LANDSLIDE? Hidden Vote?

**-Lazarus-** · 11-03-12, 09:22 AM

Oh. One more thing. You point to things like the RCP map with no toss ups like it's the gospel. Come on man. You are saying that a 1% polling margin in favor of Obama means he wins that state. Even I won't do that! Get real man.

**Fovezer** · 11-03-12, 10:06 AM

Originally Posted by -Lazarus-

I just did a long writeup on why I think Romney will win in a landslide. I've also said that Romney would win virtually every battleground/swing state, and win in a landslide. That's pretty damned specific. I also said that history would record that the turning point in the election for Romney would be the first debate. You responded by laughing at me, insulting me, and calling me names, as usual. I recorded that prediction before the debate even happened. The reversal in the poll numbers was quite dramatic, proving my prediction correct.

Even if you give ALL the battlegrounds to Romney, he still only tops out at 301 EVs. That's not a landslide unless he wins the popular vote by 7%+, and if he does that, he'll win more than 301 EVs. So what you are saying will happen doesn't match up with your prediction at all and you refuse to make any real predictions.

Originally Posted by -Lazarus-

Fov, you keep saying that if Romney wins the pollsters will all be wrong. The problem is that the only place where you will see the kind of numbers you keep pointing to is this 538 blog you keep touting. I've looked at them as well as every other source of information I could find, and frankly, at this point to me it looks like 538 is a liberal wish-tank. I like their format on their site but I kind of feel bad for them because their predictions are so far off base. It has to hurt their credibility after this election. At this point, anyone who says it will be a miracle if Romney wins like you keep saying is simply not paying attention to all the polls and other data - they have their head in the sand, as you clearly do.

Nate Silver has earned a ton of people's respect since 2008. He predicted every state except Indiana, and only missed it by 1%, and every Senate race. In 2010, he fairly accurately predicted the Republican takeover. He has very solid record and the people who dismiss him do it because they don't like what his model is showing. If you don't like that model, go look at Prof. Sam Wang's at the Princeton Election Consortium. He has a great track record, too. Predicted everything right in 2004 and missed by 1 EV in 2008.

Originally Posted by -Lazarus-

Now that said, maybe all the polls *are* wrong to a degree, some more than others. The polls are skewed in favor of Obama but not intentionally in some cases at least I think. There are two problems with most of the polls. The biggest problem with the polls in this election is that they assume voter turnout and enthusiasm will be the same for the Democrats as it was in 2008, and they adjust the data accordingly to adjust for expected turnout. The turnout and enthusiasm in 2008 for Democrats was much, much higher than it will be now. It will look more like 2010 did. There are certainly polls that support this. But most pollsters aren't accounting for that. The second problem which has happened more with liberal news organizations polling than anything, is an over sampling of one party over the other when polling. A lot of the polls (CBS, MSNBC, etc.) have sampled 1.3 Democrats for every Republican. This causes a skew toward Obama.

This is just parroting the stuff coming from poll truthers. The polls do not assume voter turnout or enthusiasm at all, what they do is measure their sample's likeliness of voting and how enthusiastic they are, but that's it. Everything else is conjecture and guessing on your part, including your assumption that 2012 will look like 2010. No presidential election year looks like a midterm year and more people vote in presidential elections than midterms, and that benefits Obama. The other point you bring up that poll truthers on the right like to complain about is that polls are "oversampling" Democrats. That's nonsense. Pollsters call a random sample of numbers and they let their sample self-report their party affiliation. They do not say "we want to have a sample of X% Democrats, X% Republicans, and X% Independents." Most polls do not weigh by party ID, either, it's just their for information. And even if you can make a valid case about a few polls "skewing" one way, that still doesn't change that ALL the polls are painting the same picture. So for Romney to win ALL the polls have to be wrong.

Originally Posted by -Lazarus-

Here's why I won't make a stupid electoral map for you, precisely predicting every state. It's 100% because you are such a dick. No other reason. If I made some map, when I am proven correct about a prediction of a landslide win for Romney but get a couple of states wrong (which I inevitably will), you will focus on that point and try to laugh and say how I was actually wrong. That is just how skewed your views are. And I won't give you the satisfaction. So sit on it and spin. I made my prediction. It's based on personal experience, election history, knowledge of human nature, and the polls. I made it before it was a safe bet. From the first moment after my prediction you have insulted me, derided me, called me names, questioned my intelligence, and his all around attacked me. I am doing this for fun during an election year. I am not a news commentator or paid prognosticator. I don't have to predict anything like you want me to and you don't have to like that. But that's the way it is. It's all recorded right here in there forums, and we'll see who was right.

