23 November 2010

WASHINGTON STATE mid-term post-mortem: FINAL ANALYSIS

WASHINGTON STATE FINAL UPDATE, 11/23/2010. Once again, I am doing this for the sake of a continuing statistical analysis. As of 23 November 2010, 10:15 GMT+1:

With only 1,548 votes left to count (two weeks ago it was 369,102, on the Tuesday following it was 237,163, on Wednesday it was 173,878, on Thursday it was 122,453, on 11/13, it was 91,878 ), Murray's lead has increased from +4.30% to +4.75%, exactly as predicted.

Murray: 1,314,023 (52.37%)
Rossi: 1,194,926 (47.63%)
Subtotal: 2,508,949
Margin: Murray +119,097 votes, +4.75% (119,097 / 2,508,949)

Of the 1,548 votes left to count, which will make 2,510,497 total for WA for 2010, those 1,548 votes make up 6 100ths of a percent (0.06%), so, even assuming a Rossi 70-30 win over the remaining votes, the percentages should stay the same and the margin will be either +4.74% or +4.75%, depending on end-total rounding.

On 11/13, I wrote about King County:

"Murray has expanded her margin to +29.50%. Very likely she will edge up close to a +30% margin in King County."

Indeed, Murray has landed at +29.84% in King County (Seattle).

Of the "Murray" counties, here are the votes left to tally:

Grays Harbor: all votes COUNTED: Murray 51.73 / Rossi 48.27 - +3.46 margin) - -0.10 for Murray since 11/13

Jefferson: 20 (currently Murray 63.01 / Rossi 36.99 - +26.02 margin)- - UNCHANGED from 11/13.

King: ALL VOTES COUNTED: Murray 64.92 / Rossi 35.08 - +29.85 margin- +0.35% margin gain for Murray over 11/13.

Pacific: 50 (currently Murray 53.15 / Rossi 46.85 - +6.30 margin)- UNCHANGED from 11/13.

Snohomish: all votes COUNTED: Murray 51.86 / Rossi 48.14 - +3.72 margin)- -0.04 margin loss for Murray since 11/13.

Thurston: 165 (currently Murray 56.14 / Rossi 43.86 - +12.28 margin)- +0.14% margin increase for Murray from 11/13.

Whatcom: all votes COUNTED:Murray 52.49 / Rossi 47.51 - +5.09 margin, +0.04% margin gain for Murray over 11/13.

Total votes left to count from "Murray" counties: 335. So, of 1,548 votes left to count overall, the other 1,213 votes are from "Rossi" counties. The "Rossi" counties with the most votes left out are Stevens and Walla Walla counties, with 780 votes total left to count, and those are both counties with +30% margins for Rossi.

In Spokane county: all votes COUNTED: Rossi 56.27 / Murray 43.75 - +12.52 margin, a -0.08% margin drop for Rossi since 11/13.

The ratio of votes from "Murray" counties left to count versus "Rossi" counties left to count was 2.97 to 1 on 11/13, it is now 1 to 3.62. This is the first time in the counting where more "Rossi" votes are out than Murray votes, but since all remaining votes only account for 0.06% of the total vote, it now makes no difference.

Two weeks ago, I wrote:


"Therefore, the statistical probability is high that Murray's vote margin will definitely go over 100,000 and her percentage margin could climb up to just under +4.5, 0.5% higher than I posted a couple of months ago."

Thursday two weeks ago I wrote:

"She is already over +100,000 votes in margin and will surely stay there. +4.5 at the end of this all looks more and more likely."

On 11/13, I wrote:

"She is now at +4.3. +4.5 is all but guaranteed now."

She now stands at +4.75%. The race was not even close at the end of the day. Just to compare and to put this into perspective, Murray won her re-election with a higher margin than George W. Bush won with in Ohio, Colorado and Nevada in both 2000 and 2004. She won better than either presidential candidate in Pennsylvania, Iowa, or Wisconsin in either 2000 or 2004. Her win in WA in 2010 was more solid than Obama's win in Florida or Ohio in 2008. It was larger than McCain's win in Montana in 2008. It comes very close to McCain's winning margin in Georgia in 2008, a state that became competitive but was never really in doubt.

--------------------------------

Just to be absolutely clear about this point: this race in 2010 in WA for the US Senate is not nearly as close as the races from 2000 (Cantwell vs. Gorton: Cantwell +0.09%) or from 1988 (Gorton vs. Lowry: Gordon +2.17%) or from 1986 (Adams vs. Gorton: Adams +1.98%). But it will be the 4th or 5th closest race in WA for the US Senate since 1962 (19 elections). If Murray climbs over +4.79% in margin, then it will be the 5th leanest election since 1962. It looks certain that it will be the 4th leanest senatorial election in WA since 1962.


