Latest Generic Polls: Ras +12%R; WSJ 6%R; Gallup 15%R; CNN 10R; Fox13R; Bloom 3R

[The Vorlon]

Labour day is sort of the "tweener" time between when it is best to use RV or LV polls.

2 months is a long time and it's still a bit early to be using LV polls, on the other hand, the die is more or less cast for November, and the relative energies of the two parties will not change all that much.... so simple RV polls don't work perfectly either.

Come October, LV polls will clearly be the way to go so we just put up with things for a month where we are in the "tweener" time.

Looking at the last few polls when you group by methodology, polls within each group agree pretty darn closely actually...

LV Polls:

Rasmussen Reports   (LV)   Republicans +12
ABC News/Wash Post (LV)   Republicans +13
WSJ-NBC - (LV) - Republicans +9

RV polls that don't push the leaners too hard

Gallup   8/23 - 8/29   1540 RV   51   41   Republicans +10
FOX News   9/1 - 9/2   900 RV   46   37   Republicans +9

RV Polls

CNN/Opinion Research   9/1 - 9/2   936 RV   52   45   Republicans +7
USA Today/Gallup   8/27 - 8/30   928 RV   49   43   Republicans +6

The WSJ Poll just released, as always, is an excellent poll worth a read...

http://online.wsj.com/public/resources/documents/LateAugustWSJNBCpoll.pdf

[The Vorlon]

Another log on the fire...

Democracy Corps (D) says GOP are up 7 among LVs, 6 among RVs

http://www.democracycorps.com/wp-content/files/dcor090210fq8.pdf

[The Vorlon]

If you also take into account that Republicans almost always outperform the generic ballot...

How many points do you add to all of these ballot results to "adjust" for this phenomenon?

I can't find a link, but I remember Gallup's top pollster say that it was along the lines of +5-ish.

For what it is worth,

in the last 4 mid term elections the GOP has, on average, done about 6% better than the final Gallup Generic ballot result.*

The broad, virtually across the board GOP advantage in the "generic ballot" is pretty much unprecedented in polling history.

CNN has the GOP +7
Democracy Corps (D) has the GOP +7
ABC/Washington Post has the GOP +13

This is uncharted territory, there really are no models to project what this means in terms of seats.



Past performance is no guarantee of future returns, Your mileage may vary, see dealer for details.

[The Vorlon]

Gallup says the GOP is back to +5 on the Generic ballot, which is, more or less, the real number among RVs.

At least for the next twenty minutes or so, the polling all looks reasonably sane and consistent....

Likely Voters

ABC News/Wash Post:   Republicans +13
Democracy Corps (D):   Republicans +7
Rasmussen Reports:        Republicans +9

Average: - GOP + 9.7%

Registered Voters:

Quinnipiac: Republicans +5
CNN/Opinion Research:   Republicans +7
FOX News:   Republicans +9
Gallup:   Republicans +5

Average - GOP + 6.5%

Polls of similar populations more or less agree, and the gap of 3.2% between LVs and RVs is consistent with historical patterns.

Stay tuned for next weeks thrilling new episode.....

[The Vorlon]

Gallup had some really strong swings in the previous few weeks.

Republicans up 10 points, then tied!?!

Their current 5 point lead for Republicans seems a little more reasonable.

Gallup is a mess... the worst pollster. On this forum, the word gallup should be forbidden.

Gallup has a lot of very bright people working for them, and they have a great deal of integrity.

That being said, Gallup is trying to use 1970s methodology in 21st century situations - with predictable results.

The REAL reason Gallup is running a daily tracking poll is so they can build up a huge dataset and actually go in and truly "fix" they way they are doing things.

When a pollster truly and profoundly $ up they can react one of two ways - denial or increased determination.

In 2000 Rasmussen totally blew it (He has Bush winning by 10% in a 3000 person sample size) - Rasmussen went in and fixed things and has done quite well since then.

Let's hope Gallup can do the same.

