Nate Silver's 538.com blog is depressing

[Pryderi]

http://fivethirtyeight.blogs.nytimes.com/

[tridim]

1. Problem is, he's using corporate funded poll results for his predictions.

Where the hell does he get his fudge factor? How does he know?

[rocktivity]

6. Yes--with the New York Times as his new "landlord"

I'm going to give him a lot more scrutiny than I have been.

:headbang:
rocktivity

[MellowDem]

7. He doesn't use partisan polls...

and of course any polling firm is a coroporation, but some poll for Dems as well and some would be described as "liberal". And he also includes polls from universities. Not only that, but his model has predicted elections very well in the past.

[tridim]

12. Has he predicted more than one election?

I thought he just started in 2008.

[MellowDem]

13. He did the elections in 2009...

but this will be his first mid-term election that I am aware of. It will be different, but his methodology hasn't been far off the mark yet. If it is, hopefully it is in the Democrat's favor.

[Codeine]

16. You're confusing "elections" with election years.

There are a shedload of elections each cycle, and he called almost all of them quite accurately.

[Mefistofeles]

18. Sabato has accurately predicted 4 elections

But I would not hesitate to bet that you don't trust Sabato either, because he is not telling you what you want to hear.

This is the thing: When others contradict our beliefs and hopes, we tend to throw them under the bus. There are many in denial about the prospects of the Democratic Party right now. But that's just my opinion. Let's see what happens in November.

[ProudToBeBlueInRhody]

24. Charlie Cook, too

All these guys had nothing but good things to say in 2008, and for some who were in the game at the time, 2006........it's funny how all the folks screaming that it's MSM disinformation had none of those issues two years ago.

Maybe, just maybe if anyone's paying attention, it will motivate someone to f'ing vote.

[scheming daemons]

2. Those numbers are better than a month ago.... moving in the right direction

We'll hold the house. The momentum is shifting.

[Nicholas D Wolfwood]

3. His House forecast is off, I believe.

Mostly because the algorithm he has relies heavily on the generic ballot polling, which stays in the system for a while. This means it includes a lot of the uglier generic ballot results we've seen, which puts a damper on better, more recent polls. That, plus the absence of heaving polling in most House races means that if any part of his forecast model (ie, House, Senate, Governor) would be wrong, it'd be the House forecast.

There are an awful lot of teabaggers running in House races - just like in the Senate, it will mitigate any gains they might have otherwise had.

[DaveinJapan]

19. Sorry, but what's "heaving polling" (I did try to look it up)? Tnx. nt

[Nicholas D Wolfwood]

20. A typo.

Should've read "heavy"

[hlthe2b]

4. Still a month away.... things can turn around

The governorships are concerning, but even if they stack the deck with teabaggers, they will QUICKLY learn that you can not run a state in the manner of their extreme rhetoric.

Time to knuckle down, though.

[nxylas]

21. Um, they *have* been running states like that

Texas, Arizona, South Carolina....

[hlthe2b]

23. I meant states with some history of actual governance.

[Drunken Irishman]

5. The problem I have with Nate Silver...

He was pretty much right on for the presidential election, but that was far easier to figure out than a midterm election.

No one really knows how a midterm election is going to turn out. You might have an idea with the way polls are trending, but it's not like a presidential election between essentially two men.

And it make sense if you think about it. A presidential election is a national race between just two candidates. It's not a hundred-plus races all over the country - many of which aren't being polled because the resources aren't there.

My district hasn't been polled in months (if at all since the primaries are over). It's just assumed our Democrat will win, even though he's in a conservative district. I wouldn't be shocked if he lost. Likewise, I'm guessing there are Republican districts that haven't been polled solely because they expect a certain outcome.

Then the polls that are done, often by local television and other local news agencies, carry a larger than normal margin of error. Sometimes Rasmussen polls a district, or CNN, but they only seem to be focusing on the contested districts.

The others? We're just guessing and speculating.

Another thing to realize is that in a presidential election, you pretty much have an idea of how most the country is going to vote. Specifically each state. We know, even before the election season begins, that the state of Utah will vote Republican 100% of the time.

It doesn't matter who the candidates are or what their platform is, the Republican will always win here.

But like I said, we have a Democratic congressman. He's in one of the most conservative districts in the country. So even here, we have Democrats.

Likewise, California ain't ever going Republican again - even though in 1989 it was considered a swing state. But it's solidified itself as a strong Democratic state and yet, they have a Republican governor. They have Republicans in congress. Up until recently, the Republicans were leading in the governor's race and in the senate race.

So it isn't as cut and dry as what you see with a presidential election. It's far easier to predict a presidential election outcome because you're only looking at most 50 different sets of data. And even then, it's not really 50 because you're already going to know about half those states will vote - in some instances, more.

