Jump to ContentJump to Main Navigation
The Ambivalent PartisanHow Critical Loyalty Promotes Democracy$

Howard G. Lavine, Christopher D. Johnston, and Marco R. Steenbergen

Print publication date: 2012

Print ISBN-13: 9780199772759

Published to Oxford Scholarship Online: January 2013

DOI: 10.1093/acprof:oso/9780199772759.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 26 July 2017

(p.225) Appendix A: Statistical Model Estimates

(p.225) Appendix A: Statistical Model Estimates

Source:
The Ambivalent Partisan
Publisher:
Oxford University Press

Chapter 3

Table A3.1. Individual-Level Antecedents of Partisan Ambivalence

Conflicting Evaluations

Consistent Evaluations

% Change

Lower

Upper

% Change

Lower

Upper

Unemployed

26.14

9.09

44.99

–7.91

–15.93

0.23

Contact w/parties

3.08

–5.23

12.24

2.00

–2.66

6.91

Group feelings

90.62

78.34

101.74

–26.78

–29.13

–24.27

Moral traditionalism

20.15

10.71

30.67

–6.77

–10.81

–2.92

Egalitarianism

39.12

24.53

55.34

–25.19

–29.73

–20.17

Worried job

–5.10

–9.94

0.07

3.64

0.58

6.91

House contact

6.74

–1.47

16.09

1.14

–3.47

5.85

Discuss politics

11.36

–0.19

23.92

3.43

–2.67

9.53

Keep up with costs

26.84

13.95

41.82

–11.57

–16.90

–6.02

Notes. ANES Cumulative File, 1980–2004. Entries are negative binomial estimates of the percent change in each component of partisan ambivalence, with confidence bounds, for a change in the respective variable from its 5th to its 95th percentile. This table is constructed from multiple models. See the text for more details on the procedure.

(p.226) Chapter 4

Table A4.1. Policy Preferences as a Function of Partisanship, 2008 ANES Panel Study

Variable

B

SE

p

Lagged preference

.52

.01

.00

Age

–.01

.03

.65

Male

.02

.01

.08

Black

–.05

.03

.07

Education

–.01

.02

.59

Income

.03

.03

.28

Unemployed

–.01

.02

.47

Republican

–.15

.05

.00

Strength

–.01

.02

.49

Interest

–.05

.04

.13

Sophistication

–.10

.04

.01

Consistent

–.05

.03

.12

Conflicting

.05

.05

.34

Rep × Strength

.06

.03

.06

Rep × Interest

.12

.05

.01

Rep × Sophistication

.13

.05

.01

Rep × Consistent

.17

.05

.00

Rep × Conflicting

–.06

.06

.32

Constant

.27

.04

.00

SD (Constant)

.07

.01

N

4,159

Respondents

520

Notes. ANES 2008 panel. Entries are restricted maximum likelihood coefficients and standard errors. The intercept is modeled as a random variable that varies by respondent. The dependent variable is measured in October, and lagged preferences are measured in January. Partisan identification, strength, and political interest are measured in January, while ambivalence and sophistication are measured in February. Model also includes fixed effects for issue (not shown).

(p.227)

Table A4.2. Economic Preferences as a Function of Partisanship and Income, 2008 ANES Panel Study

Variable

B

SE

p

Lagged preference

.58

.02

.00

Age

–.01

.04

.78

Male

.03

.02

.04

Black

–.01

.04

.85

Education

.04

.03

.16

Income

.34

.14

.02

Unemployed

.01

.03

.74

Republican

–.15

.06

.02

Strength

–.08

.07

.24

Interest

.00

.10

.98

Sophistication

–.07

.11

.53

Consistent

.11

.09

.26

Conflicting

.18

.14

.21

Rep x Strength

.04

.04

.38

Rep x Interest

.14

.07

.04

Rep x Sophistication

.07

.07

.30

Rep x Consistent

.27

.07

.00

Rep x Conflicting

–.02

.09

.80

Income x Strength

.12

.10

.22

Income x Interest

–.13

.15

.37

Income x Sophistication

.01

.16

.97

Income x Consistent

–.33

.15

.02

Income x Conflicting

–.14

.21

.50

Constant

.00

.10

1.00

SD (Constant)

.10

.01

N

1,559

Respondents

520

Notes. ANES 2008 panel. Entries are restricted maximum likelihood coefficients and standard errors. The intercept is modeled as a random variable that varies by respondent. The dependent variable is measured in October, and lagged preferences are measured in January. Partisan identification, strength, and political interest are measured in January, while ambivalence and sophistication are measured in February. Model also includes fixed effects for issue (not shown).

(p.228)

Table A4.3. Economic Preferences as a Function of Partisanship and Personal Economic Circumstances, 1992–1994 ANES Panel Study

Variable

B

SE

p

Lagged preferences

.33

.06

.00

Age

–.03

.04

.47

Male

.03

.02

.19

Black

–.10

.03

.00

Education

–.06

.04

.08

Income

.10

.04

.01

Republican

.01

.06

.93

Personal economic

.05

.14

.71

Strength

–.02

.06

.67

Interest

.06

.08

.51

Sophistication

–.04

.09

.69

Consistent

.07

.07

.32

Conflicting

–.07

.05

.18

Rep x Strength

.10

.05

.03

Rep x Interest

–.03

.06

.62

Rep x Sophistication

.07

.07

.35

Rep x Consistent

.11

.05

.04

Rep x Conflicting

–.09

.04

.03

Personal Ec. x Strength

.00

.10

.99

Personal Ec. x Interest

.00

.14

.98

Personal Ec. x Sophistication

–.03

.14

.84

Personal Ec. x Consistent

–.11

.10

.29

Personal Ec. x Conflicting

.20

.08

.02

Constant

.29

.09

.00

.43

N

434

Notes. ANES 1992–1994 panel. Entries are OLS coefficients and robust standard errors. Dependent variable is measured in 1994. All independent variables are measured in 1992.