So, for those who follow my statistics for 2012, remember this analysis. I have been spot on the mark most of the time, excluding MO in 2008 and the senatorials in NV and CO in 2010.

10 November 2010

WASHINGTON STATE mid-term post-mortem: a continuing analysis

Note: there are earlier analyses of WA on my prediction thread at USELECTIONATLAS.ORG.

WASHINGTON STATE UPDATE. Once again, I am doing this for the sake of a continuing statistical analysis. As of 10 November 2010, 09:32 GMT+1:

With 173,878 votes left to count (on Sunday, it was 369,102, on Tuesday there were 237,163 votes left to count), Murray has currently increased her margin over Rossi to +4.28%

Murray: 1,215,467 (52.14%)
Rossi: 1,115,505 (47.86%)
Subtotal: 2,330,972
Margin: Murray +99,962 votes, +4.28%



Of the 173,878 votes left to count, 80,000 of them are still left from King County to count. On Tuesday morning, there were still 125,000 votes in King County to count, so 45,000 of the 63,285 votes tallied up yesterday indeed came out of King county, where Murray has expanded her margin to +29.24%. Very likely she will move to a +30% margin in King County.

Of the "Murray" counties, here are the votes left to tally:

Grays Harbor: 15 (currently Murray 51.78 / Rossi 48.22 - +3.56 margin)

Jefferson: 350 (currently Murray 62.95 / Rossi 37.05 - +25.90 margin)

King: 80,000 (currently Murray 64.62 / Rossi 35.38 - +29.24 margin)

Pacific: 50 (currently Murray 53.15 / Rossi 46.85 - +6.30 margin)

Snohomish: 21,000 (currently Murray 51.83 / Rossi 48.17 - +3.66 margin)

Thurston: 6,000 (currently Murray 56.13 / Rossi 43.87 - +12.66 margin)

Whatcom: 1,900 (currently Murray 52.35 / Rossi 47.65 - +4.70 margin)

Total votes left to count from "Murray" counties: 109,315. The other 64,533 votes are from "Rossi" counties. The "Rossi" county with the most votes left out is:

Spokane: 26,000 (currently Rossi 56.24 / 43.76 - +12.48 margin). Rossi's margin has remained static in Spokane county, it has increased by +0.09%, within the realm of statistical noise.

The ratio of votes from "Murray" counties left to count versus "Rossi" counties left to count is 1.69 to 1. Therefore, the statistical probability is high that Murray's vote margin will definitely go over 100,000 and her percentage margin could climb up to just under +4.5, 0.5% higher than I posted a couple of months ago.

Just to be absolutely clear about this point: this race in 2010 in WA for the US Senate is not nearly as close as the races from 2000 (Cantwell vs. Gorton: Cantwell +0.09%) or from 1988 (Gorton vs. Lowry: Gordon +2.17%) or from 1986 (Adams vs. Gorton: Adams +1.98%). But it will be the 4th or 5th closest race in WA for the US Senate since 1962 (19 elections). Please also notice that names tend to appear three times or so in WA elections, for instance Gorton, Lowry, etc.....rematches in WA are a pretty normal instance.

The fact that the end-statistics for this race are moving to the outer edge of battleground territory mrgins for elections supports the claim that the democratic party has built a massive firewall for 2012 on the west coast: WA, OR, CA, HI, NV.

A fascinating note about COLORADO

I want to point to one piece of information that I do not think that anyone else has yet brought up, and this holds ominous information for the GOP in the state of Colorado:

Here is the county electoral map of COLORADO from the GE 2008.

Here is the county electoral map of COLORADO from the 2010 Senate race.

Take a hard look. The maps are almost identical. Only Chaffee County, which was a bare McCain win in 2008, was a bare Bennett win in 2010.

This is important information, for it further supports some claims I made in my analysis of CO, for instance, here in Part III of the analysis:

"How to predict as 2012 nears? Here is a table of the closest county by county races (margin: under +4%), the so-called „tipping point“ counties (winning margin: less than 4%):

Alphabetically:


CountyRegionObama %McCain %MarginO-shiftM-shiftPartisan Shift
USA
--
52.88%
45.61%
+7.27%
+4.62%
-5.12%
+9.73%
Colorado
--
53.66%
44.71%
+8.95%
+6.64%
-6.99%
+13.62%
----
--
------
------
------
-----
-----
Grand
C-NW
48.59%
49.69%
+1.10%
+5.97%
-6.30%
-12.27%
Chaffee
C-WC
49.00
49.12%
+0.12%
+6.06%
-6.46%
-12.52%
Garfield
NW
49.20%
49.21%
+0.01%
+4.51%
-4.66%
-9.17%