[The Vorlon]

Gallup
9/13-19/10; 2,925 registered voters, 2% margin of error
Mode: Live telephone interviews

2010 Congress: Generic Ballot
46% Democrat, 45% Republican


Rasmussen
9/13-19/10; 3,500 likely voters, 2% margin of error
Mode: Automated phone

2010 Congress: Generic Ballot
48% Republican, 38% Democrat

Hmm, interesting contrast we have here.  Surprised to see the Dems leading on the Gallup ballot though.

Let's put "tied" into perspective.

In 1994 Gallup had the "generic ballot" tied, and we all know how that turned out.

The AVERAGE of the last 4 mid term elections, the Gallup RV generic ballot has under-polled the GOP by 7+ %

Tied in RVs is a big night for the GOP, anything above that is really putting a hurt on the Dems.

[The Vorlon]

The generic ballot has tightened considerably in the last month: http://www.pollster.com/polls/us/10-us-house-genballot.php

Interesting, though it was clear a GOP lead as high as it was in the late-summer was not going to last. Still, as Vorlon pointed out, if the Republicans are ahead, it's a bad night for Democrats.

Gallup will shift to a likely voter model next week, and assuming Gallup uses their standard model, current data suggest a gap of about 9 points between RV and LVs in the Gallup universe.

UPDATE:

Gallup's website suggest a 10%+ GOP lead using "likely Voters"

Additionally, preliminary modeling of the likely electorate using Gallup’s traditional likely voter questions (more on this next week) suggests that if current patterns persist, Republicans could have a double-digit lead in the national House vote on Election Day, which would translate into Republicans gaining well above the number of seats necessary to control the House.

http://2010central.gallup.com/2010/09/story-of-election-disproportionate.html

It is "highly" unlikely that kind of enthusiasm gap will last till election day, I would expect it to narrow to the more normal midterm  6% (ish) gap between RVs and LVs by election day

[The Vorlon]

Vorlon, you are calling for a sixty seat Republican gain in the House?

"calling for" would imply advocacy for that outcome, and I don't actually trust the GOP enough to want them to have that many seats.  

There are about 35 Dem seats that are basically "gone", baring some major shift the race is over.  There are another 35 or so where it is very very close, more or less toss-ups  I expect the GOP to win about 3/4s of them or so.  The Dems will likely win back LA2, HI1, and Delaware AL, ~~maybe~~ Illinois 10, so yes 60 seats +/- is the middle of the range.

There are dozens of seats, mainly in the Midwest, that are all very close, so it's really hard to get an exact read right now.  A shift or even 2 or 3 points Nationally changes a couple dozen seats in the house.  It's actually quite exciting.  Depending on the turnout model, GOP gains could be anywhere from 28 to about 105 seats at the extreme ends of the spectrum.

In politics, follow the money - The GOP is shifting money from the mid-west to the coasts.  You play offense in the other guys marginals, and if the marginals are New York, California, Washington, the GOP is looking at 60 (ish) seats.

[The Vorlon]

Vorlon, you are calling for a sixty seat Republican gain in the House?

"calling for" would imply advocacy for that outcome, and I don't actually trust the GOP enough to want them to have than many seats. 

There are about 35 Dem seats that are basically "gone", baring some major shift the race is over.  There are another 35 or so where it is very very close, more or less toss-ups  I expect the GOP to win about 3/4s of them or so.  The Dems will likely win back LA2, HI1, and Delaware AL, ~~maybe~~ Illinois 10, so yes 60 seats +/- is the middle of the range.

There are dozens of seats, mainly in the Midwest, that are all very close, so it's really hard to get an exact read right now.  A shift or even 2 or 3 points Nationally changes a couple dozen seats in the house.  It's actually quite exciting.  Depending on the turnout model, GOP gains could be anywhere from 28 to about 105 seats at the extreme ends of the spectrum.

Well, I agree with a 30-35 range of "gone" seats, but I'm a less certain about another 25-30 on top of that.  105 is possible if Obama nominates bin Laden for the Supreme Court.  That 60 seems excessive.