(p.229) Chapter 5

(p.230)

Table A5.1. Bias and Responsiveness in Economic Retrospections, 1980–2004

Variable

Δ in Unemployment

GDP Growth

B

SE

p

B

SE

p

Age

.00

.01

.73

.00

.01

.74

Female

–.04

.01

.00

–.04

.01

.00

Black

–.04

.01

.00

–.04

.01

.00

Education

.05

.01

.00

.05

.01

.00

Income

.02

.01

.12

.02

.01

.12

Unemployed

–.03

.01

.01

–.03

.01

.01

Partisan strength

–.03

.01

.00

–.04

.01

.00

Interest

–.04

.01

.01

–.05

.02

.01

Sophistication

–.01

.01

.52

–.06

.02

.00

Conflicting

.11

.03

.00

.01

.04

.81

Consistent

–.12

.02

.00

–.09

.03

.00

In-Party

–.02

.01

.09

–.02

.01

.08

In-Party x Strength

.08

.01

.00

.08

.01

.00

In-Party x Interest

.07

.02

.00

.07

.02

.00

In-Party x Sophistication

.03

.02

.08

.03

.02

.07

In-Party x Conflicting

–.29

.05

.00

–.29

.04

.00

In-Party x Consistent

.26

.03

.00

.25

.03

.00

Change in Economy

–.06

.02

.01

.03

.01

.01

Change x Strength

–.01

.01

.30

.00

.00

.27

Change x Interest

–.01

.01

.29

.00

.00

.33

Change x Sophistication

–.04

.01

.00

.02

.00

.00

Change x Conflicting

–.07

.02

.00

.03

.01

.00

Change x Consistent

.02

.01

.16

–.01

.01

.10

Constant

.40

.03

.00

.32

.04

.00

SD intercept

.08

.02

.00

.08

.02

.00

SD residuals

.23

.00

.00

.23

.00

.00

N

8,159

8,159

Years

11

11

Notes. Data from the ANES cumulative file, 1980–2004; dependent variable is economic retrospections (higher values = “better”); models include random intercepts for year and are estimated with a normal link function.

(p.231)

Table A5.2. Economic Retrospections in the 2008 Presidential Election

Variable

B

SE

p

Lagged retrospection

.24

.02

.00

Age

.01

.02

.50

Male

.00

.01

.89

Black

–.01

.02

.66

Education

–.03

.02

.16

Income

.00

.02

.82

Republican

.01

.04

.74

Strength

.01

.02

.71

Interest

–.01

.03

.67

Sophistication

–.01

.03

.71

Consistent

.02

.03

.46

Conflicting

.00

.04

.94

Rep x Strength

–.01

.02

.83

Rep x Interest

.05

.04

.23

Rep x Sophistication

.04

.04

.37

Rep x Consistent

.03

.04

.46

Rep x Conflicting

–.11

.05

.03

Constant

–.01

.03

.81

.21

N

981

Notes. ANES 2008 panel. Entries are OLS coefficients and robust standard errors. Dependent variable is measured in October (higher values = “gotten better”). Lagged retrospections are measured in January. Partisan identification, strength, and political interest are measured in January, while ambivalence and sophistication are measured in February.

Table A5.3. Accuracy of Unemployment Retrospections During Reagan’s Presidency

Democrats

Republicans

Variable

B

SE

p

B

SE

p

Age

–.39

.25

.12

–.22

.32

.48

Male

.05

.10

.65

.47

.13

.00

Black

–.21

.12

.08

–.40

.28

.16

Education

.41

.21

.06

.21

.29

.46

Income

.22

.19

.26

.45

.26

.08

Unemployed

–.29

.19

.13

–.07

.33

.82

Strength

–.11

.12

.36

.37

.16

.02

Interest

–.14

.14

.32

.11

.19

.58

Sophistication

.61

.23

.01

.79

.31

.01

Consistent

–.48

.19

.01

.23

.30

.45

Conflicting

1.14

.36

.00

–.35

.32

.28

Cut 1

–.23

.17

–.30

.22

Cut 2

.45

.17

.41

.22

Pseudo-R²

.05

.10

N

646

549

Notes. 1988 ANES. Entries are ordered probit coefficients and standard errors. Higher values of dependent variable indicate more positive retrospections.

(p.232)

Table A5.4. Accuracy of Inflation Retrospections During Reagan’s Presidency

Variable

Democrats

Republicans

B

SE

p

B

SE

p

Age

.21

.25

.40

–.63

.29

.03

Male

.28

.10

.01

.42

.11

.00

Black

.10

.12

.39

.20

.28

.48

Education

.55

.21

.01

.28

.26

.27

Income

.09

.19

.64

.25

.22

.26

Unemployed

–.29

.20

.14

–.23

.31

.47

Strength

–.10

.12

.40

.13

.14

.37

Interest

–.27

.14

.06

–.07

.18

.71

Sophistication

.86

.23

.00

1.26

.28

.00

Consistent

–.37

.20

.06

.87

.27

.00

Conflicting

1.51

.35

.00

–.01

.29

.96

Cut 1

.53

.17

.24

.20

Cut 2

1.35

.18

1.01

.20

Pseudo-R²

.08

.13

N

653

548

Notes. 1988 ANES. Entries are ordered probit coefficients and standard errors. Higher values of dependent variable indicate more positive retrospections.