Of the 3 „tipping-point“ counties, all 3 were barely won by McCain . McCain lost slightly more on percent than Obama gained in all 3 counties . The largest prize of these tipping-point counties is Garfield county, which was analyzed in Part II. Garfield County has moved from a 17% hispanic population in 2000 to a 24% hispanic population in 2007.
The direction of THESE counties in 2012 will give us a very good early indication of who will win the state: Obama or his GOP opponent. All eyes will be on the two huge pick-ups from 2008: Jefferson and Arahapoe counties, but polling in areas like Ouray and also in Garfield county will also give us a strong indication of who will win the state. The other important detail to watch will be the amount of money put into the Colorado Springs media-market. If Obama is in good condition and begins an agressive campaign in El Paso county, then the GOP will be forced to do the same. If it does not, then this means that the GOP already knows that Colorado is lost for them. If however, the hispanic community sours on Obama and the GOP goes on the offensive to reclaim the pick-up counties from 2008, then this means that the GOP has a good chance of re-taking the state."

-----------------------

It is interesting to note that Chaffee county, one of the tipping point counties from 2008, actually flipped to the DEMS in the year of a massive GOP wave. The county map of COLORADO is, as far as I can tell, the most consistent map vis-a-vis 2008 of any state of size. The county configuration of CO that has been leading to a democratic majority in the state can be traced back to it's real beginnings in 1988. Assuming the increased hispanic voter turnout that I am sure will happen in 2012, CO is looking more and more as if it will become a solid BLUE state in 2012 and beyond.

The fact that Bennett able to eek out a narrow win in the face of a massive GOP onslaught, but that the electoral map of CO barely budged, tells me that in 2012, the GOP will have an uphill battle in this state. For all of those young voters who did not show up in 2010 will be there in 2012.

And the hispanic population in Garfield County is expected to go over 30% in 2012. The statistical probability that Obama picks up all three of these tipping point counties and wins the state 56-43, is very, very high.

I will be posting regularly on the potential battleground states for 2012. This is the FIRST posting in such a series of postings.

05 November 2010

The Seattle Times calls the Senate Race for Murray

They have declared her the winner. Three time loser Rossi has already conceded. I think we should call him the William Jennings Bryan of the west...

Just yesterday, I made the following statistical analysis on my map thread to USELECTIONATLAS.ORG:

As the counting in WASHINGTON STATE continues, Patty Murray has increased her lead over Dino Rossi. Currently, as of 04 Nov 2010, 04:30 EDT:

Murray 50.84%
Rossi 49.16%
Margin: Murray +1.68%

This was expected as returns from King county have yet to come in, and she is heavily favored there.

Specifically, 71% of the vote is in, but most of the extremely red counties east of the Cascade mountains have reported more than the largest counties in the state. Here a good sampling:

Small red counties:
Lincoln County (deep red) 83% - 4,200 votes total
Columbia County (deep red) 75% - 2,000 votes total
Pend Oreill County (deep deep red) 92% - roughly 5,000 votes total.
Asotin County (deep red) 76% , 6,500 votes
Kittitas County (red) 85% - 12,500 votes
Kliktikat County (moderate red) 86% - 7,300 votes


larger RED counties:

there are other very small red counties with less then 71% of the votes counted, but they are very small counties. The exception is Benton County, with 40,000 cast. In Benton county, only 63% of the vote has been counted and Rossi is landsliding here with +29%.

Grant county (deep red): 79% - 20,000 votes total

the largest counties in WA:

Pierce County (tossup, red tilt) 93% - the second largest county in WA. Pierce county is where Murray is underperforming vis-a-vis Obama in 2008. Obama won Pierce county with +12. Currently 210,000 votes cast. Rossi is ahead here by 0.5%. But his lead is so narrow and there is only 7% of the county left to count. If this trend will continue to the end, then Rossi only adds 95 votes to his margin in this county. This will barely make a dent in Murray's votes still expected to come out of King county.

Snohomish County (moderate blue) 67% - third largest county. Murray is also underperforming here vis-a-vis Obama in 2008. She is winning with 51.5%, a +3 margin. Obama won here with 58%, a +18% margin. Currently 151,669 votes counted. But if she at least maintains here to the end, then she adds another 2,283 votes to her margin in Snohomish county, which will wipe out Rossi's edge in Clark county.

Clark County (moderate red) 85%, fourth largest county. Currently, 117,000 votes cast, Rossi is winning by +7.0. Obama won here by +6. If the trend continues to the end, then Rossi increases his margin in this county by 1,300 votes, a little more than half of Murray's expected margin gain (votewise) in Snohomish county.

King County (Seattle) 62% - the largest county in WA; more than 25% of the entire state vote will come from this county.

