Initial Pollster Ratings: Winners and Losers

These are still based on preliminary results from this afternoon, so these could change, but the following is a review of 11 pollsters who had at least 5 state polls in the final 2 weeks of the election:

Average Miss:

  1. Grove Insights: 1.4%
  2. SurveyUSA: 1.8%
  3. YouGov: 2%
  4. Gravis: 2.3%
  5. Pulse Opinion Research: 2.48%
  6. We Ask America: 2.5%
  7. NBC/Marist: 2.7%
  8. PPP: 3.1%
  9. Rasmussen: 3.9%
  10. American Research Group: 4.3%
  11. Mason-Dixon: 4.9%

Partisan Index:

  1. We Ask America: Obama+0.2%
  2. SurveyUSA: Obama+0.3%
  3. Grove Insights: R+0.8%
  4. PPP: R+0.8%
  5. Mason-Dixon: R+0.9%
  6. YouGov: R+1.2%
  7. NBC/Marist: R+1.4%
  8. Gravis: R+2.1%
  9. Pulse Opinion Research: R+2.4%
  10. Rasmussen: R+3.1%
  11. American Resarch Group: R+4.2%

Pollster Score (Partisan Index + Avg. Miss)

  1. SurveyUSA: 2.1
  2. Grove Insight: 2.2
  3. We Ask America: 2.7
  4. YouGov: 3.2
  5. PPP: 3.9
  6. NBC/Marist: 4.1
  7. Gravis: 4.4
  8. Pulse Opinion Research: 4.9
  9. Mason-Dixon: 5.8
  10. Rasmussen: 7
  11. American Research Group: 8.5

So, who won and who lost? I think the three clear winners are SurveyUSA, who was 2nd in both partisan lean and average miss; Grove Insight, who had the lowest average miss and was 3rd in partisan lean; and YouGov, who has demonstrated that online polling can be equal to, if not superior to traditional phone polling.

The big losers are Rasmussen and ARG, who both missed bit AND missed big to Romney’s side, and Mason-Dixon who missed by the widest average margin of any poll, though at least they missed to both sides.

PPP was only mediocre in state polling, despite his success at the national head-to-head, coming in 8th in average accuracy and tied at 3rd in partisan lean.

Gravis also had a respectable showing after many, including myself, criticized them for being well off in their polling to start. It seems they finally got their feet under them to a reasonable extent by the end. They came in 4th in accuracy, though they ranked 8th in partisan lean.

Projection and Polling Roundup

If you missed the flurry of posts last night and this morning, you can find them all again here.

First, the big one, the Electoral College Projection. I had two projection models, both ended up coming out to Obama 303 Romney 235.

I also did some Senate projections. It came out to be a 54-46 democratic senate, though with some heavy caveats.

Of course, there is the November 5th polling report.

The final 2-week national average

And the final State of the Race report.

Also, some oddities in this election that could happen:

  1. If Obama wins re-election, it will be the first time we have back-to-back-to-back two term presidents (Clinton, George W. Bush, Obama) since Thomas Jefferson, James Madison, and James Monroe. Yes, it’s really been that long.
  2. If Obama wins, it will likely be the first time an incumbent won re-election with a lower share of the popular vote than in his first election since Andrew Jackson in 1832 (not counting FDR’s 3rd and 4th election)
  3. If Obama wins, it will likely be the first time an incumbent won re-election with a smaller number of electoral votes than his original victory since Woodrow Wilson in 1916 (not counting FDR’s 3rd and 4th election)
  4. If Romney wins, it will be the first time a sitting member of the House of Representatives (Paul Ryan) will be elected President or Vice President since Speaker of the House John Nance Garner was elected as Franklin D. Roosevelt’s first Vice President in 1932.
  5. If Romney wins, he would be the first President from Massachusetts since John F. Kennedy in 1956 and the first former governor of Massachusetts to be elected president since Calvin Coolidge in 1924
  6. Romney is potentially facing the largest home-state loss ever. The largest home-state loss in a two-man race by a major party candidate who didn’t share his home state with the winning candidate was Adlai Stevenson losing Illinois to President Eisenhower by 19.2% in 1956.