Table A5.5. Accuracy of Deficit Retrospections During Clinton’s Presidency

Variable

Democrats

Republicans

B

SE

p

B

SE

p

Age

.87

.48

.07

.25

.48

.60

Male

.00

.17

1.00

–.01

.17

.95

Black

–.25

.20

.23

–.08

.52

.88

Education

.45

.39

.24

.95

.36

.01

Income

1.60

.61

.01

.01

.51

.98

Strength

–.01

.21

.96

.40

.21

.06

Interest

.34

.25

.17

–.04

.28

.89

Sophistication

.94

.41

.02

.43

.41

.30

Consistent

1.03

.39

.01

–1.05

.37

.00

Conflicting

.63

.54

.24

.96

.45

.03

Cut 1

.53

.36

.13

.36

Cut 2

1.42

.36

.83

.36

Pseudo-R²

.15

.05

N

301

220

Notes. 2000 ANES. Entries are ordered probit coefficients and standard errors. Higher values of dependent variable indicate more positive retrospections.

(p.233)

Table A5.6. Accuracy of Crime Retrospections During Clinton’s Presidency

Variable

Democrats

Republicans

B

SE

p

B

SE

p

Age

–.34

.42

.42

.93

.44

.04

Male

.29

.14

.04

.28

.16

.09

Black

–.36

.19

.05

–.02

.49

.97

Education

.58

.31

.06

.31

.34

.36

Income

.41

.43

.34

.24

.47

.61

Strength

.20

.18

.28

.08

.20

.68

Interest

.19

.21

.36

.28

.26

.29

Sophistication

1.01

.33

.00

.41

.38

.28

Consistent

.18

.32

.57

–.11

.35

.74

Conflicting

.39

.41

.34

.83

.43

.05

Cut 1

.41

.30

.82

.34

Cut 2

1.27

.31

1.80

.35

Pseudo-R²

.09

.06

N

311

234

Notes. 2000 ANES. Entries are ordered probit coefficients and standard errors. Higher values of dependent variable indicate more positive retrospections.

(p.234)

Table A5.7. Projection and Inference in Perceptions of Presidential Candidates’ Policy Positions, 1980–2004

Variable

B

SE

p

Age

–.01

.01

.19

Female

.01

.00

.00

Black

.05

.00

.00

Education

–.01

.00

.01

Income

–.02

.01

.00

Candidate party

.11

.01

.00

R’s preference

.06

.01

.00

Strength

–.05

.01

.00

Interest

–.08

.01

.00

Sophistication

–.07

.01

.00

Consistent

–.17

.01

.00

Conflicting

.14

.01

.00

Party x Strength

–.01

.00

.06

Party x Interest

.06

.01

.00

Party x Sophistication

.14

.01

.00

Party x Consistent

.10

.01

.00

Party x Conflicting

.14

.01

.00

Pref. x Strength

.09

.01

.00

Pref. x Interest

.10

.01

.00

Pref. x Sophistication

–.02

.01

.07

Pref. x Consistent

.29

.01

.00

Pref. x Conflicting

–.45

.02

.00

Constant

.50

.01

.00

SD (constant)

.07

.00

N

57,215

Respondents

6,725

Notes. ANES Cumulative File, 1980–2004. Entries are restricted maximum likelihood coefficients and standard errors. “Candidate Party” represents processes of ideological inference; “R’s preference” represents projection. The intercept is modeled as a random variable which varies by respondent. Model includes fixed effects for issue, which vary by the party of the candidate being judged (not shown).

(p.235)

Table A5.8. Projection and Inference in Perceptions of Senators’ Roll-Call Votes, 2006

Variable

B

SE

p

Age

–.11

.04

.00

Male

–.01

.01

.51

Black

.06

.03

.02

Education

.00

.03

.99

Income

–.05

.03

.11

Senator’s party

.39

.06

.00

R’s policy position

.66

.06

.00

Partisan strength

–.09

.03

.00

Interest

–.71

.04

.00

Ambivalence

.00

.04

.98

Party x Strength

–.16

.04

.00

Party x Interest

1.14

.05

.00

Party x Ambivalence

.55

.05

.00

Position x Strength

.24

.04

.00

Position x Interest

.23

.06

.00

Position x Ambivalence

–.65

.05

.00

Constant

–.12

.05

.01

Pseudo-R²

.44

N

123,786

Notes. 2006 CCES. Entries are probit coefficients and standard errors. “Senator’s party” represents processes of ideological inference; “R’s policy position” represents projection. Model includes fixed effects for issue, which vary by the party of the candidate being judged (not shown).