Murray is landsliding here with +25.2%, 62.6% to 37.4%. Till now, 418,937 votes have been counted in King county. 418,937 is 62% of 675,705. If Murray maintains this percentage and margin, then King county would look like this when all the votes are cast:

Murray 422,991
Rossi 252,714
Vote margin: 170,277

current (actual) vote margin in King County:

Murray +105,664

So, if conditions stay absolutely the same in King County, then Murray will add another 64,613 votes to her total. Obviously, Murray is currently also underperforming here. Obama won King county with almost 70%, a +42% margin over McCain. Those 64,000+ votes to pad her margin will more than wipe out any gains that Rossi will still get east of the Cascades.

I could extrapolate every county, but this would be quatsch. Without knowing exactly which precincts are out and their tendencies, it would be a shot in the dark. The big point is that a lot of King county is still out, and that will more than make up for the difference in red counties with votes yet to come in.


She may go as high as +2.5 at the end of the day. Wait and see.


-------------------------------------------------

And indeed, her margin as of today is: +2.5

SMILE.

When the dust has settled...

...from the mid-term elections, I will be doing an intensive analysis of:

1.) The Tea-Party candidates who won, and those who lost.

2.) The bad polling in a number of states and the worst perpetrators.

3.) Predictions on where the congress, now split, will go.

There were two outstanding articles about Obama, Capitalism and GM that came out. Hold onto your pants:

The Economist has published an APOLOGY to President Obama.

And this editorial pretty much says it all.

02 November 2010

Mid-term Election, final prognostications

Senate: When all the votes are counted: 51 D - 49 R.


On 10/29, I posted a trend-table reflecting the changes in polling averages for key senate seats between 10/14 and 10/29. Since RCP and pollster.com use slightly differing methodologies, I have taken the mean value (average) of those two and compared them.
Here is how things looked on 10/29:

State RCP 10/14 P.COM 10/14 Mean RCP 10/29 P.COM 10/29 Mean Diff Trend
NV
D +0.5
D +0.2
D +0.35
R +4.0
R +3.0
R +3.50
R +3.85
GOP
IL
D +0.5
R +0.7
R +0.10
R +2.8
R +1.2
R +2.00
R +1.80
GOP
WV
+/- 0.0
D +0.8
D +0.40
D +4.8
D +1.7
D +3.25
D +2.85
DEM
CO
R +3.0
R +3.9
R +3.45
R +1.6
R +1.2
R +1.40
R -2.05
DEM
WA
D +4.0
D +3.2
D +3.6
D +1.6
D +0.9
D +1.25
D -2.35
GOP









CA
D +3.7
D +4.4
D +4.05
D +6.5
D +4.5
D +5.50
D +1.45
DEM
KY
R +5.3
R +7.1
R +6.20
R +9.8
R +7.0
R +8.40
R +2.20
GOP
PA
R +7.5
R +6.2
R +6.85
R +4.0
R +3.1
R +3.55
R -3.30
DEM
WI
R +7.3
R +7.8
R +7.55
R +6.6
R +8.0
R +7.30
R -0.25
-static-
MO
R +9.4
R +9.2
R +9.3
R +10.4
R +7.6
R +9.00
R -0.30
-static-
CT
--
--
--
D +11.0
D +9.8
D +10.40
--
---
NC
R +15.7
R +13.1
R +14.40
R +11.8
R +13.8
R +12.80
R -1.60
DEM
OH
R +15.2
R +15.1
R +15.15
R +17.8
R +16.0
R +16.90
R +1.75
GOP
FL
R +16.2
R +14.1
R +15.15
R +13.3
R +13.7
R +13.50
R -1.65
IND
DE
D +18.3
D +19.1
D +18.70
D +15.8
D +14.4
D +15.10
D -3.60
GOP


And now, here is how things have changed from 10/29 to 11/02. I have adjusted the table to go in ascending order of mean value for 11/02, meaning from the statistically narrowest race (WA) to the widest blowout of the states analyzed (OH). A trend in BOLD means a shift in trend from one party to the other in the last 5 days. Note: A trend does not guarantee a win for the team that is trending. A trend is always in context of the margin itself. For instance, in DE, the mean is trending +1.10 in the direction of the GOP, but Coons is still stomping O'Donnell by a (perfectly synchronized, I might add) a +14.00 mean.

First, take a look at the numbers.