As I said, 105 is a really "extreme" scenario - maximum GOP turnout, very depressed Dem turnout, a "big" event worth a couple extra points, so "shock and awe" event.

In terms of counting 35 or so...

these I am pretty sure of, and then there are another 40 or so that are very much in play... in a wave year the wave party will get most of the marginals...

I am sure there are a bunch in the mid-west nobody is watching that will go GOP as well.

1 - TN6
2 - LA3
3 - AR2
4 - NY29
5 - MD1
6 - OH15
7 - NM2
8 - OH1
9 - KS3
10 - IN8
11 - CO4
12 - NH1
13 - NH2
14 - ND99
15 - TN8
16 - FL24
17 - MS1
18 - NV3
19 - VA5
20 - VA2
21 - FL8
22 - MI1
23 - NY24
24 - PA11
25 - WV1
26 - MI7
27 - WA3
28 - PA7
29 - IL14
30 - TX17
31 - AR1
32 - AL2
33 - IN9
34 - ID1
35 - SC5
36 - PA3

[The Vorlon]

You must be logged in to read this quote.

Hmmmmm.......

Gonna wanna look real close at these numbers.....

+18.......................... ?

The kindest thing I can think of saying is that this far out Gallup's likely voter model is... over sensitive.... and that it performs better as we get closer to the election....

(rubs eyes, checks his vision, takes drugs....)

[The Vorlon]

PSRA ‘jumps the shark’

In a recent poll for Newsweek, Princeton Survey Research Associates really outdoes itself.

Long know as one of the most left-leaning pollsters, in its 10/1/10 release, PSRA would have us believe that 96% of Democrat Registered Voters indicate that they are supporting or leaning to supporting the Democrat candidate for Congress, with only 2% supporting or leaning to support the Republican candidate!

In 2008 (according to Edison exit polls), 92% of Democrat voters supported the Democrat and 7% the Republican candidate.  That’s pretty much the same as 2006, when 93% supported the Democrat candidate (again, according to Edison).  In 2004, it was 90%.

Both for adults and registered voters, PSRA gives the Democrats an eight point lead, whereas Pollster’s aggregation of polls gives then a five point lead among adults, and a one point lead among ‘registered and likely voters.’

These ‘quirks’ in PSRA’s methodology probably explain, (at least in part) why it gives a generic Democrat advantage of 5 points while Opinion Dynamics and Rasmussen indicate a 6 point Republican advantage.

When checking RealClearPolitics generic ballot page, the last time any of the polls they use came up with a Democrat advantage of more than 2 points, was back in July!

PSRA....  almost as amusing as Zogby... not quite, but they are trying hard.

I got your email, I have made afew inquiries

[The Vorlon]

Yep, and now you know why I don't particularly like Gallup.  Since 2008 2000, they have been giving really strange numbers.

Fixed.

Actually in 2000, they were within 1-2 points.

At the end, yes. Their closing numbers are usually reasonable. But they had crazy numbers in October - like Gore up 12 points one day, and Bush +10 two days later.

Gallup have a very, very sensitive "likely voter" model.

It works pretty well with a week to go.

It doesn't work when it is more than a week to go.

We are more than a week to go => It does not work

QED

[The Vorlon]

Yeah but is there any reason why it would narrow significantly in the next few weeks?

The die is more or less cast.

I expect the RV to LV gap to tone down to about 7 or 8 points, a slight drift back, but not a sea change.

[The Vorlon]

New Gallup:

http://www.gallup.com/poll/143777/GOP-Holds-Solid-Leads-Voter-Preferences-Week.aspx

RV
GOP 48-43 (was GOP 47-44)

LV (1)
GOP 53-42 (was GOP 53-41)

LV (2)
GOP 56-39 (was GOP 56-39)

Gallup has shown pretty much the same result three weeks in a row now.

The Gallup turnout model/likely voter screen starts to stabilize more or less now, so I don't expect any huge swings from Gallup any more. (Next week's result? - Dem +13 or something....)

GOP moved two at the registered level, dems picked up 1 in the LV models....

