Campaign Dynamics: The Race for Governor

Appendix D. Complete Models Examining Votes for Democratic Gubernatorial Candidates

APPENDIX D Complete Models Examining Votes for Democratic Gubernatorial Candidates Appendix D includes the three complete models used to produce tables 9.4 through 9.6.They include control measures for whether a voter was Jewish, the age of the respondent, whether the Democrat or the Republican candidate was an incumbent, the amount spent by each candidate during the primary and gen eral -election campaigns, a measure of the electorate’s political ideology, three separate dummy variables for the three presidential-election years included in this analysis, and a dummy variable for the South, defined as former Confed erate states. Each of these variables is measured as a dummy variable except age, state ideology, and campaign spending. Age is measured as a categorical variable ranging from 1 to 4. The state ideology measure is Erikson, Wright, and McIver’s (1993) measure. Its observed range is from⫺27.93 to ⫺.81, with increasingly negative numbers indicating increasingly more conservative state electorates. Campaign spending is the dollar amount spent, adjusted for infla tion and population. In addition, it is expected that the impact of spending will vary with the incumbency status of the candidates involved. Thus, the two campaign -spending variables are multiplied by whether the respective candidate spending the money is an incumbent.The sample size for each model is 93,858. The behavior of these control variables is lar gely as would be expected and is consistent across all three models. Jewish voters are more likely to vote for Democratic candidates, controlling for the other factors in the model. Older voters tended to support Republican candidates more than did younger voters. Incumbents of both parties tended to have an advantage over challengers. Spending behaved as predicted—spending more money helped candidates win votes, but that benefit was diminished if the candidates were incumbents As state electorates become increasingly liberal, Democratic candidates found it harder to get votes. This apparently counterintuitive finding corresponds to that re ported by Erikson,Wright, and McIver (1993) and likely results from the same phenomenon they describe. Democratic candidates did significantly less wel 188 in 1984 than in presidential of f-year elections, not surprising given Reagan’ s overwhelming presidential victory that year . The Bush election in 1988 and the Clinton victory in 1992 produced coattail effects in the expected direction, but they were not statistically significant. Finall , the South is no longer a safe haven for state-level Democratic candidates. Once other factors, including party identification and race, have been controlled, voters in the South were les likely to cast their ballots for Democratic candidates. Appendixes 189 TABLE D1. Complete Model Used to Produce Table 9.4 Variable Coefficien Constant ⫺0.1216 Democrat 1.3094*** Republican ⫺1.2031*** Family income ⫺0.0732*** Black 0.9416*** Catholic 0.0520 Jewish 0.5429*** Female 0.1132*** Age of respondent ⫺0.0558*** Democratic incumbent 0.4072* Republican incumbent ⫺0.4307* Democrat spending 0.511*** Republican spending ⫺0.0244 Democratic incumbent ⫻ Democratic spending ⫺0.0083 Republican incumbent ⫻ Republican spending 0.0198 State ideology ⫺0.0168** 1984 dummy variable ⫺0.5069** 1988 dummy variable ⫺0.3005 1992 dummy variable 0.1467 South dummy variable ⫺0.3833** N ⫽ 93,858 *p ⬍ .1 **p ⬍ .05 ***p ⬍ .01 190 TABLE D2. Complete Model Used to Produce Table 9.5 Variable Coefficien Constant ⫺0.0363 Democrat 1.2482*** Republican ⫺1.1168*** Family income ⫺0.0544*** Black 0.7383*** Catholic 0.0633 Jewish 0.5199*** Female 0.0773** Age of respondent ⫺0.0602*** Democratic incumbent 0.2974 Republican incumbent ⫺0.7486*** Democrat spending 0.0486*** Republican spending ⫺0.0211 Democratic incumbent ⫻ Democratic spending ⫺0.0091 Republican incumbent ⫻ Republican spending 0.0466 State ideology ⫺0.0163** 1984 dummy variable ⫺0.5760*** 1988 dummy variable ⫺0.0880 1992 dummy variable 0.0181 South dummy variable ⫺0.3371** Gender-related issue(s) stressed 0.3649*** Gender-related issue(s) stressed ⫻ female 0.0594 Partisanship stressed 0.1067 Partisanship stressed ⫻ Democrat 0.1154 Partisanship stressed ⫻ Republican ⫺0.2223** Race-related issue(s) stressed ⫺0.3793*** Race-related issue(s) stressed ⫻ black 0.3597** Abortion stressed ⫺0.1100 Abortion stressed ⫻ Catholic ⫺0.0767 Class-based issues stressed ⫺0.0893 Class-based issues stressed ⫻ family income ⫺0.0662** N ⫽ 93,858 *p ⬍ .1 **p ⬍ .05 ***p ⬍ .01 191 TABLE D3. Complete Model Used to Produce Table 9.6 Variable Coefficien Constant 0.0600 Democrat 1.2772*** Republican ⫺1.1006*** Family income ⫺0.0532*** Black 0.7640*** Catholic 0.0645...