Two-Week Average for October 23 – November 5: Obama 47.9%, Romney 47.1%

The last two week average was an exact tie, so it was important to see if anyone got momentum in the final two weeks. It looks like Obama got that momentum, although somewhat slightly, building up a 9/10ths of a percent lead on Governor Romney on the eve of the election:

Obama: 47.9% (+0.7%)
Romney: 47.1% (-0.1%)

Margin: O+0.8% (O+0.8%)

This average included 22 national polls:

  • Associated Press/Gfk from October 19-23 (Romney 47, Obama 45)
  • National Public Radio from October 23-25 (Romney 48, Obama 47)
  • CBS/New York Times from Oct. 25-28 (Obama 48, Romney 47)
  • National Journal from Oct. 25-28 (Obama 50, Romney 45)
  • High Point University from Oct. 22-30 (Obama 46, Romney 43)
  • Fox News from Oct. 28-30 (Obama 46, Romney 46)
  • Washington Times/Zogby from Oct. 29-31 (Obama 49, Romney 49)
  • Purple Strategies from Oct. 31 – Nov. 1 (Obama 47, Romney 46)
  • Pew Research from. Oct. 31 – Nov. 3 (Obama 48, Romney 45)
  • NBC/Wall Street Journal from Nov.1-3 (Obama 48, Romney 47)
  • ABC News/Washington Post from Nov. 1-4 (Obama 50, Romney 47)
  • Democracy Corps from Nov. 1-4 (Obama 49, Romney 45)
  • Gallup from Nov. 1-4 (Romney 49, Obama 48)
  • Monmouth University from Nov. 1-4 (Obama 48, Romney 48)
  • PPP/Daily Kos/SEIU from Nov. 1-4 (Obama 50, Romney 48)
  • American Research Group from Nov. 2-4 (Obama 49, Romney 49)
  • CNN/ORC from Nov. 2-4 (Obama 49, Romney 49)
  • Rasmussen from Nov. 2-4 (Romney 49, Obama 48)
  • UPI/CVoter from Nov. 3-5 (Obama 49, Romney 48)
  • Gravis Marketing from Nov. 3-5 (Obama 48, Romney 48)
  • IBD/TIPP from Nov. 3-5 (Obama 50, Romney 49)
  • Politico/GWU/Battleground from Nov. 4-5 (Obama 47, Romney 47)

The closeness of the national popular vote can be see via the 7 ties among the 22 polls conducted. However, Obama appears to have the advantage by virtue of leading in 11 polls versus Romney’s 4, and two of the four Romney polls are the two oldest polls in the list.

Obama certainly has some worry signs as the average still has him only around 48%, but the fact that the trend appears to be in his direction goes in his favor. His average is slightly higher than it was during the September 25th to October 8th period, which split between pre-first debate and post-first debate. It’s also 1.3% higher than it was through most of the summer.

Romney’s average is the same as last week, demonstrating that while he hasn’t lost any support in the past week, he also hasn’t gained any ground either.

To compare against previous elections, the final average in 2000 was Bush leading by about 2.5%, 46.2% to 43.6%. The actual election results were about 3% off, but there were  also a full 10% undecided even going into election day in 2000, which is twice the number as today.

2004 saw Bush leading Kerry by 1.2% in the polls, 48.9% to 47.1%, and ultimately winning by 2.4% on election day. Bush gained 1.8% over his final 2-week average while Kerry gained 1.2%. Unlike this time, it was Kerry who appeared to have the momentum in 2004 while Bush was stalled, but Bush pretty evenly split the final undecided vote pretty evenly nevertheless.

2008 saw Obama leading by 7.2% in the final average and winning by the same 7.2% margin, with both candidates gaining 1.9% on election day over their final polling average.

In all three cases – 2000, 2004, and 2008 – the incumbent party improved on it’s final polling average in the final election day results, so there should be no surprise if Obama does the same. The question is how much do those votes break. Assuming 1% vote for third parties, Romney needs late deciders to break about 60-40 or so his way to win the popular vote based on this average. That ratio will go up if the 3rd party share is higher than 1%.

Final Electoral College Projections

And here we are: the final electoral college projections. It’s been an interesting seven months tracking polls and we’re finally here at the end, technically on election day, with some people in New Hampshire having already cast their votes.