(p.236)

Table A5.9. Projection and Inference in Perceptions of Presidential Candidates’ Policy Positions, 2008

Variable

B

SE

p

Lagged perception

.32

.01

.00

Age

.00

.02

.92

Male

–.02

.01

.00

Black

.03

.02

.14

Education

.02

.02

.27

Income

–.02

.02

.48

R’s position on issue

–.09

.04

.03

Party of candidate

–.02

.04

.54

Strength

–.01

.03

.61

Interest

–.09

.03

.01

Sophistication

–.07

.03

.03

Consistent

–.12

.03

.00

Conflicting

–.05

.04

.25

R’s x Strength

.05

.04

.16

R’s x Interest

–.02

.04

.60

R’s x Sophistication

.07

.04

.11

R’s x Consistent

.20

.04

.00

R’s x Conflicting

–.10

.05

.05

Party x Strength

–.03

.03

.29

Party x Interest

.11

.03

.00

Party x Sophistication

.09

.03

.01

Party x Consistent

.07

.03

.02

Party x Conflicting

.11

.04

.01

Constant

.43

.04

.00

SD (constant)

.05

.01

N

6,796

Respondents

475

Notes. 2008 ANES panel. Entries are restricted maximum likelihood coefficients and standard errors. The intercept is modeled as a random variable that varies by respondent. Dependent variable is respondent’s placement of the candidate in the October wave. Lagged perceptions are measured in June. Partisan identification, strength, and political interest are measured in January, while ambivalence and sophistication are measured in February. Model includes fixed effects for issue, which vary by the party of the candidate being judged (not shown).

(p.237)

Table A5.10. Accuracy of Presidential Perceptions, 1980–2004

Variable

B

SE

p

Age

–1.51

.11

.00

Female

–.30

.05

.00

Black

.32

.07

.00

Education

.59

.09

.00

Income

.26

.09

.00

Mismatched

–.19

.10

.06

Strength

.15

.07

.02

Interest

1.13

.10

.00

Sophistication

1.32

.10

.00

Consistent

2.17

.12

.00

Conflicting

.40

.18

.02

Strength x Mismatch

–.45

.09

.00

Interest x Mismatch

–.59

.14

.00

Sophistication x Mismatch

–.02

.13

.88

Consistent x Mismatch

–2.02

.17

.00

Conflicting x Mismatch

2.09

.23

.00

Constant

–.69

.10

.00

N

32,400

Respondents

6,935

Notes. ANES Cumulative File, 1980–2004. Entries are restricted maximum likelihood coefficients and standard errors. The intercept is modeled as a random variable that varies by respondent. Dependent variable is coded “1” for correctly placing the Democratic candidate to the left of the Republican. “Mismatched” is coded “1” for citizens whose own position on the issue conflicts with their party’s position. Model includes fixed effects for issue.

Table A5.11. Accuracy of Senatorial Perceptions, 2006

Variable

B

SE

P

Age

.23

.03

.00

Male

.23

.01

.00

Black

–.05

.02

.03

Education

.25

.02

.00

Income

.27

.03

.00

Mismatch

–.28

.04

.00

Strength

.14

.02

.00

Interest

1.03

.03

.00

Ambivalence

.14

.02

.00

Strength x Mismatch

–.17

.03

.00

Interest x Mismatch

–.33

.04

.00

Ambivalence x Mismatch

.32

.03

.00

Constant

–1.12

.03

.00

Pseudo-R²

.10

N

174,789

Notes. 2006 CCES. Entries are probit coefficients and standard errors. Dependent variable is coded “1” for a correct placement of the Senator on the roll-call vote. “Mismatch” is coded “1” for citizens whose own position on the issue conflicts with their party’s. Model includes fixed effects for issue.

Chapter 6

Table A6.1. Timing of Vote Choice, 1980–2004

Variable

B

SE

p

Age

–.01

.00

.04

Male

–.06

.10

.53

White

.64

.22

.00

Education

.03

.03

.30

Income

–.03

.07

.61

Democrat

.38

.13

.00

Partisan strength

–.92

.08

.00

Interest

–.83

.18

.00

Sophistication

.21

.17

.21

Consistent

–1.34

.25

.00

Conflicting

1.41

.34

.00

Constant

–1.02

.28

.00

N

6,132

Notes. ANES cumulative file, 1980–2004. Entries are maximum likelihood logit estimates with estimated standard errors.

(p.238)

Table A6.2. Choice Set Size, 1992

Variable

B

SE

p

Age

.00

.00

.05

Male

–.01

.05

.86

White

.16

.09

.08

Education

–.02

.02

.40

Income

.04

.03

.17

Democrat

.03

.05

.55

Strength

–.12

.06

.06

Interest

–.07

.11

.54

Sophistication

–.01

.12

.96

Consistent

–.19

.11

.09

Conflicting

.29

.17

.08

Constant

.50

.17

.00

N

1,099

Notes. 1992 ANES. Entries are maximum likelihood ordered logit estimates with estimated standard errors.

Table A6.3. Uncertainty in Vote Choice, 1980–2004

Variable

Presidential

House

B

SE

p

B

SE

p

Panel A. Utility Model

Age

.00

.00

.30

.00

.00

.44

Male

–.01

.02

.58

.02

.02

.31

White

–.29

.05

.00

–.25

.05

.00

Education

.01

.01

.04

–.01

.01

.27

Income

–.02

.01

.07

–.03

.01

.01

Democrat

.90

.10

.00

.71

.09

.00

Constant

–.40

.08

.00

.06

.09

.52

Panel B. Variance Model

Strength

–.46

.07

.00

–.57

.07

.00

Interest

–.28

.11

.01

.06

.11

.58

Sophistication

–.35

.10

.00

–.18

.09

.04

Consistent

–1.74

.12

.00

–.69

.10

.00

Conflicting

1.95

.18

.00

.63

.16

.00

N

5,014

6,512

Notes. ANES cumulative file, 1980–2004. Entries are maximum likelihood heteroskedastic probit estimates with estimated standard errors. The standard errors for vote choice in presidential elections are clustered by state; those for vote choice in House elections are clustered by district. The election years 1998 and 2002 drop out due to the absence of a measure of partisan ambivalence. Year dummies are not shown.

