Presidential Capital and the Supreme Court Confirmation Process
tion is made. First, we include a measure of the president’s public approval rating at the time of the nomination. Second, we include a measure of qualifications for each nominee to account for the fact that presidents may find it easier to sell highly qualified nominees by going public.22 Finally, we include a measure of the number of years a president has left in his current term to control for “honey- moon” or “lame-duck” effects (Massaro 1990).23
Table 2 presents the results of the GEC model with the number of references to political capital as the dependent variable.24 Parameter estimates for our three ideological distance variables are in the hypothesized direction and statistically significant. These results indicate that as the distance between the nominee and the pivotal Senator increases, the president is more likely to invoke public state- ments on behalf of a nominee. The same relationship holds for the distance between the Court median and the pivotal Senator, which suggests that presidents are more likely to use political capital when trying to sell a nominee who will not move the Court closer to the pivotal Senator’s preferred position. Finally, as the distance between the president’s own ideology and the pivotal Senator increases, the president is more likely to make public statements in support of a nominee. Thus, as we expected, the ideological alignment of the key institutional players has a clear effect on the presidents’ propensity to go public during this process.25
Our statistical findings also have clear substantive meaning. Holding all other variables at their mean value, the predicted number of public statements that
22 Qualifications are based on scores derived from the Supreme Court Compendium (Epstein et al. 1996, 329). Like Segal/Cover Scores that measure the ideology of Supreme Court nominees, the qual- ifications measure is based on a content analysis of editorial judgments about the nominees in four major newspapers (two liberal leaning papers—the New York Times and the Washington Post, and two that lean conservative—The Chicago Tribune and the Los Angeles Times).
23 Descriptive statistics for each of these variables can be accessed in Appendix 1 (available at www.journalofpolitics.org).
24 In Appendix 2 (available at www.journalofpolitics.org), we report the findings for the GEC model without the outlier (Bork), as well as the results of negative binomial regression models with and without the Bork nomination included. We also provide results using an alternative measure for ide- ology—the Bailey/Chang scores discussed in footnote 19. We do so to demonstrate the robustness of the findings even without the outlying observation included in the model, and when using alternative modeling or measurement techniques.
25 The reader could argue that these variables do not fully capture the conditions under which the president would go public. Indeed, it is possible that the ideological distance variables do not actu- ally measure whether a nominee is in trouble in the Senate. To test this hypothesis we ran our model with an additional variable: the predicted number of “no” votes the president could expect in the Senate (Segal and Spaeth 2002, 216). Doing so does not change our results—in fact the revised model produces even more precise estimates for our key variables, while the new control variable does not reach an acceptable level of statistical significance (results available at www.journalofpolitics.org). In short, even when controlling for this plausible alternative explanation our model holds. We do not include this variable in the reported model due to the loss of two observations (Clark and Minton) and because we believe that our model in Table 2 captures the essential dynamics of the process.