There will be two final projections: the first is based straight off the average and the other is a modified 2-to-1 scenario. In that scenario, I take the number of unallocated voters – that is – voters that don’t say they are voting for Obama or Romney. I then subtract 1% from that total to account for third party voting. I then split up the remaining vote 2/3 for Romney and 1/3 for Obama in accordance to the so-called “incumbent rule.”

Before I get to each state projections, here are the electoral vote projections for each of the two projections I am doing:

Straight Average Projection:

Obama: 303, Romney: 235

Adjusted 2-to-1 Scenario:

Obama: 303, Romney: 235

As it happens, both scenarios actually give the same electoral vote count, albeit the margins in some of the states are closer in the second scenario.

*states with an asterisk have had no 2012 polls conducted  so projections are based off of 2008 results

#One of Maine and Nebraska’s electoral votes have been split out and projected on their own.

The following are “lean” states, which are states that are under 5% in both projections:

Lean Obama (87 Electoral Votes)

State Average Margin 2-to-1 Margin EVs
Colorado O+1.9% O+0.6% 9
Iowa O+2.6% O+1.1% 6
Maine-CD2 O+4% O+3.3% 1
Nevada O+3.6% O+2.6% 6
New Hampshire O+2.5% O+1.4% 4
Ohio O+2.6% O+1.4% 18
Pennsylvania O+4.1% O+2.6% 20
Virginia O+1.2% O+0.04% 13
Wisconsin O+4.2% O+3.1% 10

Lean Romney (44 Electoral Votes)

State Average Margin 2-to-1 Margin EVs
Florida R+0.8% R+1.6% 29
North Carolina R+2.4% R+3.4% 15

Next are “weak” states, which are states that are at or over 5% in at least one of the two projections but under 10% in both projections:

Weak Obama (38 Electoral Votes)

State Average Margin 2-to-1 Margin EVs
Michigan O+5.7% O+4.4% 16
Minnesota O+6.7% O+5.1% 10
New Mexico O+8.3% O+5.8% 5
Oregon O+6.1% O+3.7% 7

Weak Romney (40 Electoral Votes)

State Average Margin 2-to-1 Margin EVs
Arizona R+6.6% R+7.4% 11
Georgia R+7.4% R+8.2% 16
Montana R+8.6% R+9.8% 3
Nebraska-CD2 R+2.5% R+5% 1
South Carolina* R+9% R+9% 9

Finally, these are “strong” states, or states where a candidate leads by 10% or more in at least one of the projections:

Strong Obama (178 Electoral Votes)

State Average Margin 2-to-1 Margin EVs
California O+17.4% O+14.9% 55
Connecticut O+11.7% O+10.2% 7
Delaware* O+25% O+25% 3
District of Columbia O+80% O+79% 3
Hawaii O+27.5% O+26% 4
Illinois O+16.3% O+15.4% 20
Maine O+10.5% O+8.7% 3#
Maryland O+20.9% O+18.8% 10
Massachusetts O+18.6% O+17.3% 11
New Jersey O+12.1% O+9.8% 14
New York O+26.3% O+24.9% 29
Rhode Island O+21.9% O+18.1% 4
Vermont O+36.9% O+32.9% 3
Washington O+13.1% O+11.7% 12

Strong Romney (151 Electoral Votes)

State Average Margin 2-to-1 Margin EVs
Alabama* R+21% R+21% 9
Alaska* R+21% R+21.7% 3
Arkansas R+26.4% R+29.7% 6
Idaho R+36% R+39.1% 4
Indiana R+12.5% R+13.9% 11
Kansas R+20% R+22.4% 6
Kentucky R+14% R+16.4% 8
Louisiana R+18.7% R+21% 8
Mississippi* R+13% R+13% 6
Missouri R+8.9% R+10% 10
Nebraska R+16.3% R+17.6% 4#
North Dakota R+16.3% R+18.3% 3
Oklahoma R+29% R+33.1% 7
South Dakota R+10.1% R+12.1% 3
Tennessee R+24.9% R+33.1% 9
Texas R+16.2% R+17.8% 38
Utah R+44.5% R+46.1% 6
West Virginia R+14% R+17.1% 5
Wyoming* R+32% R+32.3% 3