State RCP 10/29 P.COM 10/29 Mean RCP 11/02 P.COM 11/02 Mean Diff Trend
WA
D +1.6
D +0.9
D +1.25
D +0.3
D +0.7
D +0.50
-0.75
GOP
CO
R +1.6
R +1.2
R +1.40
R +3.0
R +1.6
R +2.40
+1.00
GOP
IL
R +2.8
R +1.2
R +2.00
R +3.3
R +1.7
R +2.50
+0.50
GOP
NV
R +4.0
R +3.0
R +3.50
R +2.7
R +3.0
R +2.85
-0.65
DEM
WV
D +4.8
D +1.7
D +3.25
D +4.5
D +3.0
D +3.75
+0.50
DEM
PA
R +4.0
R +3.1
R +3.55
R +4.5
R +4.0
R +4.25
+0.70
GOP
--
------
------
---
---
---
---
---
---
CA
D +6.5
D +4.5
D +5.00
D +6.5
D +6.0
D +5.75
+0.75
DEM
CT
D +11.0
D +9.8
D +10.40
D +8.7
D +6.3
D +7.50
-2.90
GOP
WI
R +6.6
R +8.0
R +7.30
R +7.7
R +7.8
R +7.75
+0.45
GOP
KY
R +9.8
R +7.0
R +8.40
R +11.0
R +8.4
R +9.70
+1.30
GOP
MO
R +10.4
R +7.6
R +9.00
R +10.4
R +9.2
R +9.90
+0.90
GOP
NC
R +11.8
R +13.8
R +12.80
R +12.8
R +12.5
R +12.65
-0.15
-static-
DE
D +15.8
D +14.4
D +15.10
D +14.0
D +14.0
D +14.00
-1.10
GOP
FL
R +13.3
R +13.7
R +13.50
R +17.0
R +14.5
R +15.75
+2.25
GOP
OH
R +17.8
R +16.0
R +16.90
R +18.8
R +17.8
R +18.30
+1.40
GOP


On paper, WA no appears to be the closest race out there, and indeed on election night, this could be so. But we may not know for up to 3 weeks. Earlyvoting.net has an outstanding small write-up on this:

I have been communicating with friends and colleagues in Washington about how quickly ballots arrive and are counted in the state.   The figures are fascinating, and if it is true that the Washington race may end up being crucial in determining control of the Senate, then the national media has better be prepared to wait for results.
Past history has been that 10% of WA voters return their vote by mail ballots immediately upon receipt. 30-40% of the ballots are returned by the Friday before the election (as of today, 38% of the ballots have been processed), and by Monday, approximately 50% of ballots will have been cast.
The remaining 50% come in Tuesday through Friday. And because Washington is a big military state, UOCAVA ballots arrive for weeks.
The 8pm returns from Washington will be meaningless.  Firm conclusions about the Washington Senate race will not be possible until Wednesday or Thursday at the earliest.  And it's quite possible that we'll have to wait the full 21 days until Washington certifies its results.“

In reality, however, CO may be the closest race going on. Not only that, CO is expected to have 70% voter turnout in this year, bursting all records from the past for mid-term elections. The GOTV in CO will make all the difference in the world.

What the table does confirm is the GOP wave making a roughly 1.20% expanse on the average across the board going into the finish. Look at OH, KY, MO, CO (not FL, it is hard to gauge a three-man race against a two man race). Also notice that the margin average in OH is only somewhat slightly ahead of the projected generic of GOP +15 that gallup produced.

For the full analysis, go to my last map at USELECTIONATLAS.ORG

31 October 2010

Clinton- Boccieri-Ohio-and babies!!!

This is a classic moment, a real keeper. I laughed so hard, my coffee came out through my nose!



CANTON, Ohio — Ohio Rep. John Boccieri ran offstage Saturday during a speech by Bill Clinton after the congressman received word that his pregnant wife had gone into labor.

Boccieri quickly left the podium Saturday afternoon while the former president addressed a crowd of about 1,000 people in Canton, about an hour south of Cleveland.

"The baby is now being born!" Clinton announced as the crowd erupted with cheers. "You'd be amazed how many times I take a picture with a very pregnant woman and then she immediately gives birth."

Boccieri, a Democrat, is locked in a closely watched race with Republican Jim Renacci. Boccieri's wife, Stacey, gave birth to their fifth child, Emma Nancy, on Saturday night.

Addressing both sets of grandparents, who were among the crowd standing in the parking lot, Clinton joked: "I'd like some credit for your fifth grandchild being brought into the world."

Clinton couldn't resist one more wisecrack before continuing his speech.

"We got another Democrat," he said. "I wish we could register that baby before it's too late."

Can pollsters be consistently off?

Answer: YES.

Rasmussen is an outstanding example thereof.

Back to statistics, and important ones at that. Here is a look back at 2008 in key states, with an eye especially on the verifiable inaccuracy of RASMUSSEN in battleground states or in states we knew would be DEM locks. For proof, here is the link to my "Poll Convergence No. 12" - my final poll convergence, just to refresh some memories and to prove that I am making none of this stuff up.