(p.239)

Table A6.4. Voter Turnout, 1980–2004

Variable

Presidential

House

B

SE

p

B

SE

p

Age (in decades)

.71

.13

.00

–.29

.48

.55

Age squared

–.05

.01

.00

.03

.03

.28

Male

–.33

.06

.00

–.05

.09

.61

White

–.14

.11

.19

.04

.04

.34

Education

.26

.03

.00

.04

.05

.47

Income

.27

.04

.00

.05

.17

.75

Internal efficacy

.00

.02

.95

.05

.08

.52

External efficacy

.13

.03

.00

.02

.05

.71

Contact

.73

.07

.00

.49

.70

.48

Strength

.64

.17

.00

.05

.32

.88

Interest

1.68

.15

.00

.31

1.51

.84

Sophistication

1.21

.13

.00

.75

1.05

.48

Consistent

.46

.19

.02

.13

.08

.09

Conflicting

–.10

.27

.70

–.06

.23

.80

Constant

–4.59

.33

.00

–.32

3.91

.93

N

6,032

1,904

Notes. ANES cumulative file, 1980–2004. Table entries are maximum likelihood logit estimates with cluster-corrected standard errors (clustering by year).

Table A6.5. Media Usage, 1980–2004

Variable

B

SE

p

Age

.01

.00

.00

Male

.10

.08

.23

White

–.07

.10

.47

Education

.13

.02

.00

Income

.10

.04

.01

Internal efficacy

.03

.03

.35

External efficacy

.07

.03

.03

Contact

.33

.10

.00

Strength

–.12

.03

.00

Interest

2.07

.18

.00

Sophistication

.69

.14

.00

Consistent

.46

.05

.00

Conflicting

.75

.17

.00

Constant

–3.30

.22

.00

N

6,342

Notes. ANES cumulative file, 1980–2004. Table entries are maximum likelihood logit estimates with cluster-corrected standard errors (clustering by year).

(p.240)

Table A6.6. Voting Behavior in Presidential Elections, 1984–2004

Variable

B

SE

p

Age

–.05

.14

.73

Female

–.03

.06

.56

Black

–.85

.10

.00

Education

.18

.11

.08

Income

.28

.12

.02

Unemployed

–.07

.17

.69

Econ. retrospections

.82

.31

.01

Republican

.54

.17

.00

Economic preferences

.01

.42

.98

Social preferences

.50

.32

.12

Partisan strength

–.39

.25

.11

Interest

–.71

.39

.07

Sophistication

–1.65

.35

.00

Consistent

–.96

.52

.07

Conflicting

1.61

.62

.01

Rep x Strength

1.19

.15

.00

Rep x Interest

–.06

.23

.78

Rep x Sophistication

.58

.21

.01

Rep x Consistent

4.78

.33

.00

Rep x Conflicting

–4.96

.35

.00

Retros x Strength

.03

.26

.90

Retros x Interest

.37

.40

.37

Retros x Sophistication

.31

.38

.41

Retros x Consistent

–.36

.57

.53

Retros x Conflicting

–.31

.63

.62

Econ x Strength

.17

.38

.66

Econ x Interest

.76

.57

.18

Econ x Sophistication

1.46

.51

.01

Econ x Consistent

–2.91

.84

.00

Econ x Conflicting

2.00

.94

.03

Social x Strength

–.63

.28

.03

Social x Interest

.40

.42

.35

Social x Sophistication

.89

.39

.02

Social x Consistent

–.10

.62

.87

Social x Conflicting

.34

.68

.62

Constant

–.65

.30

.03

Pseudo-R2

.60

N

5,289

Notes. ANES cumulative file, 1980–2004. Entries are probit coefficients and standard errors. Year dummies are not shown.

(p.241)

Table A6.7. Voting Behavior in the 2008 Presidential Election, 2008

Variable

B

SE

p

Lagged evaluations

.48

.03

.00

Age

–.02

.03

.55

Male

–.01

.01

.64

Black

–.07

.02

.00

Education

–.03

.02

.13

Income

–.05

.03

.10

Economic retrospections

.08

.03

.01

Republican

–.04

.06

.51

Issues

.05

.15

.71

Partisan strength

.00

.03

.94

Interest

–.09

.06

.16

Sophistication

–.19

.06

.00

Consistent

–.13

.07

.05

Conflicting

.08

.09

.36

Rep x Strength

.14

.04

.00

Rep x Interest

.12

.06

.03

Rep x Sophistication

–.01

.06

.93

Rep x Consistent

.32

.06

.00

Rep x Conflicting

–.29

.08

.00

Issues x Strength

–.12

.09

.17

Issues x Interest

.04

.14

.77

Issues x Sophistication

.27

.14

.06

Issues x Consistent

–.09

.15

.56

Issues x Conflicting

.19

.20

.32

Constant

.35

.07

.00

R2

.73

N

985

Notes. 2008 ANES Panel. Entries are OLS coefficients and standard errors. All variables scaled from 0 to 1.