So you don't want to make one because you're afraid your feeling will get hurt? You don't want to predict a Romney landslide in map form because you think I'll pick on you for getting one or two states wrong? If that happens, I'm not going to give a shit about one or two states. The real reason you won't make hard predictions is because you're afraid what will happen if you're wrong.

Originally Posted by -Lazarus-

Oh. One more thing. You point to things like the RCP map with no toss ups like it's the gospel. Come on man. You are saying that a 1% polling margin in favor of Obama means he wins that state. Even I won't do that! Get real man.

That's one example of many, leading by a 1% margin means he's the favorite, not that he's certain to win. But of all those states, only Colorado has a margin of 1% in his favor. All the other states are over that.

**AetheLove** · 11-03-12, 10:27 AM

Originally Posted by -Lazarus-

Fov, you keep saying that if Romney wins the pollsters will all be wrong. The problem is that the only place where you will see the kind of numbers you keep pointing to is this 538 blog you keep touting. I've looked at them as well as every other source of information I could find, and frankly, at this point to me it looks like 538 is a liberal wish-tank.

What is it about fivethirtyeight's methodology that you find most troubling?

Æ

**-Lazarus-** · 11-03-12, 10:36 AM

Originally Posted by Fovezer

This is just parroting the stuff coming from poll truthers.

Really? "poll truthers"? Do you even read what you wrote before you click "submit reply"? You keep saying "lol" and insulting everyone else and saying how conservatives are sounding desperate. Do you understand that it is your own posts which sound increasingly desperate?

I can't take anything you say seriously any more. I don't even know why I respond to you. Take your freaking blinders off.

**Fovezer** · 11-03-12, 10:55 AM

Originally Posted by -Lazarus-

Really? "poll truthers"? Do you even read what you wrote before you click "submit reply"? You keep saying "lol" and insulting everyone else and saying how conservatives are sounding desperate. Do you understand that it is your own posts which sound increasingly desperate?

I can't take anything you say seriously any more. I don't even know why I respond to you. Take your freaking blinders off.

Ah, I see you found an easy cop-out so you can ignore actually responding to anything I said. Poll truthers are people who ignore the polls or fault them for allowing their sample to self-ID as whatever they want and then claim they "oversampled" people.

Now for some comedy relief:
"Skewed" Presidential Polls - The Colbert Report - 2012-27-09 - Video Clip | Comedy Central

**-Lazarus-** · 11-03-12, 11:19 AM

Originally Posted by AetheLove

What is it about fivethirtyeight's methodology that you find most troubling?
Æ

Thanks for asking AE, it's refreshing not to see hyperbole in front of a question.

First of all the blog is liberal - Nate himself is a declared liberal. So calling the blog liberal biased isn't unfair.

There are several things I find troubling, but for me the most troubling is the way fivethirtyeight uses polls.

First, it gives much more weight than it should to polls that are still well within the margin of error. A 1% lead for Obama in a poll to Mr. Silver means a decided advantage for Obama.

Second, look at this snip from the blog from today:

LANDSLIDE? Hidden Vote?-fivethirtyeight-jpg

Silver also does things like calling Maine, Michigan and Minnesota (I highlighted them) swing states, and consistently using the polling advantage Obama has in those states to pad his battleground state assessment. This is just not sound reasoning and it pads his assessment of the election outlook overall.

Also, Mr. Silver cherry picks polling data - I think to support his liberal bias. Silver himself even said he was cherry picking to make one of his points for Obama in this recent article: Oct. 30: What State Polls Suggest About the National Popular Vote - NYTimes.com

Another quote from this article where Mr. Silver reveals how he gives too much weight to polls that are within the margin of error:

Mitt Romney and President Obama remain roughly tied in national polls, while state polls are suggestive of a lead for Mr. Obama in the Electoral College.

Most of the battleground state polls at that time were within the margin of error. But silver sees a 1 or 2% lead for Obama in a poll as statistically significant. This is troubling to me.

When I see things like I have mentioned here, knowing that Nate Silver is a declared liberal, it frankly causes me to question Mr. Silver's sincerity in his research.

**-Lazarus-** · 11-03-12, 11:22 AM

Originally Posted by Fovezer

Ah, I see you found an easy cop-out so you can ignore actually responding to anything I said. Poll truthers are people who ignore the polls or fault them for allowing their sample to self-ID as whatever they want and then claim they "oversampled" people.
Now for some comedy relief:
"Skewed" Presidential Polls - The Colbert Report - 2012-27-09 - Video Clip | Comedy Central

More Saul Alinsky. Sorry not playing your game. Facts are facts. And reality is a bitch.

