Nate Silver is beginning to hedge.

[RandySF]

If you’re a regular reader of FiveThirtyEight you’ll know that our Senate forecast has said pretty much the same thing every day. When we officially launched our model in early September, it gave Republicans a 64 percent chance of winning a majority in the Senate. Today, the number is similar: 63 percent.

Republicans’ odds have never been higher than 66 percent — a figure they reached late last week — or lower than 53 percent. The informal model updates we published going as far back as March also had Republicans as 55 or 60 percent favorites.

To an extent, this stability reflects the noise-reducing features of the FiveThirtyEight model. Our program examines the polls for signs of statistical bias, and weighs them more heavily when they have larger sample sizes, better methodologies and better track records — which can reduce the impact of outliers. The FiveThirtyEight model is also fairly conservative in estimating the uncertainty associated with each race and the disposition of the Senate overall. At times in the past, the polls in most swing states have been biased in the same direction (either toward Democrats or Republicans).

But this degree of stability is unusual. In pretty much every election we’ve covered, the polls have more clearly broken toward one or another party by this point.

http://fivethirtyeight.com/datalab/senate-update-the-republican-advantage-is-consistent-but-not-decisive/

[unblock]

1. the main problem is that the races are not entirely independent

i don't think 538's model really factors in the possibility the some nationwide event could influence many elections at once.

if each race is independent, then it would be quite a feat for several elections to all go in the democrats' favor. but if they are correlated, it's wouldn't be nearly as much of a surprise, as a single event could sway several elections our way.

for this reason, i think the 63% number seems to overstate the odds of a republican win overall. they may have more than 50%, but their chance might be a lot closer to, say, 54% if the model took this into account.

[Cha]

2. "Too Close to Call" GOTV2014!!!!!!

"The bottom line to remember is this: No one – not Nate Silver, not Sam Wang and certainly not me – can tell you with any great confidence at this point who is going to control the Senate in three weeks. Given the statistical uncertainty in the polling data, the closeness of several races and the possibility that unforeseen events can shift a small but decisive number of voters’ decisions, it would be foolish to suggest otherwise."

http://www.usnews.com/opinion/blogs/opinion-blog/2014/10/10/neither-nate-silver-nor-sam-wang-can-be-sure-of-2014-senate-predictions

[Logical]

3. I trust him 100%. I think he only care about the math. Nothing else. nt

[randome]

4. Yes but...

...his strange spat with whatever-that-other-polling-site-is puts a chink in his armor for me. I admit I want him to be wrong this time around when he has been saying that the GOP will take the Senate.

But I'm starting to think some of his magic is revealing itself to be what magic always is -part illusion. He may be as flawed as anyone and I'm prepared for that even as I'm prepared for him to be right.

And for someone who buoyed our hopes last time, it's a little odd that he has chosen to commit more fully to something as useless as sports statistics.
[hr][font color="blue"][center]"Everybody is just on their feet screaming 'Kill Kill Kill'! This is hockey Conservative values!"[/center][/font][hr]

[Logical]

6. I also hope he is wrong this time. I still can't believe most this country is GOP. nt

[elias49]

8. I don't believe it is.

But the GOP votes.

[Ecumenist]

10. AND they suppress the vote as well....

ALL the more reason to make sure that we all vote and drag folks, KICKING AND SCREAMING if you have to, to the polls.

[Logical]

11. Good Point! Young Dems don't vote in high percents either. nt

[brooklynite]

14. Right now "whatever-that-other-polling-sit-is" says the same thing...

Bottom line is that the result is going to be close, and could go either way. All Wang and Silver are doing is applying some different principles in their models.

[Bluenorthwest]

7. His track record on midterms and Congressional races in general is not spectacular but it does make

up for that with brevity, as he's done so few of them thus far in his career. 2010, underestimated GOP wins in the House by 9 seats, he was wrong about 3 in the Senate out of 37 seats up. 2012 he was correct in 31 out of 33 Senate States, in both cases he predicted for Republicans and Democrats won.
So he's done two, and been off by 2 or 3 each time.

[Wounded Bear]

5. One other thing is the time zones...

A win in the east, say NC or GA, could influence late western votes in CO and AK.

I refuse to believe this is a slam dunk for the Pubbies.

[AtomicKitten]

9. Nate Silver is good at data-crunching.

In fact, he's second to none. But there are other factors in play such as the voter suppression laws that remain intact. Readers may recall the miraculous 2002 midterm election when several states exhibited mysterious double-digit swings in polling overnight that handed the Senate to the Rs. Rs are slippery bastards that will stop at nothing to win.

[KittyWampus]

12. "Rs are slippery bastards that will stop at nothing to win".

How true.

[brooklynite]

13. No, he's revising his model based on new data.

Unless you're arguing that Sam Wang is "hedging" as well...