While on the subject of Democratic turnout coming up short of expectations, the Million Muppet March was far from a million

 

http://newsbusters.org/blogs/tom-blumer/2012/11/04/lame-million-puppet-march-draws-hundreds

[BRAWNDO]

OC very impressive analysis