I all cases except where explicitly mentioned otherwise, I am referring to FINAL POLLS that were in the final calculations.
------------------------------------

CALIFORNIA, 2008: the three most accurate polls over Obama vs. McCain were FIELD, PPIC and SUSA:

Field, Final Poll:
Obama 55 / McCain 33: Margin, Obama +22

PPIC, Final Poll:
Obama 56 / McCain 33: Margin, Obama +23

SUSA, Final Poll:
Obama 60 / McCain 36: Margin, Obama +24

Actual Result:
Obama 60.94% / McCain 36.91%: Margin, Obama +24.03%

Please note that the Field poll understated Obama's win by 2 points, PPIC understated it by 1 point.

In all three cases, the polls were of REGISTERED VOTERS, not of LIKELY VOTERS.

Interestingly enough, this was the ONLY race of the night where RAS overpredicted Obama's win. The final RAS showed Obama with a +27 margin.

I tend therefore to think that Boxer will win with over 9 points. Even in a republican year like 2004, Kerry won CA with 10 points.

-------------------------

In WASHINGTON in 2008, SUSA, Strategic Vision (R) and Elway were the closest in predicting the race.

SUSA, final poll:

Obama 56 / McCain 40: Margin, Obama +16

SV, final poll:

Obama 55 / McCain 40: Margin, Obama +15

Elway:

Obama 55 / McCain 36: Margin, Obama +19

Actual Result:

Obama 57.34% / McCain 40.26%: Margin, Obama +17.08%

RAS missed the margin in WA by 6 points. RAS showed only an 11 point spread. By RAS's own definition of success (no more than one point in either direction), it wildly missed it's goal in WA.

The only pollsters in two months to show Rossi barely ahead in WA in 2010 have been RAS and (yes, you guessed it: FOX). If RAS has Rossi ahead by 1, then in reality, Murray is probably ahead by 3.5 to 4 and will land at between +3.5 and +4 once every ballot is counted.

-----------------------------

NEVADA, 2008:

The polling in NV was way OFF in 2008, mostly because the hispanic factor was incorrectly calculated.

So, here, first the actual results:

Obama 55.15% / McCain 42.65%: Margin, Obama +12.49%

In this case, AP-GFK nailed the results:

AP:

Obama 52 / McCain 40: Margin, Obama +12

In an alternate-reality twist of events, ZOGBY came the next closest, showing a +10.8 win for Obama:

ZOGBY:
Obama 53.2 / McCain 42.4: Margin, Obama +10.8

Suffolk showed it as a 10 point race: Obama 50 / McCain 40.

From that point on, everyone else was way, way off. Both PPP and RAS gave Obama only a +4 lead, meaning, both were off by 8.5% in Margin. CNN showed only a +7 lead. In fact, the polling was so off that the average I came up with was only Obama +6.83%, roughly half of the actual average.

CAVEAT: in this case, the AP-GFK poll dump of was of likely voters, and of the 8 states polled, they came the closest in CO, NV and VA. It is one of the only pollsters to do LV calculations and come close. It is more of the exception than the rule here...

Well, what to predict? 14% unemployment in NV, an absolute tea-party whacko on the GOP side, an unpopular Senator on the other side and way too much outside money flowing into the state. CNN, Mason-Dixon and RAS all show Angle at +4. I cannot ignore these stats. It will be interesting to see what PPP says in it's final poll. I may have to flip this state to the GOP. That being said, the early voting totals look very good for the democrats and if the hispanic element has truly been mispolled, then this race may be one of those "egg-on-my-face" races for a number of pollsters.

--------------------------

COLORADO, 2008:

President Obama won CO with +8.95% (53.66% to 44.71%).

CNN and AP came the closest: CNN showed the race at +8 for Obama, while AP showed the race at +9 for Obama. PPP overstated the margin by 1 point, showing Obama at +10. But the topline percentage for Obama by PPP was the most accurate.

The most inaccurate poll of CO was (yes, you guessed it): RASMUSSEN, which showed Obama at +4, less than half of the actual margin. National Journal also showed only a 4 point race (48-44). Marist showed it at +6, Mason Dixon had it at +5. Again, as in the case of NV, but not so extreme, the hispanic vote was not properly calculated into the polls for CO.

The trend here is very obviously to the DEMOCRATS, but it remains to be seen if it is enough. If RAS is as far off in CO as he was in 2008, then Bennet is sure to win. The final PPP and final CNN totals may be very enlightening.

--------------------------------

And just some more tidbits for you:

Obama won WISCONSIN in 2008 with +13.90%, the largest landslide for any presidential candidate in this state since 1964. RAS showed the race at +7 for Obama. Once again, right around half of the actual margin.