(p.242)

Table A6.8. Voting Behavior in US House Elections, 1984–2004

Variable

B

SE

p

Age

.00

.11

1.00

Female

.06

.04

.14

Black

–.39

.09

.00

Education

.15

.08

.05

Income

.16

.09

.08

Unemployed

.18

.14

.18

Economic retrospections

.46

.25

.07

Republican incumbent

1.07

.21

.00

Democratic incumbent

–.73

.20

.00

Republican

.58

.14

.00

Economic preferences

.56

.35

.11

Social preferences

.11

.25

.65

Partisan strength

–.82

.21

.00

Interest

–.01

.35

.98

Sophistication

.31

.32

.33

Consistent

–.33

.36

.36

Conflicting

–.53

.49

.28

Rep. Inc. × Strength

.12

.17

.47

Rep. Inc. x Interest

–.06

.27

.83

Rep. Inc. x Sophistication

–.52

.25

.04

Rep. Inc. x Consistent

.12

.28

.66

Rep. Inc. x Conflicting

–.37

.37

.31

Dem. Inc. x Strength

.04

.16

.80

Dem. Inc. x Interest

–.06

.26

.82

Dem. Inc. x Sophistication

–.29

.24

.23

Dem. Inc. x Consistent

.29

.27

.29

Dem. Inc. x Conflicting

–.21

.36

.55

Retros x Strength

.14

.19

.44

Retros x Interest

.32

.30

.29

Retros x Sophistication

–.40

.29

.16

Retros x Consistent

–.87

.33

.01

Retros x Conflicting

.63

.43

.14

Rep. x Strength

.96

.12

.00

Rep x Interest

.00

.18

1.00

Rep x Sophistication

.04

.17

.80

Rep x Consistent

1.80

.21

.00

Rep x Conflicting

–1.72

.25

.00

Econ x Strength

.53

.29

.06

Econ x Interest

–.21

.45

.64

Econ x Sophistication

.19

.41

.64

Econ x Consistent

–.88

.49

.07

Econ x Conflicting

2.04

.65

.00

Social x Strength

–.07

.20

.72

Social x Interest

.01

.31

.97

Social x Sophistication

.38

.29

.19

Social x Consistent

.05

.34

.89

Social x Conflicting

.66

.45

.14

Constant

–1.07

.29

.00

Pseudo-R2

.46

N

7,003

Notes. ANES cumulative file, 1980–2004. Entries are probit coefficients and standard errors. Year dummies are not shown.

(p.243)

Table A6.9. Defection Among Unsorted Partisans, 1980–2004

Variable

Unsorted Economic

Unsorted Social

Unsorted on Both

B

SE

p

B

SE

p

B

SE

p

1988

.30

.22

.17

1992

.80

.28

.00

–.16

.18

.38

.24

.49

.62

1996

.21

.28

.46

–.26

.20

.19

–.05

.48

.91

2000

.58

.28

.03

–.12

.17

.49

–.08

.55

.89

2004

–.39

.23

.09

–.52

.16

.00

–1.04

.40

.01

Age

–.33

.35

.34

.05

.29

.87

.43

.83

.60

Female

.00

.14

.98

–.09

.12

.43

–.14

.29

.62

Black

.22

.26

.38

.14

.22

.51

.27

.50

.59

Education

.62

.26

.02

.15

.22

.50

.93

.56

.10

Income

–.37

.29

.20

–.09

.24

.73

.78

.61

.20

Strength

.64

.19

.00

1.08

.16

.00

1.26

.47

.01

Interest

.00

.28

.99

–.08

.24

.75

.38

.65

.56

Sophistication

–.13

.25

.61

.07

.23

.76

–.98

.60

.10

Consistent

3.13

.43

.00

2.87

.34

.00

2.01

.95

.04

Conflicting

–2.84

.39

.00

–2.78

.32

.00

–2.07

.68

.00

Constant

–.14

.34

.69

.17

.26

.52

–.67

.67

.32

Pseudo-R2

.33

.32

.41

N

561

831

137

Notes. ANES cumulative file, 1980–2004. Entries are probit coefficients and standard errors. Dependent variable is coded “1” if respondent voted for the in-party, and zero if he or she voted for the out-party. Economically unsorted respondents were unsorted on two of the three issues examined, while socially unsorted respondents were unsorted on both of two issues examined.

(p.244)

Table A6.10. Partisan Ambivalence, Defection, Split-Ticket, and Third-Party Voting

Variable

Defect Pres.

