Timothy R. Johnson and Jason M. Roberts
outside the spatial model in order to win a confirmation (Moraski and Shipan 1999). Thus, we hypothesize that when presidents are ideologically distant from the Senate they are more likely to “go public” in order to win confirmation for a Supreme Court nominee.11
We do not argue that these ideological relationships alone explain presidential and senatorial behavior on Supreme Court nominees. For instance, many of the Senate arguments against President Johnson’s attempted elevation of Abe Fortas to Chief Justice surrounded the timing of the nomination (Krutz, Fleisher, and Bond 1998). As Senator Griffin (R-MI) observed, “Never before has there been such obvious political maneuvering to create a vacancy so that a ‘lame duck’ President can fill it . . .”12 In fact, we are quite certain that, in many instances, factors outside the spatial model such as gender (O’Connor), race (Thomas), and religion (Ginsburg, Scalia) may have affected the president’s choice of nominee as well as the Senate’s reaction to this choice. However, the extent to which we can demonstrate aggregate patterns in spite of other factors only enhances the robustness of our theory.
Finally, we hypothesize that if presidents go public to help their nominee, their efforts should have an effect on the process—that is, going public should help their nominees win confirmation. This final prediction derives from the analysis throughout the previous section, as well as from the more general nomination and confirmation literature. We know, empirically, that presidents can help their nom- inees by publicly touting them during their initial nomination speeches (Krutz, Fleisher, and Bond 1998), and that success in critical nominations is tied to pres- idential resources (Ruckman 1993). It follows, then, that going public should increase the success rate of Supreme Court nominees.
Data and Methods
To test our hypotheses we analyze all presidential addresses and press confer- ences from the time a nomination is made until the Senate takes action for all Supreme Court nominees between 1949 and 1994. We code every sentence of presidential public statements, and every answer presidents give to questions from the press, that pertains to the Supreme Court during this time period.13 Appendix
11 Throughout the paper, distance is operationalized as the absolute value of the difference in 1st dimension DW-NOMINATE coordinates (Poole and Rosenthal 1997).
12 Quoted in Congressional Quarterly Weekly Report 1968 (1834). 13 There are many potential operationalizations of this variable beyond the one we employ here. For instance, one could count speeches rather than statements. However, we believe our method is an appropriate strategy because we are interested in the exact amount and nature of support given to a nominee. In other words, we are interested in the information contained in the presidential speeches, rather than the simple fact that a nominee is mentioned. Additionally, note that our categories are mutually exclusive. This means that we do not double count statements. We do, however, count when presidents make several different statements to support their nominee in one speech (although this rarely happens). Finally, note that we estimated the same model as we present in Table 2, but substi- tuted the number of speeches that support a nominee (excluding press conferences) for the total