In this case, both Strategic Vision and National Journal pegged the race at +13 for Obama.

Likewise, in neighboring MICHIGAN, Obama won with 16.44%, the largest landslide for any candidate in this state since 1984. RAS showed the race at +10, 6 points off. But Selzer, who is from the region and used one of the larger samplings of any pollster, REGISTERED VOTERS, mind you, pegged his win at +16. Mitchell Research came in next at +14. EPIC/MRA (from MI), showed it at +12. Both PPP(D) and SV (R) showed the race at +13. However, the Big Ten 10 from 10 days earlier wildly overstated the race, showing Obama at +22.3.

Tracking after the election showed quite clearly that Detroit and Dearborn were understated in the statistics.

FLORIDA, 2008: Obama won FL by +2.81%. RAS showed the race at McCain +1. Of the last 21 polls of FL before election day, only RAS and DATAMAR showed a McCain lead. SUSA nailed the final margin at +3 for Obama. PPP and Qunnipiac were also very close, showing +2 for Obama.

NORTH CAROLINA is a fascinating case, as Obama won NC with his leanest margin of the night: +0.33%, so it would be very understandable were a pollster to err 1 percent on either side. RAS and SUSA erred 1 percent toward McCain. PPP and ARG showed +1 Obama. My average showed Obama +0.69%. This proves once again that averages do work. The differential between my average and the actual result was -0.33%, within the realm of statistical noise.

For the same reason, MISSOURI is just as fascinating, for it was McCain's leanest win of the night and it was the leanest win of all: McCain +0.13%.

Both RAS and SUSA showed the race as an absolute tie: RAS has it at 49-49 and SUSA had it at 48-48. PPP missed in in this case, showing Obama at +0.8, which means that PPP missed by 0.93%, slightly under 1%. Zogby was off, showing +1.7 for Obama, while Insider Advantage was off in McCain's direction, showing it at McCain +3. At the end of the day, my polling average was 0.46% for McCain. The differential between my average and the actual result was -0.33%, within the realm of statistical noise. PLEASE NOTE THAT BOTH DIFFERENTIALS - for NC and MO, respectively - ARE ABSOLUTELY IDENTICAL TO EACH OTHER, meaning that my average for these two states was 0.33% too generous to Obama in NC and 0.33% too generous to McCain in MO. Those are damned good statistics!

I called MO for Obama and it was the only state I mis-called in 2008, to note.

INDIANA is one of the most fascinating of all, for I indeed called IN for Obama on election night, in spite of a poll convergence average of +1.18 for McCain.

In IN, RAS showed the race at +3 for McCain. In reality, Obama won by +1.03%, so RAS was off by 4 points. The real weirdo in the statistic here was ZOGBY with his online junk: he showed a McCain win of +5.3 (I did a massive write up on this in 2008 and my 1/2 weightings of ZOGBY polls thereof).

PPP and Selzer (who knows the region better than any other pollster) both nailed it at Obama +1 and PPP had the slightly better topline % for Obama. But the fact that both Downs/SUSA and ARG both showed absolute ties in a core GOP state told me that the general direction was undoubtably for Obama. There were also some wild, wild outliers: the BIG TEN poll from 10/24 showed it at Obama +9.5, while Zogby from that very day had it at McCain +10.5. So, my poll average showed McCain at +1.18, but without ZOGBY in the mix, it was Obama +0.29. Then I looked hard at the Chicago media market, which feeds over into NW IN and I decided for Obama, correctly.

The point is, however, that RAS once again missed it, and in a core GOP state.

OHIO, 2008: Obama won OHIO with +4.58%, better than both of Bush 43's margins in 2000 and 2004.

RAS showed the race as an absolute tie: 49-49. Strategic Vision and Mason Dixon were at McCain +2. Zogby, SUSA and PPP showed it at Obama +2. The pollster who came the closest was CNN, which showed the race at 51-47, a 4 point spread for Obama. The CNN poll was not in my final calculations. The UCINCY poll from 11/01-02 was the next closest, showed Obama at +5.8.

ILLINOIS, 2008: Obama's win in his "home state" is more historic than most realize. With a whalloping +25.11% winning margin, his landslide in IL was the biggest since Calvin Coolidge in 1924 and the largest democratic landslide of all time in IL.

RAS's final poll showed Obama at +22, so RAS was only off by 3. However, the two previous RAS polls are very enlightening: the RAS from 10/15 showed Obama at "only" +17, and the RAS from 09/20 showed him at +16. Those are major discrepancies in a state where most other pollsters were showing Obama at +20 or more. The RAS shortly after the selection of Sarah Palin as McCain's running mate showed the race in IL at +13, practically half of the margin. But the greater sinner here was ARG, which actually showed this race at +6 for Obama on 09/17. This was never ever even once a single digit race in this state.