In fact, a presidential election boils down to a few key swing states.

You focus on those states, predict the winner and you've just predicted who wins the presidential election.

But there are so many races out there that are underpolled, overpolled, overestimated and underestimated that there is a true unknown factor here.

How many people will turn out? This is an issue in presidential elections, but not nearly as much because we at least know a good number of people will turn out. In a midterm? It's a crapshoot.

In 1994, people thought Republicans would win - but they didn't know it would be as big as it turned out. That's a situation we could see in 2010. I don't know.

In 2002, the polls indicated Republicans might win, but it wasn't as conclusive or decisive. A few Democrats who were leading in the polls (Walter Mondale, Max Cleland) lost an it led to the Democrats losing seats - which bucked the historical trend.

In 2006, polls suggested Democrats would be in a pretty good position. But I don't know if anyone thought they'd clean up like they did.

In fact, it's possible we're underestimating the Republican support and they clean our clocks all around like in 1994. Then again, it's possible we underestimate the Democratic support and they get out there and keep both the House & Senate.

Right now, that's what it banks on - support for the Democrats. If Democrats GET OUT THE VOTE, we win. It's that simple. There are more Democrats than Republicans and the independent vote, while disappointed with the party, isn't strong enough to throw the election if most Democrats vote.

But will they?

That's the question. That's something Nate Silver can't answer. Because each midterm is totally different than the last. It's not a presidential election where you know there is going to be a good number of people voting.

We don't know what to expect for that and the reason I listed above.

[Mefistofeles]

15. What if I told you that Larry Sabato was extremely accurate in 2002 and 2006?

Would you believe his current 47 seat loss predictions for the Democratic Party?

[Drunken Irishman]

17. Sabato even admits this race could swing either way due to how close many races are...

That's the point. Many of these races are within the MOE. They are extremely close. Yes, they very well could break for the Republicans in the end, leading them to a landslide. However, I suspect we'll find a few surprises of our own in November.

I also don't think this is as similar as 2006 and 2002. In 2006, you had an extremely unpopular president. Pres. Obama isn't anywhere near the levels of Bush in that regard. He's closer to Reagan when Reagan suffered midterm losses in 1982.

In 2002, Pres. Bush had high approval ratings as we were only a year removed from 9/11.

This race is entirely different. Pres. Obama's approval is generally either just below 50% or just above it.

Generally that means the party in power either has manageable losses or wins seats. That doesn't appear to be the case right now. But that's what I'm saying...we just don't know what to expect.

[BzaDem]

22. I think you paint turnout models as a lot less certain than they are.

I mean, you are correct that no one knows for SURE who turns out, but that's not saying much. No one knows anything for SURE.

We do though have a very good idea using likely voter models. These models have been used in the past, they have been honed over the years and decades, and they have been pretty accurate.

This midterm election will be different than past ones. But so what? Every midterm is different than past ones. The point is, factors in the likely voter models DETECT the differences between this year and past ones.

The truth is that right now, if the election were held today, we would lose the House and come close to losing the Senate, because Republicans would turnout at a much higher rate than Democrats (according to likely voter models that are often accurate).

[Dawgs]

8. Don't tell anyone, but he's liberal.

Maybe he's trying to get the hopes up for the Republicans.

[The Northerner]

9. Considering how accurate is past predictions were

I'm feeling a bit worried but there are still about 6 weeks remaining.

[ClarkUSA]

10. 4.5 weeks is a lifetime in politics. Sign up to GOTV & turn your depression into a positive action.

Click on my sig if you want to help the Democratic party improve their chances.

If you want to, that is. It's your choice.

[Toucano]

11. I am beginning to think Dr. Dean is right.

We're not going to lose neither one.

The momentum is on our side, and the mainstream media story has been wrong for the entire election season.

[ProudToBeBlueInRhody]

25. I'd feel more confident if Dr. Dean was in charge

But no major political figure is going to kick dirt on their party's coffin before the election, even if they see the pallbearers are pulling it from the hearse.

[Arkana]

14. Look closer at the data for some of the House races--Nate hasn't updated that in a month or two.

[grantcart]

26. there is no real good polling about the House races

That would require multiple polling to establish trends in each of the 100 contested races.

My district is solidly Republican but the Democratic candidate is going after the incumbent (Bilbray) who is a terrible campaigner and has made some gaffes.

There is no independent polling here so we have no idea but it is possible that he is in trouble, just can't tell.

Also the Republicans and Tea Party has started way to early. The campaign is now just getting started and a lot of things are going to happen between now and election day - just ask Meg Whitman.