Here's a pretty strong rebuttal to Nate Silver's predictions by electoral-vote.com.

[craigmatic]

Statistician Nate Silver has built a statistical model of the 435 House races and by running simulations with the model, has determined that the Republicans have a 2-in-3 chance of getting a majority in the House. For people not familiar with stochastic modeling, a lot depends on the details of the model. Silver has included every parameter and bit of data he could get a hold of, which is generally a good thing, but the devil is in the details. For example, current polling shows that when asked: "Will you vote for the Democrat or the Republican?" (without naming the specific local candidates), the GOP is up by 8%. Such a factor can be included in the model. The hard part is figuring out how heavy to weigh it.
http://www.electoral-vote.com/

At least there's some hope. I didn't think the repubs could take the house. They'd need to run the table and not lose a seat but they'll fall a handful short.

[flamingdem]

1. Thanks for some good news nt

[craigmatic]

2. I just checked his model and he has us at exactly 218 seats not counting the toss ups.

[DrToast]

4. But EV shows us losing the Senate?

That seems hard to imagine we'd keep the House and lose the Senate.

[Mass]

13. EV results on their home page aer very problematic. They use the latest poll for a state .

So, it is subject to wild changes when a new series of polls from a different pollster comes out.

[jenmito]

16. They have DE as "Strong Republican," but it looks llike O'Donnell will beat Castle who will then

lose to the Dem. Coons.

[DrToast]

3. I don't generally make it a habit to bet against Nate, but...

There are a lot of good points in that article.

[DFLforever]

5. He was off in the MN governor's primary

After the polls had closed he was predicting Dayton losing by a couple of % points. Instead Mark won by 7,000 votes, enough to avoid a recount.

His MN model was imprecise.

And I don't see Bachmann winning by 16 points as he predicts. By 6 points maybe, but not 16.

[DemocratSinceBirth]

9. I Don't Think He Claims To Be Perfect

If he predicta 50.1%-49.9% outcome and its 49.9%-50.1% that just illustrates the limits of any model.

Also, I see, intrade and the Iowa political markets are "predicting" the Republicants have a 70% chance of capturing the House.

[Hansel]

23. I didn't know enough about Minnesota then to make his prediction

Dayton's running mate is very popular in the Duluth area which is a highly Democratic area of the state with traditionally strong voter turnouts. Their numbers come in much later after polls close. His main challenger was only slightly more popular in the Twin Cities than Dayton and was no where near far enough ahead of him to make up for the Duluth avalanche that was about to hit. If he had taken these factors into consideration, like our local political analysts did, he would have known that Dayton was going to win.

I wonder how many other of these nuances that Nate has missed. Sometimes relying more on statistics than your understanding of local politics can be a little tricky.

[flamingdem]

6. I'm betting against Nate this time nt

[Beetwasher]

18. No Need To Bet Against Him, Remember, His Analysis Is A Snapshot Of THIS MOMENT IN TIME

And at the time of his analysis, things were in fact looking pretty dismal. However, it will evolve and be modified as we progress towards the election. Watch him adjust it more favorably for Dems in the coming weeks.

[DrToast]

24. Good point. I agree.*

*

[angee_is_mad]

7. Kick

and keep kicking till November

[krispos42]

8. People usually think "their" congresscritter is okay

But that "somebody else's" congresscritter needs to go.

[FBaggins]

10. No so much a "rebuttal" as an explanation for why they differ.

Their explanation does as much to call their own ratings in to question as to "rebut" Silver's.

They point out that the generic polling is tough to weight for particular races... but Silver doesn't use it as the predominant part of his model... it's one of several factors. But that's not a big deal either way. What I do find interesting is that their model just automatically assigns a victory to any race where the incumbent last won with 55% or more of the vote... which is far more arbitrary than necessary.

Yes, Silver is almost certainly wrong about the LA-2nd (though that's really Cook's fault more than Silver's), but that's not nearly as wrong-headed as the notion that an incumbent can't lose if he last received 55% of the vote.

I'm also not sure that I'm anxious to jump on their bandwagon when they essentially predict a senate loss. I suppose it's possible, but I don't see any reasonable outcome that results in losing the senate but keeping the House.

[ClassWarrior]

11. There's only hope if we work our asses off. No need for "pretty strong rebuttals" to...

...let us off the hook. What's needed is a lot of volunteering for progressive Dem candidates. Period.

NGU.

[Ramulux]

12. Nate is good for the most part

But he is very off in a lot of house races. He allows polls and models to dictate results in districts where there are other obvious mitigating factors.

[Mass]

14. Not a rebuttal. He explains how these models work. Interesting to read, and of course,

nobody should confuse these models results with the election results. They only give an indication for where the polls trend at this point. Important to know, but not fixed.

[backscatter712]

15. Nate Silver's predictions are based on how the election would go, BASED ON TODAY'S POLLS.

So Nate's math says absolutely nothing about tomorrow's polls, of the events of the next few weeks.

In other words, anything can happen. And seeing that only a week ago, we've gotten past Labor Day, I'd say that the Democrats are only now warming up.

I predict that the GOP's leads, where they still have them, are going to erode.

Obama's very careful not to blow his wad too early, which meant he wasn't doing much campaigning in August.

[JNelson6563]

17. I think many will be proven wrong on election day.

And I will pointing and laughing at them.

Julie

[DrSteveB]

19. Do you agree with electoral-vote.com's prediction that Republicans will win control of the Senate?

Or do you disagree.

[jenmito]

20. I know this post isn't to me, but I disagree with them that Repubs. will take the Senate because

they have CASTLE being the Repub. going against (and beating) the Dem. Coons. But I think O'Donnell will win the Repub. primary and then lose to Coons, which will give the DEMS. the majority.

[craigmatic]

25. I disagree but at least the Electoral-vote.com's model changes frequently concerning the Senate

polls and it'll look much different in the coming weeks.

[harun]

21. They'll probably end up one short and then I'll say

[Bluenorthwest]

22. Nate is good at what he does, but what he does is limited

And in my opinion, not plugged into the real world in many major ways. While it profits the opining professionals to pretend the campaigns have been ongoing and we are now close to the day, the fact is the campaigns are just starting, and we are very far from the day.
Personally, I see the inclusion of 'generic data' to be an intentional tactic meant to avoid naming the Republican candidates. If you know both candidates, why ask generic?
The data that matters with Nate is as follows. You take all of his guesses, set apart the correct ones. Now note both the date of the correct guess in relation to the election in question, and note Nate's guesses previous to that day. By the time he's hitting the nail on the head, we all are, without his strained models and workings. Look at his rate of being right this far out. Not a month out. His data is already ancient, from the past, in a present that is moving very quickly.
The experts are often wrong, which is why there are experts instead of an expert. Which is why election nights are not relaxing evenings awaiting a foregone conclusion. Which they most certainly are not. Election night tells you what you really think about Nate and polls and all of that, which is it is balderdash, tea leaf reading. Bone throwing. All dressed up for the 21st.