Sorting and the Split-Ticket A ppendix Table A1. Coding Information and Variable Descriptions……………………………….1 Figure A1. Distribution of Sorting Scores……………………………………………………..3 Table A2. Fixed Year Effect Estimates for “Table 1”……………………………………..4 Table A3. Alternative Modeling Specification…………………………………………….…5 Table A4. Out-Party Voting Across House and Presidential Offices...………...……6 Table A5. CCES Tabular Marginal Effect Estimates………………………………..……7 Table A6. Regression Estimates for Effect of “Change” in Sorting over Time…….8 0 Sorting and the Split-Ticket Table A1. Descriptive and Coding Information for Variables Included in Manuscript N Mean Std. dev Mi n Max Split-ticket vote 13636 0.1929 0.3946 0 1 Sorting 18633 0.2569 0.2367 0 1 Out-party incumbent 26318 0.2529 0.4347 0 1 Indifferent 26318 0.3315 0.4707 0 1 Policy-balancing 24263 0.5583 0.1239 0 1 Index of issue position strength 26318 0.4495 0.2527 0 1 26318 0.5024 0.5 0 1 Variable Partisan strength Political knowledge Operationalization 1 = for vote for any combination of Dem-Rep for Presidential and House of Rep., 0 = straight-ticket, Dem-Dem or Rep-Rep, vote Captures overlap and strength of partisan and ideological selfidentification, ranging from 0 to 1; see Figure A1 for full distribution of scores and manuscript for full operationalization Folded partisan identification, where 0 = Independent, 0.33 = leaning partisan, 0.66 = weak partisan, and 1 = strong partisan 1 = incumbent running for House of Representatives is of opposite party as R; otherwise coded 0 1 = R responding that they did not like or dislike anything about the parties; otherwise coded 0 Party balancing = abs(Rideo - Cideo) – abs(Rideo – (GOPideo + Dideo)/2), where: Rideo = respondent’s self-placement within ideological space; Dideo = respondent’s placement of Democratic Party within ideological space; GOPideo = respondent’s placement of Republican Party within ideological space; Cideo = the value of Dideo or GOPideo that most closely approximates the respondents ideological self-placement, Rideo An index of extremity of responses to policy statement, including: government health insurance, aid to blacks, government provision of jobs, defense spending, and abortion. Responses--usually five-part disagree/agree options--are folded so that neutral opinions are low values and strong opinions are high values; simple additive index is then created such that variable ranges from 0 = neutral on all issue positions to 1 = has strong opinion on all issues 1 = correctly identifying which party controlled House of Rep.; otherwise =0 1 2 Table A1. Descriptive and Coding Information for Variables Included in Manuscript continued... Variable N Mean Std. dev Min Max Operationalization Education 26092 0.5082 0.2891 0 1 White 26318 0.5857 0.4926 0 1 Ranges from 0 = less than 8th grade education to 1 = advanced post-bach degree 1 = self-identify as white, 0 = otherwise Black 26318 0.1036 0.3048 0 1 1 = self-identify as black, 0 = otherwise Old South 26318 0.2854 0.4516 0 1 1 = state included in original Confederacy, 0 = otherwise Urban 26318 0.2171 0.4123 0 1 1 = lives in urban area, 0 = otherwise Age 26146 0.361 0.2144 0 1 Individual's age ranges from 0 (17) to 1 (94) Male 26318 0.4481 0.4973 0 1 1 = self-identify as male, 0 = self-identify as woman Income 24251 0.4592 0.2863 0 1 Income quintiles, ranging from 0 (lowest) to 