Monday, June 30, 2008

Senate Projections are Live

As should be apparent from the new chart that appeared on the site in the last few moments, our Senate tracking is ready to go. We currently project the Democrats to gain 6.7 seats. However, the distribution is rather asymmetric: four current GOP-held seats (VA, NM, NH, CO) are exceptionally likely to go to the Democrats, two others (AK and MS-B) are more or less toss-ups (we project the Republican as a very slight favorite in both), and then there is a "long tail" of seats that the Democrats may have an outside shot at. Among Democratic-held seats, meanwhile, only Louisiana looks to be under any threat; we give the GOP about a 20 percent chance of winning it.

We give the Democrats about a 20 percent chance of reaching 60 seats counting Joe Lieberman and Bernie Sanders, and a 13 percent chance counting Sanders but not Lieberman. The Democrats are nearly certain to hold on to an outright majority of seats, whether or not Lieberman's seat is counted.

There is, obviously, a lot of explanation required for what goes into all of this, which will come in doses over the next day or two.

-- Nate Silver at 10:21 AM  

Labels: ,

42 comments

Anonymous said...

". . . whether or not Liberman's seat is counted.

NOTE TYPO ON LI(E)BMAN.

EXCELLENT NEWS TO SEE THESE NEW PROJECTIONS.

June 30, 2008 10:37 AM  
Mark said...

I agree with your view of North Carolina. While I think the race is certainly winnable for the Democrats, a lot of people have been over-estimating the Dems' chances of knocking off Sen. Dole. Winning that race is going to be very tough, particularly with a mostly-unknown State Senator.

Keep a special eye on Oregon and Maine, which I think will predict how well Democrats do in November. Both Sens. Collins and Smith are very popular, moderate Senators in blue states that will go heavily to Obama. If either of them loses, or perhaps both, it is going to be a Democratic Senate landslide that will wash away every vulnerable Republican incumbent.

June 30, 2008 10:40 AM  
Anonymous said...

Any efforts at reading tea leaves and seing what would happen if certain senators left the senate to join an Obama administration? just curious...

June 30, 2008 10:43 AM  
Anonymous said...

You're gonna start giving Cook and Rothenberg a run for their money!

June 30, 2008 10:46 AM  
ty said...

What's the basis for the projection of Idaho's Senate race at 82% Republican retention? I've not seen any polls that show this one at all close, though I'd definitely like it to be.

June 30, 2008 10:49 AM  
Juris said...

CLARIFICATION SUGGESTION:

Now that you have both Senate and Presidential poll detail, you might change the heading of the "Poll Detail" for the Presidential polls to read "Presidential Poll Detail."

Also, the ordering of the various tables and graphs on the right side is a bit confusing since it jumps from pres to senate to pres.

June 30, 2008 10:50 AM  
jsh1120 said...

Right on I think, Nate. From the antecdotal evidence from Alaska one picks up here in the Seattle area, I'm inclined to think that Stevens may well be defeated along with his colleague, Young, in the House. The stench of corruption in the GOP in Alaska is pervasive.

I know less about what is going on in Mississippi, but strong African American turnout for Obama might just add that seat to the Democratic side, as well.

I wish I were as optimistic about KY, OR, and MN. Obama might help in the latter two sates, but Smith's efforts in Oregon and Franken's inherent weakness as a candidate in MN make those states a reach. As for KY, I have no idea what's going on there. I'd only hope that the anti-incumbent, anti-Washington sentiment might tip the balance against McConnell.

June 30, 2008 10:55 AM  
Another Mike said...

I'm excited to see the Senate projections. Some questions that I hope you will be able to answer:

1. How do you do the regression for Senate races--what factors count and how much? same for every Senate race or different regression models in different races? to what extent is the model empirically based on past election results?

I thought I generally understood how you did the regression model for the primaries and less confidently how you do it for the general. But, it seems like the Senate races will require something completely different.

2. I understand the snapshot is the current status and the projection is the future prediction, but on what basis are you making it? It seems to be very different than what you're doing in the general as sometimes the frontrunner increases his lead. I'm sure you have some good common sense basis for it that is just eluding me.

I know you said details will come out in due course, so please do not consider this a nag post. Just wanted to hit some areas that I hope you will elucidate when the time comes. Much thanks again on the hard work.

June 30, 2008 11:02 AM  
Rhode Island X said...

Great to see this up! Comprehensive sources on the senate races are few and far between.