[The Vorlon]

FWIW, I think pollsters are over estimating turnout.  I think the GOP is under-polling by about 3% across the board.

[The Vorlon]

The broad consensus seems to be a 5 or 6 point spread in favour of the GOP, from what I can see. That's discounting the more outlandish polling.

This is what is sitting in the current RCP average.

Rasmussen Reports: - Republicans +9
Gallup (LV Lower Turnout): - Republicans +14
Gallup (LV Higher Turnout)*: - Republicans +9
Newsweek   10/20 - 10/21   773 LV   45   48   Democrats +3
Politico/GWU/Battleground: - Republicans +5
Pew Research: - Republicans +10
Associated Press/GfK: - Republicans +7
FOX News: - Republicans +9

To quote that old Sesame Street song.. "One of these things, is not like the others...."

So let's just toss Mr. Newsweek into the bottom of the birdcage where it belongs....

Gallup's much maligned "low turnout" model is probably the outside worst case for the Dems, basically assumes no re-energizing of the Dem bass, which seems unlikely, although I would point out Rassmussen finds an 18% GOP lead among those most closely following the election which "kinda" validates the Gallup number as a worst case scenario.

Rasmussen Reports: - Republicans +9
Gallup (LV Lower Turnout): - Republicans +14
Gallup (LV Higher Turnout)*: - Republicans +9
Politico/GWU/Battleground: - Republicans +5
Pew Research: - Republicans +10
Associated Press/GfK: - Republicans +7
FOX News: - Republicans +9

Using the time honored tradition of tossing out the high and the low, and averaging the rest, this leaves us with:

PEW => GOP +10
Rasmussen, Gallup, Fox at GOP +9
Associated Press at +7

So GOP up upper single digits seems about right, assuming no "structural" flaw in all this polling, like assuming an improbably large turnouts in certain demographics based upon 2008 voting patterns......

[The Vorlon]

Gallup's low turnout model assumes a 43% turnout. So, that poll is either wrong because the turnout will be higher than 43%, or wrong even though the turnout assumption is correct, or it is right. Which is it, Vorlon?  I myself strongly doubt the turnout will be above 43%, and I suspect it will be lower.

I am on record as saying I think the GOP, broadly speaking, is being under polled by about 3% or so.

I's not so much the 43% or so I quibble with, it's the composition of that 43%.  Baking into the cake turnout demographics from 2008 seems incorrect to me in 2010.

I could be very wrong, we will know in 8 days. 

It will vary from race to race naturally, but I just don't think modeling in the 2008 turnout among young people, minorities, etc makes sense in any midterm, and certainly not 2010.

[The Vorlon]

Clearly Gallup was not baking into the cake turnout composition models given a given turnout percentage from 2008! I suspect it is  more like 2004, even though that was a presidential year.

let me rephrase, upon re-reading I don't think what I was trying to say got said correctly.

Most "likely" voter screens rely, at least in part, upon past voting behavior.

In 2008 Obama was able to bring substantial numbers of "occasional" voters to the polls, by contrast, many typically pretty reliable GOP voters stayed home.

The next effect of this is that I believe 2010 "likely" screens are being tainted by an unusual 2008.  The net effect is to bump some normally "likely" GOP voters out of the pool, and promote some atypical Dem voters into the pool.

For these reasons, I think a lot of "likely" screens are a tad off this year.

[The Vorlon]

Gallup's low turnout model assumes a 43% turnout. So, that poll is either wrong because the turnout will be higher than 43%, or wrong even though the turnout assumption is correct, or it is right. Which is it, Vorlon?  I myself strongly doubt the turnout will be above 43%, and I suspect it will be lower.

I am on record as saying I think the GOP, broadly speaking, is being under polled by about 3% or so.

So do you think we are likely to see the Republicans win the house with a 12 point margin? Then we would likely see the Republicans gain close to 70 seats and your prediction would be way off.

