APPENDIX TO CHAPTER 4: SOLICITER GENERAL INFLUENCE AND AGENDA SETTING

Operationalizing the Separation of Powers

In note 16 in Chapter 4, we describe the ongoing debate over the legislative decision-making process, which directly affects our ability to operationalize any separation-of-powers variable. The key disagreement in the literature stems from whose preferences within the legislature are pivotal. At present, four viable candidates exist: the Senate filibuster pivot and the veto override pivots (Krehbiel 1998), the chamber median (Riker 1962; Krehbiel 1995), party gatekeepers (Cox and McCubbins 2005), and the judiciary committee (see, generally, Smith 1989). An additional complication stems from debate over whether to include the president in the latter three of these spatial models. Some studies include the president (Owens 2010), while others do not (Harvey and Friedman 2006).

Lacking private information about which of these constitutes the “right” model, we take an approach of complete agnosticism while conducting our data analysis. In particular, we reestimated our underlying model using each of the unique combinations of legislative preferences and inclusion of the president (the filibuster–veto pivots perspective assumes the presence of the president). We then used these parameter estimates to simulate marginal effects. Effect estimates in hand, we then simply pooled all sets of simulation results together before computing the percentile values. The Constrained Court variable never achieves statistical significance in the expected direction. Accordingly, our 95 percent confidence interval for the marginal effect of the Court being constrained on the Office of the Solicitor General's (OSG's) influence is very wide (i.e., [-0.43, 0.07]).