An Integrative Approach to Modeling Presidential Success in Congress

Abstract

We propose and test an integrative multivariate model predicting aggregate annual presidential success on House and Senate roll calls from 1953 to 1994. We incorporate presidency-centered, Congress-centered, agenda, and timing variables in our analysis. Our results show that there are important similarities and dissimilarities across legislative chambers. For both chambers, the percentage of co-partisans of the president and when they came into office are important. A presidential honeymoon with Congress, however, is evident in the House of Representatives, but not in the Senate. Also, the House equation does a substantially better job of explaining the variation in presidential success than does the Senate equation. We argue that an integrative approach offers a more satisfactory statistical and substantive explanation of presidential success than an approach based on asingleperspective. The presidency-centered view of presidential-congressional relations emphasizes the importance of presidential skills, popularity, and personality in determining presidents' legislative success (Barber 1993; Hager and Sullivan 1994; Kellerman 1984; Neustadt 1960; Sullivan 1991). Sullivan (1991), for example, finds that presidents have the ability to increase their initial levels of support in Congress. Moreover, he finds that presidential popularity influences the presidents' ability to sway members from their initial positions. A view competes with this perspective, which highlights the partisan and ideological composition of Congress as the primary determinant of legislative victory or defeat (Bond and Fleisher 1990; Edwards 1989; Oppenheimer 1993). The acceptance of the Congress-centered view is based on the dominance of party affiliation in individual-level regression models predicting presidential support, and of chamber party composition in aggregate-level models of presidential agreement or success in getting legislation passed; in contrast, presidency centered variables (such as the president's Gallup approval ratings or perceived levels of legislative skill) have weak and inconsistent effects in such models. Obviously, neither perspective should be overemphasized, and some scholars have called for a more balanced, tandem perspective (Jones 1994; Peterson 1990). According to this point of view, it is rare for either Congress or the president to win or lose a legislative battle unambiguously, and each player has a significant role in determining the outcome. These studies tend to combine the quantitative approach emphasized by the congress-centered literature with the qualitative approach stressed by the presidency-centered scholars. In recent years, political scientists have introduced yet another way to view presidential congressional relations. They suggest looking beyond the participants in the legislative process and examining the nature of the process itself. This perspective-which we might call the structural perspectiveincludes a variety of factors. There are temporal factors, such as the trends and cycles in presidential influence and the election calendar. Presidents, for example, might be advantaged by an influx of new members who are elected along with the president. These new members might be especially loyal to the wishes of the president. After an intervening election has passed, however, the impetus of new members of the president's party to be especially supportive might dissipate. Institutional factors, such as the differences in rules across legislative chambers, might contribute to an explanation of presidential success. Last, agenda factors, such as whether the issue under consideration is foreign or domestic policy or if the issue is one favored or initiated by the president as opposed to wholly independent congressional proposals might play a role. While some of these variables have been examined in earlier work on the presidency, the research that specifically employs this perspective is of reasonably recent vintage (for example, Brace and Hinckley 1992; Covington, Wrighton, and Kinney 1995; Jones 1994; Light 1989; Skorownek 1993; Sullivan 1991). …