I haven't updated my numbers for quite a while. -

Things are very fluid  in the House, so I will post something late Monday as a "final" prediction. -

I am thinking 52ish in the house right now, There are about 20 races I want to have an actual look at before I come to a final number.

As I posted maybe 2 or 3 months ago "the House is gone" so I really have not been following things on a race by race level in the House.

I am also taking a very careful and hard look at the Washington State Senate race, it's very close, and I am going over the polls to try and figure out which way that one will go. - I am probably at 50+2/48 in the senate, but I want to double check Illinois, Colorado, and Washington - also, I hope we get a few more senate polls over the weekend.

[The Vorlon]

Gallup's low turnout model assumes a 43% turnout. So, that poll is either wrong because the turnout will be higher than 43%, or wrong even though the turnout assumption is correct, or it is right. Which is it, Vorlon?  I myself strongly doubt the turnout will be above 43%, and I suspect it will be lower.

I am on record as saying I think the GOP, broadly speaking, is being under polled by about 3% or so.

So do you think we are likely to see the Republicans win the house with a 12 point margin? Then we would likely see the Republicans gain close to 70 seats and your prediction would be way off.

I went through the House seat by seat, and I get to GOP +69 in the House actually, not sure I really believe it, I will look again after all the weakend polls drop.

The Senate is pretty stable, the House is collapsing for the Dems.

Fox is showing GOP +13 on the Generic now.

[The Vorlon]

And oh yes, you subtract GOP losses, be it 2 or 5 or whatever, from these totals.

The Dems win back Louisiana 2nd District - Cao beat "William Jefferson Cold Hard Cash" (as in $90K in the freezer) in 2008, and even then he only won by 3% - This is a almost a Dem lock to take back.

Delaware At Large, the seat of incoming Delaware Senator  tea-partied historic foot note Mike Castle will go back to the Dems, though it might be semi-close. (not as in gotta watch it, but a respectable loss)

Hawaii 1 is actually very close, I think Djou holds on given how good a GOP year it will be.

Kirks old seat in Illinois should be quite close - the polling has been so erratic as to be useless in that one, but Obama carried it by 23%, so it's pretty uphill for the GOP.

The Dems winning 3 seats seems ~~about~~ right.

[The Vorlon]

To put the same idea in a slightly different form, in 2008 the Dems won the House by about 11% in the total vote.

if we take the generic ballot as being +9 to the GOP side (this is an example, pick your own number to use) that is a shift in the general environment of about 20% or so.

if we assume (a huge assumption, but go with the flow here) that every seat shifts by about 20%, any democrat who won by less than 20% is, at least conceptually, vulnerable.

This is the list of seats the Dems won by under 20%

Ohio - 18th district
Iowa - 2d district
Pennsylvinia - 7th district
Kentucky - 3d district
Ohio - 10th district
Kansas - 3d district
Arizona - 1st district
New York - 1st district
Oregon - 5th district
Florida - 24th district

New York - 25th district
Pennsylvinia - 12th district
Conneticutt - 5th district
Illinois - 14th district
Pennsylvinia - 8th district
New Hampshire - 2d district
New York - 19th district
Georgia - 8th district
Iowa - 3d district
Texas - 23d district

Pennsylvinia - 10th district
Colorado - 4th district
Arizona - 8th district
New mexico - 2d district
Virginia - 11th district
Pennsylvinia - 4th district
New mexico - 1st district
North Carolina - 8th district
Mississippi - 1st district
Ohio - 16th district

California - 11th district
Michigan - 9th district
Arizona - 5th district
Maine - 1st district
Florida - 22d district
Wisconsin - 8th district
New York - 24th district
Texas - 17th district
New York - 29th district
New Hampshire - 1st district
Nevada - 3d district

Ohio - 1st district
Virginia - 2d district
New Jersey - 3d district
Florida - 8th district
Alabama - 5th district
Pennsylvinia - 11th district
Pennsylvinia - 3d district
Michigan - 7th district
Idaho - 1st district
Conneticutt - 4th district

Ohio - 15th district
Maryland - 1st district
Alabama - 2d district
Virginia - 5th district

Now obviously this is a very, very, rough and crude tool.

