Tuesday, September 11, 2012

Obama’s DNC Bounce?

Here is how the past few election cycles voted based on exit polls in terms of political ideology:

Year

Democrat %

Republican %

Independent %

Notes

2000

39

35

27

Gore Wins by 0.1% Popular Vote

2002

38

40

22

Republicans pick up a few House seats

2004

37

37

26

Bush by 3%

2006

38

36

26

Democrats make big gains in the House

2008

39

32

29

Obama by over 7%

2010

35

35

29

Republicans make big gains in the House

Average

37.6666667

35.83333333

26.5

Democrats by 1.88%

The difficult thing to measure is what is going to be the breakdown in political ideology in the 2012 Election. Here are some poll figures and some simple manipulations to better understand the data and to break it down by voting ideology, ethnicity, and gender (those results are in bold font):

Poll

Democrat %

Republican %

Independent %

Obama %

Romney %

Margin %

Date

IPOS/Reuters

46

43

11

48

43

Democrat 3%

10-Sep

Gallup

38

34

28

50

44

Democrat 3.7%

Daily

Rasmussen

34

36

30

48

45

Republican 1.5%

Daily

NBC

34

27

36

48

44

Democrats 7%

21-Aug

FOX

42

38

18

44

45

Democrats 4%

PEW

33

22

42

51

41

Democrats 11%

CNN

40

29

31

52

46

Democrats 11%

9-Sep

CBS

31

29

34

46

45

Democrats 2%

26-Aug

TIPP

36

31

30

46

44

Democrats 5%

9-Sep

Battleground

44

40

15

48

47

Democrats 1%

Survey USA

33

27

40

46

43

Democrats 6%

30-Aug

AP/GfK

31

23

30

47

46

Democrats 8%

LA Times

39

33

26

48

46

Democrats 6%

YouGov

45

36

20

45

45

Democrats 9%

20-Aug

ABC

32

26

37

49

48

Democrats 6%

9-Sep

Democracy Corps

47

26

26

49

47

Democrats 21%

18-Aug

Resurgent Republic

37

30

30

46

45

Democrats 7%

18-Aug

American Research Group

38

34

28

46

49

Democrats 4%

7-Sep

Poll Average

37.7777778

31.33333333

28.44444444

47.61

45.17

6.444444444

2.444

2012 Model (Prediction)

38

34

28

1.0059

1.085

0.984375

Poll Scale

47.89

49.01

Poll

DO %

RO %

IO %

DR %

RR %

IR %

MO %

FO %

MR %

FR %

WO %

NWO %

WR %

NWR %

M %

F %

W %

NW %

IPOS/Reuters

Gallup

Rasmussen

NBC

48

52

74

26

FOX

88

4

32

5

92

42

40

48

48

42

36

73

53

17

PEW

CNN

97

2

40

3

96

54

48

55

47

44

42

88

55

11

CBS

89

5

40

7

90

41

TIPP

85

6

41

9

88

41

40

51

48

40

37

80

52

12

Battleground

78

22

Survey USA

87

7

37

6

87

40

40

50

44

38

39

73

49

18

48

52

73

27

AP/GfK

49

51

74

26

LA Times

46

54

76

24

YouGov

48

52

73

27

ABC

Democracy Corps

61

39

Resurgent Republic

90

5

37

6

93

45

46

54

74

26

American Research Group

85

5

44

11

92

49

41

51

54

45

39

84

57

16

48

52

76

24

Averages

88.7142857

4.857142857

38.71428571

6.714

91.143

44.57

41.8

51

48.2

41.8

38.6

79.6

53.2

14.8

47.57

52.43

73.22

26.78

2008 Averages

39

32

29

19.65

27.03

22.65

22.154

29.92

17.91

41.23

3.33

46.3

53.7

76.3

23.7

2010 Averages

35

35

29

47.3

52.7

78

22

2012 Model (Prediction)

38

34

28

38

34

28

47

53

77.5

22.5

33.71142857

1.651428571

10.84

2.5514

30.989

12.48

0.988

1.011

1.058

0.84

2012 Model Results

46.2028571

46.02

0.183

46.97

44.54

2.429

44.96

44.75

0.2093

Key

DO = Democrats for Obama

Obama

Romney

RO - Republicans for Obama

Poll Ave

47.61

45.167

IO - Independents for Obama

Poll Scaled

47.89

49.011

DR - Democrats for Romney

Poll Affiliation

46.2

46.02

RR - Republicans for Romney

Poll Gender

46.97

44.542

IR - Independents for Romney

Poll Ethnicity

44.96

44.755

MO - Males for Obama

FO - Females for Obama

Average

46.73

45.899

MR - Males for Romney

FR - Females for Romney

WO - Whites for Obama

NWO - Non Whites for Obama

WR - Whites for Romney

NWR - Non Whites for Romney

F - Female

M - Male

W - Whites

NW - Non Whites

85

6

41

9

88

41

To determine the ideological breakdowns of some polls I needed to carefully decipher these numbers from the questions within the polls (the information is not readily available). For instance, the latest CNN poll had Obama up by 6 points. However, the poll claims that Romney win independents by 14 points, for this to happen the poll breakdown must give the Democrats an 11 point advantage.

Although Republicans are not overly excited by Romney, polls indicate they are more energized than Democrats because they do not want another 4 years of Obama. The generic congressional ballot between Democrats and Republicans is even and that is good news for Republicans because they historically trail in this poll. Are the poll models above correct? The Gallup poll is a bit concerning for Romney, but it is not as bad as it could be. The average of the last polls by 13 firms (7 since the DNC) has Obama ahead by nearly 2.5 points. Obama may be much more popular than Romney, but the economy is in the dumps and not getting better so it is hard to believe that Obama will win independents as he did in 2008. However, many of these polls seem to be oversampling Democrats by large margins when compared to historical trends outlined above. What’s worse, many of these same polling outlets are making the same mistake in state election polls. My best guess at an ideological breakdown in 2012 is 38-D/34-R/29-I. If that is the case when this election modeled using these numbers we get the following:

Obama – 47.9% / Romney 49%

Also, using the above data in the polls when broken out by party, gender, and race (not all polls give this data) then my models show the following (I am expecting 53% / 47% margin for women and a 77.5% / 22.5% margin for white versus non-white voters in 2012).

By Party: Obama 46.2% / Romney 46%

By Gender: Obama 47% / Romney 44.5%

By Ethnicity: Obama 45% / Romney 44.8%

There you have it. I have modeled (scaled the highly liberal polls) the polls 5 ways and Obama comes out on top in 4. If I average all 5 methods then we are really in a dead heat: Obama 46.7%, Romney 45.9% (Obama has a 0.8% lead).

2 comments:

  1. I like your numbers better than what I'm seeing in the news, Patrick.

    ReplyDelete
  2. I am trying to decipher the fine details of the polls. I need to make this file a bit more readable, it does not transfer very well. I will post it again in a few weeks after the bounce wears off.

    ReplyDelete