1 (highest) Protestant 26318 0.5522 0.4972 0 1 1 = self-identify as Protestant, 0 = otherwise Notes: Dummy “year” variables not included 3 Sorting and the Split-Ticket Figure A1. Distribution of Sorting Scores 1972-2012 2,750 2,500 (N) Observations 2,250 2,000 1,750 1,500 1,250 1,000 750 500 250 0 16 14 Percent 12 10 8 6 4 2 0 0.000.050.080.090.120.160.190.200.280.310.530.621.00 Source: 1972-2012 Cumulative ANES Time-Series 4 Table A2. Year Fixed Effects Estimates Corresponding to “Table 1” in Manuscript Model 1 b s.e. Model 2 b s.e. 1972 1976 1980 1.672** 1.109** 1.291** 0.137 0.112 0.115 1.632** 0.992** 1.188** 0.126 0.113 0.114 1984 1988 1992 1996 1.079** 1.088** 0.972** 0.607** 0.110 0.107 0.111 0.099 1.025** 1.066** 1.000** 0.620** 0.109 0.106 0.108 0.101 2000 2004 2008 0.663** 0.688** 0.571** 0.105 0.102 0.103 0.688** 0.610** 0.450** 0.103 0.103 0.100 Source: 1972-2012 Cumulative ANES Time-Series Notes: *p<0.05, **p<0.01 5 Table A3. Alternative Modeling Specification with Additional Demographic Variables Sorting Partisan strength Out-party Incumbent Indifferent Balance Issue strength Political knowledge Education White Black Old South Urban Age Male Income Protestant Constatnt Model 1 -2.288** (0.18) ----- Model 2 ----- 1.192** (0.12) 0.197 (0.12) 0.369 -1.417** (0.10) 1.138** (0.09) 0.166 (0.10) 0.755** (0.20) -0.269* (0.12) -0.250** (0.06) 0.159 (0.14) 0.031 (0.09) -0.648** (0.17) 0.437** (0.11) 0.089 (0.08) 0.126 (0.26) 0.004 (0.08) 0.148 (0.16) -0.07 (0.04) -2.427** (0.16) 9,783 (0.17) -0.438** (0.08) -0.214** (0.06) -0.011 (0.11) 0.065 (0.10) -0.483** (0.17) 0.441** (0.11) 0.107 (0.07) 0.094 (0.20) -0.07 (0.07) 0.066 (0.12) -0.146** (0.05) -2.000** (0.14) 11,983 Source: 1972-2012 Cumulative ANES Time-Series Notes: Models are replications of Table 1 in manuscript, save with the addition of extra control variables per reviewers’ request. Substantively, coefficients for main explanatory variables change very little. *p<0.05, **p<0.01 6 Table A4. House vs. President Out-Party Voting? Examining the Composition of the Split-Ticket Out-party defectors within ticketsplitter category 80 70 Percent 60 50 40 30 20 10 0 Congressman President 7 Table A5. Marginal Effects of Partisan Strength and Sorting on Likelihood of Casting Split-Ticket Across Four Ticket Pairings, 2010 CCES Partisan strength Sorting Governor- Governor- Governor- Governor- Average US House US Senate State House State Senator M.E. -0.069 -0.057 -0.069 -0.081 (0.01) (0.01) (0.01) (0.01) -0.122 -0.099 -0.125 -0.139 (0.01) (0.01) (0.01) (0.01) -0.069 -0.121 Source: 2010 CCES Notes: Marginal effect of partisan strength and sorting calculated using respective “Model 1” and “Model 2” in each matching logistic regression model found in Table 2 in manuscript. These effects calculated holding other variables at means; standard errors in parentheses. 8 Table A6. Logistic Regression Estimates for Change in Ticket-Splitting from 1992-1996 b s.e. Sorting increase -1.49** 0.63 Sorting decrease -0.30 0.52 Predicted probability of split-ticket when Male -0.86* 0.50 ….no increase in sorting = 14.0% Education -0.04 0.16 ….increase in sorting Knowledge 0.05 0.26 White -0.66 1.10 Black -0.71 1.33 Policy balance -0.08 0.16 Old South -0.58 0.78 Constant -1.34 1.40 X2 11.34 Pseudo R2 0.07 N 476 Source: 1992-1996 ANES Panel Study Notes: *p<0.05, **p<0.01 9 = 0.04%
© Copyright 2024 Paperzz