Defect House

Split-Ticket

Third-Party

B

SE

p

B

SE

p

B

SE

p

B

SE

p

Age

.00

.00

.75

.00

.00

.08

.00

.00

.25

–.03

.01

.00

Male

–.09

.10

.36

–.06

.08

.42

–.06

.09

.47

.41

.16

.01

White

1.00

.18

.00

.83

.17

.00

.66

.22

.00

2.05

.53

.00

Education

–.05

.04

.17

–.01

.02

.63

.01

.03

.82

–.09

.06

.12

Income

.01

.05

.91

.00

.03

.89

.04

.05

.37

.13

.09

.13

Democrat

1.09

.10

.00

–.09

.18

.60

–.05

.13

.70

.18

.16

.27

Out-party familiarity

1.17

.13

.00

Major candidate dislike

2.30

.29

.00

Strength

–.89

.13

.00

–.78

.08

.00

–.65

.11

.00

–1.15

.22

.00

Interest

–.43

.20

.03

.07

.11

.54

–.04

.18

.81

–.35

.34

.31

Sophistication

–.44

.19

.02

–.21

.13

.12

–.45

.18

.01

–.10

.35

.79

Consistent

–4.73

.32

.00

–1.39

.21

.00

–1.10

.20

.00

–1.38

.42

.00

Conflicting

4.53

.10

.00

1.40

.28

.00

1.04

.27

.00

1.50

.56

.01

Constant

–1.35

.32

.00

–.85

.30

.00

–1.57

.35

.00

-2.62

.68

.00

N

5,014

6,512

4,285

1921

Notes. Table entries for defection in presidential elections are maximum likelihood logit estimates with estimated standard errors. Table entries for defection in House elections are maximum likelihood hierarchical logit estimates with their estimated standard errors. The hierarchical model contains a random intercept across states and congressional districts (estimated variance component is .025, ns).

(p.245)

Table A6.11. Effects of Anxiety and Enthusiasm on Vote Intention, 1980–2004

Variable

B(SE)

Party ID

1.06 (.14)*

Issues

2.09 (.40)*

In-anxiety

.60 (.32)^

Out-anxiety

.28 (.25)

In-enthus

–.56 (.28)*

Out-enthus

–1.05 (.31)*

PID x In-Anx

–.65 (.19)*

PID x Out-Anx

.52 (.19)*

PID x In-Enthus

1.89 (.20)*

PID x Out-Enthus

–1.79 (.20)*

Issues x In-Anx

–.56 (.62)

Issues x Out-Anx

–.96 (.54)^

Issues x In-Enthus

–1.05 (.59)^

Issues x Out-Enthus

3.80 (.62)*

Candidate evaluations

7.23 (.21)*

Age

.02 (.01)

Gender

.02 (.04)

Black

–.32 (.07)*

Hispanic

–.02 (.02)

Education

.19 (.08)*

Income

–.05 (.08)

Interest

–.04 (.07)

Cut 1

–4.77 (.24)*

Cut 2

.51 (.02)*

N

8,512

Notes. ANES cumulative file, 1980–2004. Entries are ordered probit coefficients and standard errors.

*p 〈 .05; ^p 〈 .10.

(p.246)

Table A6.12. Effects of Anxiety and Enthusiasm on Candidate Evaluations, 2008

Variable

B(SE)

Lagged evaluation

.38 (.04)*

Age

–.04 (.02)

Male

–.02 (.01)^

Black

.00 (.02)

Education

–.06 (.02)*

Income

–.06 (.03)*

Republican

.29 (.04)*

Issue preferences

.19 (.08)*

In-anxiety

.04 (.08)

Out-anxiety

–.09 (.04)*

In-enthusiasm

–.11 (.04)*

Out-enthusiasm

.11 (.06)^

Rep x In-Anx

–.05 (.10)

Rep x Out-Anx

.05 (.06)

Rep x In-Enth

.23 (.06)*

Rep x Out-Enth

–.44 (.08)*

Issues x In-Anx

–.03 (.16)

Issues x Out-Anx

.16 (.10)^

Issues x In-Enth

–.09 (.11)

Issues x Out-Enth

.27 (.16)^

Constant

.19 (.04)*

R2

.71

N

1,079

Notes. ANES 2008 Panel. Entries are OLS coefficients and standard errors.

*p 〈 .05; ^p 〈 .10.

(p.247) Chapter 7

Table A7.1. Partisan Change, 1992–1994

Variable

Unsorted Respondents

Sorted Respondent

Health Insurance

Moral Traditionalism

Health Insurance

Moral Traditionalism

B

SE

p

B

SE

p

B

SE

p

B

SE

p

Age

1.60

.99

.11

–.17

.70

.81

–.27

.49

.59

.38

.63

.54

Male

.17

.50

.74

–.03

.32

.93

.06

.25

.82

.09

.31

.79

Black

.68

.92

.46

.35

.53

.51

–.35

.42

.40

–.11

.55

.85

Education

–1.24

.83

.13

–.28

.67

.68

–.31

.52

.56

–.38

.58

.51

Income

2.32

1.02

.02

.24

.66

.71

–.06

.49

.90

.52

.59

.38

Strength

–.63

.70

.37

–1.19

.47

.01

–1.33

.31

.00

–1.53

.47

.00

Interest

.28

.74

.71

–.74

.51

.14

–.45

.39

.25

–.37

.50

.46

Sophistication

–.86

1.00

.39

–.98

.76

.20

–.28

.54

.61

–.23

.70

.74

Consistent

–1.81

1.14

.11

–.79

.74

.29

–.76

.50

.13

–1.20

.68

.08

Conflicting

3.38

1.26

.01

1.37

.98

.16

.51

.75

.50

.88

.88

.32

Constant

–2.02

.99

.04

.55

.55

.32

.24

.39

.53

–.57

.50

.26

Pseudo-R2

.28

.15

.17

.18

N

85

145

314

278

Notes. ANES 1992–1994 Panel. Entries are probit coefficients and standard errors. Dependent variable is coded “1” for citizens who changed partisan identity, and “0” for those who remained stable from 1992 to 1994.