**Fovezer** · 11-03-12, 11:53 AM

Originally Posted by -Lazarus-

More Saul Alinsky. Sorry not playing your game. Facts are facts. And reality is a bitch.

More comedic relief!

**WickedTribe** · 11-03-12, 12:09 PM

Originally Posted by -Lazarus-

Oh. One more thing. You point to things like the RCP map with no toss ups like it's the gospel. Come on man. You are saying that a 1% polling margin in favor of Obama means he wins that state. Even I won't do that! Get real man.

Romney was at +1% nationally when you were declaring a landslide for him. So yeah, you will do that.

**AetheLove** · 11-03-12, 12:52 PM

Originally Posted by -Lazarus-

First of all the blog is liberal - Nate himself is a declared liberal. So calling the blog liberal biased isn't unfair.

[...]

When I see things like I have mentioned here, knowing that Nate Silver is a declared liberal, it frankly causes me to question Mr. Silver's sincerity in his research.

That's not a criticism of the method, that's just saying you don't trust liberals (however you define liberal).

Not trusting the source is legit. There's plenty of sources I don't trust. I think your use of the label "liberal" as a scarlet letter is misguided, but that's how you judge trust and anyone who wants yours will have to clear the bar you set.

Originally Posted by -Lazarus-

Silver also does things like calling Maine, Michigan and Minnesota (I highlighted them) swing states, and consistently using the polling advantage Obama has in those states to pad his battleground state assessment. This is just not sound reasoning and it pads his assessment of the election outlook overall.

[...]

Another quote from this article where Mr. Silver reveals how he gives too much weight to polls that are within the margin of error:

Mitt Romney and President Obama remain roughly tied in national polls, while state polls are suggestive of a lead for Mr. Obama in the Electoral College.

Most of the battleground state polls at that time were within the margin of error. But silver sees a 1 or 2% lead for Obama in a poll as statistically significant. This is troubling to me.

I'm not sure I understand - you're not happy that Silver is saying that Michigan, Maine, and Minnesota are swing states? You'd rather he just assign their electoral college votes directly to Obama?

I can see that you understand the difference between the popular vote and the state-by-state voting (where electoral college votes get decided (except for two states which are not winner-take-all)); but then why do you care if these states are called "battleground" ? The label doesn't affect the data. Either a state can be predicted to vote a particular way or it can't. He picks out the states he thinks are close because he does further analysis in those states. For example, he tries to assess where the best bang for the ~~marketing~~ campaign dollars are.

A 1% or 2% lead can easily be statistically significant. It might also be too small to matter. The Law of Large Numbers and the Central Limit Theorem are real things, and learning how to sort out when you have a large enough sample to make finer-grained predictions is important if you want to do this sort of work. Having one poll show candidate A with a narrow margin might not mean much. Having 10 different polls all show a narrow margin to candidate A and only 1 poll showing a narrow margin to B means a lot more.

Originally Posted by -Lazarus-

Also, Mr. Silver cherry picks polling data - I think to support his liberal bias. Silver himself even said he was cherry picking to make one of his points for Obama in this recent article: Oct. 30: What State Polls Suggest About the National Popular Vote - NYTimes.com

I just clicked through to read that article. Did you?

Here's the sentence I think you're referring to:

Yes, I am deliberately cherry-picking a bit. But the discrepancy seems to hold if you look at the data in a more comprehensive way.

He is trying to find if the margin in the national polls and the apparent differential from 2008 can help him tease a bit more out of the current state polls. In particular, he wonders if Obama is underperforming in states we list as deeply blue. He's questioning Obama's lead in the states we all assume he's leading. He uses a few examples to illustrate the point; and he uses the examples that most clearly illustrate the differential.

That's what you want in an example - one that clearly shows the point.

But then he goes on to look at ALL the data.

This is how I read it:

"But what about the states we all think Obama will win going away? Is that so certain? For example, there was one poll out of Oregon in which he did poorly. Even if he wins the election, could he lose the popular vote?"

There are other ways to read it. I can't think of any way to read it where he's actually cherry-picking his data to skew his results.

Anyway, Nate Silver doesn't need me to defend him. Liberal or not, his future career as an electoral statistician depends on him doing a good job. Maybe he's trying to cash-in on his 15 minutes by purposefully singing a sweet song to Democratic voters in order to get a large, quiet, back-door Christmas present from The Liberal Conspiracy. Maybe it'll be 30 pieces of silver with a bow on it.

But I don't see it.

None of that says that I think he's "right". I think there is only one poll which really matters, and it happens on Tuesday. Playing pre-game odds maker is fun for the rest of us (and serious business for the campaigns).

Cheers,

AetheLove