But the pattern of RAS is interesting. IN 2008, he consequentially and consistently deflated DEM numbers, inflated GOP numbers, especially in so-called BLUE states, and then, very close to the election, suddenly corrected his numbers come closer to the norm from everyone else in the known universe. IL is an outstanding example of this, which I why I cast serious doubt on RAS's number in IL for the senatorial in 2010.

A state that went for its favorite son by 25.11 points is not likely to elect a republican to his seat in 2010. There is, simply put, a DEM machine in IL that is too large and too well-oiled for this to happen, no matter how flawed the DEM candidate may be. And since the GOP candidate is equally flawed, the DEM still has the advantage. And apparently, this really is a 3 or 4 man race in IL, a fact that RAS loves to ignore. I am 100% sure that this will be a DEM win on Tuesday when all the votes are in, and I stand by my claim. Wait and see. I have the same feeling about this state as I had about INDIANA in 2008.

----------------------------------

In PENNYSLVANIA, President Obama won with a +10.31% margin, the largest margin for any candidate in this state since 1972.

PA was the most polled state in the Union in 2008, both in the primaries and then again in the GE.

Quinnipiac's final total nailed Obama's win at +10. Zogby's final was the next closest, at +9.8 for Obama. SUSA came it at +9 for Obama. PPP understated the win, had Obama at +8

The Muhlenberg final poll (Muhlenberg does daily 3-day running rollover totals, as it is also doing in 2010, AND RAS were both off by 4 points: they showed Obama at +6 and both had identical topline. Mason-Dixon was the most off in this state: it had the race at only +7 for Obama.

But the Muhlenberg is worth noticing, for it shows a sinus curve, just as it is now showing in the Toomey-Sestak race.

Here is the table from Pennsylvania from 2008, excluding all other pollsters except Muhlenberg. Take a look at the margins on the right and notice the up-and-down sinus curve nature of these roll-over polls:


Pollster
Date
Obama
McCain
Other
Und.
Margin
11/03
52
46
1
2
6
11/01
52
45
--
3
7
11/01
52
44
1
4
8
10/31
53
43
1
3
10
10/30
54
41
1
4
13
10/29
53
42
1
4
11
10/28
53
41
1
5
12
10/26
53
40
2
5
13
10/25
52
41
1
6
11
10/24
52
40
2
6
12
10/22
52
41
2
6
11
10/21
52
42
2
5
10
10/20
53
41
2
5
12
10/19
53
41
3
6
12
10/17
53
39
2
7
14
10/16
53
37
3
7
16
10/15
52
38
4
6
14
10/14
51
38
3
4
13
10/13
51
38
3
4
13
10/12
51
39
3
5
12
10/11
52
40
3
5
12
10/08
50
38
3
9
12
10/07
48
38
3
10
10
Muhlenberg
10/06
49
38
3
9
11
10/02
50
41
2
7
9
09/29
49
42
2
7
7
09/26
47
43
--
10
4

Had Muhlenberg done a run out complilation of it's end poll, showing just the last two days and then the last day, I bet the numbers would have gone: +7, +9.

And now, the Muhlenberg polls for PA-SEN, 2010, which you can see as a straight line here.

We see the same sinus curve, which we also expect for rollover polling. I do not know if that is enough to Sestak to win, but the race is clearly much closer than RAS has it: +4 for Toomey.
------------------------------------
I have been waiting until the midterms are over with and then I will be publishing a complete and exhaustive statistical analysis of the how the pollsters did in 2008 and in 2010, in regards to the final analysis. And I am sure that specific patterns will emerge, patterns that may dismay the one or the other side in all of this.

FUTUREWATCH:

To be brutally honest, we know live in an age of data information where a MOE of +/-3.5% is absolute nonsense! It is a copout.

With the kind of micro-data-management of which we are now capable, a MOE of +/-1% or less should now be possible. This however, requires a larger sampling and more consequent weighting, plus it requires us to consider how people are living these days: cell phone only users and the demographic they are associated with must now be considered. Although I think that ZOGBY with his online interactive polls is way off for now, I can clearly see that in 50 years, when more than 99% of the Union is online for practically everything, that online polling will become a science, and a very accurate one at that.


I am going to stick my neck out there and predict that RAS is going to have a lot of egg on his face once the totals from the 2010 Mid-Term elections are all in and finalized. And I think the Tea Party should be thanked for this, for their noisy participation in this cycle has more likely than not awakened a hell of a lot of DEMS who may have slept through this one but will now quietly got to the polls on Nov. 2nd. I am one of them, only I voted early...