Chapter 8

(p.248)

Table A8.1. Effects of Partisanship and GDP Growth on Retrospections Across Level of Partisan Strength

Variable

Leaning

Weak

Strong

B

SE

p

B

SE

p

B

SE

p

Age

–.02

.02

.35

–.03

.02

.04

–.02

.02

.27

Female

–.03

.01

.00

–.05

.01

.00

–.05

.01

.00

Black

–.05

.01

.00

–.04

.01

.00

–.05

.01

.00

Education

.04

.02

.01

.05

.01

.00

.06

.01

.00

Income

.02

.02

.19

.01

.01

.59

.02

.01

.17

Unemployed

–.04

.02

.03

–.06

.02

.00

–.04

.02

.03

In-party

.01

.02

.59

.03

.02

.08

.03

.02

.11

GDP growth

.02

.01

.19

.03

.01

.00

.03

.01

.01

Interest

–.11

.03

.00

–.04

.02

.15

–.05

.03

.09

Sophistication

–.04

.03

.24

–.03

.02

.24

–.05

.03

.06

Consistent

–.17

.04

.00

–.10

.03

.00

–.10

.03

.00

Conflicting

.19

.05

.00

.10

.04

.02

–.10

.05

.06

In-Party x Int.

.07

.03

.02

.06

.03

.03

.12

.03

.00

In-Party x Soph

–.02

.03

.61

.03

.02

.25

.04

.03

.12

In-Party x Conf.

–.34

.05

.00

–.23

.04

.00

–.21

.05

.00

In-Party x Cons.

.26

.04

.00

.18

.03

.00

.20

.03

.00

GDP x Int.

.02

.01

.01

.00

.01

.48

.00

.01

.75

GDP x Soph

.02

.01

.00

.01

.00

.05

.01

.01

.17

GDP x Conf.

–.01

.01

.35

.01

.01

.46

.05

.01

.00

GDP x Cons.

.01

.01

.14

.01

.01

.27

.00

.01

.98

Constant

.34

.05

.00

.30

.05

.00

.30

.05

.00

SD (constant)

.10

.02

.09

.02

.10

.02

SD (residual)

.23

.00

.22

.00

.23

.00

N

3,498

4,719

4,332

Years

11

11

11

Notes. Entries are restricted maximum likelihood coefficients and standard errors. Model includes random intercepts for year. Dependent variable is economic retrospections, where higher values indicate more positive assessments. GDP, gross domestic product.

(p.249)

Table A8.2. Voting Behavior Dynamics Across Partisan Strength

Variable

Weak

Strong

B

SE

p

B

SE

p

B

SE

p

Age

.13

.24

.59

.04

.22

.87

–.27

.30

.38

Female

–.15

.10

.12

.07

.09

.45

–.17

.12

.18

Black

–1.07

.19

.00

–.89

.17

.00

–.59

.20

.00

Education

.17

.18

.36

.19

.17

.26

.21

.23

.37

Income

–.11

.20

.59

.57

.18

.00

.15

.26

.55

Unemployed

–.32

.29

.26

.09

.27

.76

–.13

.34

.70

Retrospections

.25

.54

.65

1.09

.44

.01

.85

.68

.21

Republican

.64

.28

.02

1.24

.23

.00

2.03

.42

.00

Economic issues

–.80

.73

.28

.48

.56

.39

.24

.94

.80

Social issues

.87

.59

.14

–.02

.43

.96

.30

.67

.66

Interest

–.66

.70

.35

–.86

.62

.17

–.41

.84

.63

Sophistication

–2.35

.61

.00

–1.19

.56

.03

–1.25

.72

.08

Consistent

–2.98

1.10

.01

–.62

.83

.45

–.69

.96

.47

Conflicting

4.31

1.08

.00

2.08

1.04

.05

–3.22

1.70

.06

Rep x Interest

–.45

.41

.26

.28

.36

.43

–.76

.55

.17

Rep x Soph.

1.01

.36

.01

–.20

.33

.55

1.56

.48

.00

Rep x Cons.

5.99

.59

.00

4.93

.54

.00

3.14

.68

.00

Rep x Conf.

–6.47

.60

.00

–4.61

.57

.00

–3.31

.94

.00

Retros x Int.

.78

.75

.30

.55

.65

.39

.79

.87

.36

Retros x Soph

1.13

.70

.11

–.35

.62

.57

–.46

.76

.55

Retros x Consis.

–.30

1.08

.78

.48

.94

.61

–.55

1.08

.61

Retros x Conf.

–2.55

1.13

.03

–.17

1.06

.87

1.03

1.49

.49

Econ x Int.

1.26

1.06

.24

.47

.91

.60

.66

1.24

.60

Econ x Soph.

2.06

.98

.04

1.62

.82

.05

.32

1.01

.75

Econ x Cons.

–1.11

1.68

.51

–3.68

1.34

.01

–1.61

1.58

.31

Econ x Conf.

.94

1.60

.56

.03

1.56

.98

7.85

2.52

.00

Social x Int.

.24

.82

.77

.71

.69

.30

–.44

.88

.62

Social x Soph.

.19

.74

.80

1.17

.63

.06

1.53

.80

.06

Social x Cons.

1.38

1.22

.26

–1.12

1.02

.27

–1.00

1.13

.38

Social x Conf.

–.35

1.17

.76

.23

1.13

.84

2.87

1.72

.10

Constant

–.20

.50

.70

–1.23

.44

.01

–1.31

.70

.06

Pseudo-R2

.49

.49

.81

N

1,381

1,783

2,125

Notes. Entries are restricted maximum likelihood coefficients and standard errors. Model includes random intercepts for year. Dependent variable is economic retrospections, where higher values indicate more positive assessments.

(p.250)