Sam Spade's 2010 Predictions

[cinyc]

With Scott Brown's election, Republicans don't need to win 10 Senate seats to take control of the Senate.  They need 9 seats plus a Lieberman defection or 8 seats plus a Lieberman defection and a scared Democrat Senator defecting in order to save himself in 2012.    If Republicans win 9, Joe Lieberman becomes the most important man in the Senate.

The most likely outcome right now is that Republicans pick up 6, I think.

As for the potential 10th seat, Gillibrand is vulnerable if Pataki or someone credible decides to run.  Feingold tends to run in close races, and could lose to Thompson, if he decides to run.

[cinyc]

I am going to bury this here for the moment, in the absence of any better place. What does one think of Jay Cost's deus ex machina that one does a correction factor for all polls, so that the partisan turnout split for the given jurisdiction, matches whatever it was in 2004?  So if the split in a poll is more Dem say than in 2004 in California, you just readjust the partisan weights in the poll to a more GOP hue to get a more accurate picture of what would actually happen there was an actual vote.

2004 was a Presidential election year.  I have no clue why people are citing the 2004 numbers for an off-year election when partisan turnout will be higher at the expense of self-described independents.
 
In my opinion, given this will be a wave election, 1994 would be a more relevant measure if the overall partisan makeup if the states hadn't changed so much in the past 16 years.

[cinyc]

I am going to bury this here for the moment, in the absence of any better place. What does one think of Jay Cost's deus ex machina that one does a correction factor for all polls, so that the partisan turnout split for the given jurisdiction, matches whatever it was in 2004?  So if the split in a poll is more Dem say than in 2004 in California, you just readjust the partisan weights in the poll to a more GOP hue to get a more accurate picture of what would actually happen there was an actual vote.

2004 was a Presidential election year.  I have no clue why people are citing the 2004 numbers for an off-year election when partisan turnout will be higher at the expense of independents.  

In my opinion, given this will be a wave election, 1994 would be a more relevant measure if the overall partisan makeup if the states hadn't changed so much in the past 16 years.

Yes, quite right.  But there were no exit poll data released in 2002, and I can't find anything for 1998 either, so there you go.

There were no exit polls in 2002 after the 2000 exit poll debacle in Florida - except in California.

Exit polls from the 90s are still available on CNN's website:

http://www.cnn.com/ELECTION/1998/states/AL/#ppolls

(That's for Alabama - change state to whatever you want; click on the Governor or Senate office for the 1998 exit poll for that race - links to prior year exit polls are near the bottom of the state pages).  

U Conn's Roper Center has the raw exit poll data from 1976-2008:
http://www.ropercenter.uconn.edu/elections/common/state_exitpolls.html

I suppose using Presidential election year data is easier than off-year data because it's available for all states.  In the past two decades, only 2004, 1998 and 1992 would be an exact match to 2010's regularly-scheduled races - and all but 1998 were Presidential election years anyway.

[cinyc]

What do you think is the reason behind some of the strange latest Cook ratings changes? Oberstar being moved to Lean D? Carnahan being moved to Lean D? I suppose it has to do with the latest fundraising numbers but even then it seems like a stretch to say that Carnahan and Oberstar are in trouble based on how much CoH they have. They should be on a Watch List at best. The other changes that confused me were FL-12, PA-4, and NY-20.

Early voting in NV isn't looking bad for Reid.

Oberstar had a disastrous Duluth debate performance, where he called some of his constituents members of the flat earth society.   The sense is the race is tightening.  

There was an internal poll for Rep. Russ Carnahan's Republican opponent, showing Rep. Carnahan's lead shrunk to 9.

NY-20 seems to be tightening, with both the NRCC and DCCC throwing money at the race.  

I don't know about FL-12 or PA-04.  What did Cook do in PA-04?  I believe there was a recent independent poll showing Altmire with a big lead.

Last I heard, early voting in NV was looking terrible for Reid, when compared with 2008's early voting turnout.

[cinyc]

There was an internal poll for Rep. Russ Carnahan's Republican opponent, showing Rep. Carnahan's lead shrunk to 9.

Uh, that's the kind of poll result that you move a seat from Lean D to Safe D for. Not the other way around.

Not when everyone's expectation was that Carnahan would easily win reelection.  A single-digit lead shows uncertainty.

You must be logged in to read this quote.

I wouldn't give anything on such rumours (or more frequently "rumours".) That goes to the poster you replied to as well, of course.
[/quote]

These are not rumours, "rumours", rumors or even "rumors".  In the states that have party registration, the early/absentee vote composition is a fact - one that can be compared to the early/absentee vote composition from this time last cycle, or at the very least, at the end of the cycle or registration.  Politico did just that and says the early vote composition didn't look good for Harry Reid compared to party registration.  More Republicans and fewer Democrats have early voted as of the time the article was published.

[cinyc]

Early voting: http://www.accessclarkcounty.com/depts/election/english/Pages/home.aspx
New registrations: http://media.lasvegassun.com/media/pdfs/blogs/documents/2010/10/21/CCVP-2010-Q3_FINAL.pdf (showing great things for Democrats)

Showing great things for Democrats?  Hardly.  In 2008, the Clark County Early voters were 52.0% registered Democrats, 30.6% Republicans and 17.4% independents.  If my math is right, Obama ended up winning about 64% of the two-party Clark County early vote.

So far this cycle, Clark County Early voters are 46.3% Democrats, 38.0% Republicans and 15.6% independents.  Democrats are down 5.7 points from 2008, Republicans up 7.4 points and Independents down 1.8 points - a major swing away from the Democrats in the early voting composition.  And if my math is right, in 2008, Obama did about 9 points better in the Clark County Early Vote than the election day tally.  If a similar pattern persists in 2010, Harry Reid might not even win Clark County, let alone the rest of the state.

Edited to add:  This really belongs on the Early Voting thread, not cluttering Sam's thread.  I've copied it there.  Please continue the discussion on the Early Vote thread.

[cinyc]

Didn't the Dems do the same thing in 2006 and 2008? Did it help them bring in more seats?

Personally, I doubt that any amount of money could significantly shift the outcome on a macro scale in the last ten days. Early voting is already going on, and you can only polish crap candidates so much. It might swing a few races, but I'm really skeptical that these last-minute multi-million dollar expenditures actually do much other than firm up the races that are already shakily leaning towards them.

Remember - not every state has early voting, particularly in the Northeast.  It's more of a Western and Southern phenomenon.  Expenditures in areas without early voting could go further.  Voters in those areas don't need to make our minds up until election day, unless they vote for-cause-absentee.

[cinyc]

If you want to get a sense of the enormity of Sam's predictions - or just where the target seat are - the seats Republicans are seeking are in blue; Democratic targets are in red (HI-01 isn't on the map and LA-02 is barely visible):



The 10/25 calls for the GOP targets were:


(Dark Blue = Going R; Purple =Toss up; Dark Red= Watch List - others in between)

And the 10/25 calls for the DEM targets were:

(Dark Red = Going D, etc.) 

Purple HI-01 is off the map; LA-02 is barely visible.

[cinyc]

I see you've put up your final Congressional predictions.  Did you add any Congressional seats to the lists since 10/25?  PA-13 and MI-15 appear to be off the list.