Incumbents retiring is an obvious adjustment.

Also somebody who won by say 25% against a sacrificial lamb might, at least conceptually, be vulnerable against a well funded opponent who ran an actual campaign....

If you're a Dem, I'd rather be in California than Ohio, and the list of adjustments goes on and on...

but at a macro level 50-70 is not an insane value....

[The Vorlon]

Gallup - GOP + 15
Fox - GOP + 13
Rasmussen - GOP + 12
CNN - GOP +10
PEW - GOP + 6
CBS NY Times - GOP + 6
McClatchy/Marist - GOP + 6
ABC - GOP + 4
Bloomburg - GOP + 3

The "Good Citizen" vote is breaking heavy to the GOP (except on the left Coast, looks like Murray "barely" survives)

[The Vorlon]

Please enlighten us as to what Obama could have and should have done to stop this.

A jobs/stimulus bill that focused on, well, jobs and stimulus instead of mostly being a massive payoff to core democratic voter groups.

Not wasting 18 months on a healthcare plan that they ended up stuffing down people's throats in some pretty undemocratic ways.

Making a few concessions to get at least a few token Republicans onside for major bills. (When you can't pick off any of Collins, Snowe, Graham, Lugar, Murkowski, etc... you know it's a pretty far out there bill)

The unemployment graph, you know, the one where unemployment never tops 8% if we pass the stimulus...?

Recovery Summer...?

[The Vorlon]

Not wasting 18 months on a healthcare plan that they ended up stuffing down people's throats in some pretty undemocratic ways.

The filibuster is pretty much the antithesis of a democratic procedure. That doesn't make it a bad thing, mind you, but if we had applied democratic principles, the health care plan would have passed easily and probably done more. Giving a reduced minority a veto that prevents the majority from debating an issue or voting on it is not democratic. Again, that is different from saying it's a bad thing or good thing; clearly everyone whose party is in the minority thinks it is awesome, while those in the majority fulminate against it. But you can hardly blame the large Democratic majority from using the rules of the Congress to accomplish what Republicans, using the rules of the Congress, strove to prevent.  

Who do you think is responsible for the health care bill taking about 12 months? Was it the Democrats 100%, 90%, 80%?

Being an advocate of limited government, I like the intent of the filibuster actually.  The filibuster just about forces everybody, including the majority party, to make some compromises.

In politics, intensity counts - the "pro life" movement is a great example - Polling fairly solidly shows that a majority of the nation is "pro choice" (in many of it's swishy forms) but the sheer intensity of the feelings on the "pro-life" side means they do win some concesions from the system.

Collins, Snowe, etc...  at least one or two of them could have been brought on board fairly easy and fairly early if the Democrats had been willing to make a few minor concessions.  The final Health Care bill was 2000 pages long, if you had let Collins write 4 pages of it, Snowe 6 pages, and Lindsay Graham another 5 pages it would have been 99% the same as the final product, and it would have passed in the fall of 2009.

The GOP, correctly it appears, felt the "issue" was better to have than the compromise in terms of the politics, which is why, even more so, the Dems should have compromised.  If you make the GOP an "offer they can't refuse" and they refuse it anyway, you win politically..

If Obama had legitimately engaged the moderates in the Senate early in 2009 they could have had a bill by fall.

Clinton did this amazingly well in 1995 - he said "yes" to the GOP welfare reform bill (or at least appeared to, he actually got a lot of changes to the final product) - What was the GOP to do after he said "yes" - stop supporting their own bill?

If Obama had taken the speeches of Collins, Snow, etc all, and said... "yup you've made 15 good points, we've accepted them all, the final bill answers your objections, let's vote.."  the thing would have been done in Oct 2009.



By the time it got to be 2010, certainly the GOP was playing politics, the longer they could make the DEms thrash about on Health Care the better the GOP arguments on  he economy started to work - A president has, maybe, a year to get things done, they need to do what it takes to get them done in that year.