The Collins and Smith races, as Mark said, will be interesting to watch. I could see them falling similar to Chaffee (RI) to Whitehouse in the event of a solid Obama victory. He was the most liberal Republican--going as far as not voting for George W. Bush in 2004--but was rejected by RI voters with a strong anti-Republican wave.

June 30, 2008 11:14 AM  
Juris said...

OH-OH, something got scrunched at the top of the page (where you have the FAQ etc). I assume it will be fixed.

More generally, you may want to move to a different format, where you put ALL SENATE STUFF ON ANOTHER PAGE (also would give more advertising space), while on the FRONT page you would have only a Senate Synopsis with a LINK to the SENATE PAGE.

June 30, 2008 11:15 AM  
Juris said...

NOW that you have that second page, why not put the "Senate Polling Permathread Update" ALSO on the Senate page with the data (while still keeping something on the front page), so that there's always a story to accompany the charts?

Right now all I see on that page is a big McCain ad right in the middle of your Senate page.

I assume this is all transitional, and keep up the improvements! Just a quick reaction from me above.

June 30, 2008 11:19 AM  
Alexander said...

Great work and interesting results. Just as others have said, I'm very curious to find out more about the mechanics behind all this. But I'm sure that information will come in due time. For now, let me just say thank you for the work you've put into this.

June 30, 2008 11:21 AM  
Anonymous said...

Several suggestions for your model:
1. Take corruption into account, as well as how the presidential candidate is doing in the state right now (as opposed to in 2004.) I don't see why your regression model gives Stevens a solid +7.3 advantage, when Obama is 5 points behind in the state. See MT-Sen, 2006 for example. Tester was very liberal - but he still got elected.
2. Take name recognition into account. Many challengers should be expected to make up a lot of room as they increase their name recognition; you should try to include that into your projection (as opposed to your snapshot.)

June 30, 2008 11:24 AM  
Alex said...

You might want to put that first red bar on the Senate seat count at 51 instead of 50. The 51 seats gets party-lockstep votes and the organization resolution (which disburses committee chairs, etc) passed, while 60 is the number for cloture to break filibusters. Having 50 doesn't do too much good unless you count Lieberman (who may be on his way out of the party come January).

June 30, 2008 11:27 AM  
Juris said...

Maybe Nate was being optimistic in assuming that Obama would win and his VP would be the president of the Senate and break a tie?

But I agree with Alex that for now at least you might set the red mark at 51, not 50.

June 30, 2008 11:30 AM  
Juris said...

Nate, think of how you want your readers to operate. Now when I read your daily poll updates I usually scroll over to the charts on the right for further detail, and then back to the story.

I think the same logic should apply to the Senate races, and that's why I suggest that that second page be labeled "SENATE RACES" and that the daily or weekly stories related to the Senate be on the same page as the Senate Poll Data and all relevant charts, so that your readers can scroll over to check them as they read your stories.

Another way to do this is for you to in effect have a 'Features' header on page one, with links to latest stories on President and General 2008 Election, and to Senate stories. And anyone clicking on the Senate stories would then go to the Senate page and have all relevant charts accompanying any story.

Something like that. . .

Thanks for considering this. Right now, page 2 has mostly presidential stuff on it, and that's not what people might be expecting. Again, I know this is probably transitional, but I just wanted to suggest another way to do it.

June 30, 2008 11:39 AM  
Diggsb said...

Here's a crazy thought:

Could you add up-ticket coattails to your Senate regression model?

ie: Senate races would be affected by how well or poorly either Presidential candidate is doing at any given time.

This might require some delicacy, but could be interesting...

Now how about those House races, huh? Just kidding! Seriously though...

June 30, 2008 11:45 AM  
moondancer said...

While the numbers are disappointing, I still am hoping for the unseating of the two hypocrites, Smith and Coleman. I have faith in Minnesotans to send Coleman to K street where he really wants to be anyway.

June 30, 2008 11:50 AM  
Anonymous said...

many republicans are coming back in 2009

no '60' senate seats for democrats

hehe

June 30, 2008 12:05 PM  
Mark said...

Maine and Oregon are so interesting because the results in each will demonstrate just how far Democratic gains will go. Both Collins and Smith remain popular in their states, and the question is whether enough voters hate the GOP brand enough to vote for the Democratic challengers (Tom Allen and Steve Merkeley). The comparison to Lincoln Chafee is a good one: Chafee had an approval rating at 60 percent+, and still lost because the voters just could not pull the lever for a Republican. Given Maine's more moderate stanc