Virginia Polls: Romney Has Substantial Lead in Southern Swing State
It is time to start re-evaluating how we use and look at polls. We should no longer consider polls where the internal data is not freely made available to the public. We spend way too much time paying attention to margins of error and who is up depending on the companies weight instead of applying that data to a wide range of likely voter turnout models.
The following three latest Virginia polls, Roanoke College, Gravis, and CBS/NY Times, offer substantial insight to voting characteristics and provide a valid snapshot of how party affiliation breaks down. The remarkable thing is one poll has Romney +5, another has it as a tie, and another has Obama at +2 — but all the data is remarkably consistent across all three polls in terms of party affiliation relating to voting tendencies. So how can three polls with consistent data provide such a broad range of predictions? The answer is in the weights. Unless you think Virginia is going to vote D+10 or greater, Romney is going to win Virginia based on these polls.
Want proof? Take a look the breakdown of percentages with regards to party affiliation and look how remarkably consistent the data is even with regards to Virgina.
Roanoke
Romney
Obama
Rep
95
5
Dem
2
94
Ind
59
33
Gravis
Romney
Obama
Rep
96
4
Dem
4
95
Ind
58
31
CBS/NY Times
Romney
Obama
Rep
93
6
Dem
3
96
Ind
57
36
Average
Romney
Obama
Rep
94.67
5.00
Dem
3.00
95.00
Ind
58.00
33.33
Want to see how it breaks down when you weight the polls?
Virgina Poll Weighted Breakdown
D+3
Weight
Romney
Obama
Rep
33
31.24
1.65
Dem
36
1.08
29.45
Ind
31
17.98
10.33
50.30
41.43
D+4
Weight
Romney
Obama
Rep
33
31.24
1.65
Dem
37
1.11
35.15
Ind
30
17.40
10.00
49.75
46.80
D+2
Weight
Romney
Obama
Rep
33
31.24
1.65
Dem
35
1.05
33.25
Ind
32
18.56
10.67
50.85
45.57
D+6
Weight
Romney
Obama
Rep
30
28.40
1.50
Dem
36
1.08
34.20
Ind
34
19.72
11.33
49.20
47.03
If we are going to avoid the fat tails and focus on where the consistency in data across polls occurs, we will find more nuggets of truth. Is it fantasyland a true battleground state is D+10? Likely, but entirely possible, but it is just as likely a battleground state is R+2 than it is D+10. Until there comes a set of data in recent polls in Virgina which show a dramatic shift in independent voters, Romney is safely ahead in Virgina if indeed Virginia is a battleground state.
Is it possible Obama could still win Virginia? Yes, but it depends likely on a combination of lower independent voter turnout and a high democratic turnout. It is not until you reduce the share of the independent vote to 18% percent of the total vote in Virginia in a D+4 scenario that Obama overtakes Romney. It also doesn't go for Obama without a 40% turnout for Democrats and a D+7 advantage.
D+4 Weight Romney Obama Rep 39 36.92 1.95 Dem 43 1.29 40.85 Ind 18 10.44 6.00 48.65 48.80 D+7 Weight Romney Obama Rep 33 31.24 1.65 Dem 40 1.20 38.00 Ind 27 15.66 9.00 48.10 48.65
Election Day is what matters, and for a poll to accurately reflect what happens on Election Day it has to match the turnout model with its weights.
