8. One D Is Not Enough: Measuring Conditional Party Government, 1887-2002

What explains party unity? A test of competing theories

“…To test if the various theories apply similarly to both parties, I run separate models for each party. The dependent variable used in this analysis is the average party unity score for each party per congressional term, the same variable used in previous studies of party unity (Aldrich et al, 2007;Koger and Lebo, 2017;Lebo et al, 2007;Rohde, 1991). As discussed earlier, congressional term is the most appropriate unit of analysis because the institutional changes that would give party leaders greater ability to unite their party will most likely occur at the beginning of a congressional term.…”

“…One theory that accounts for both effects is Conditional Party Government (CPG). This theory is best outlined by David Rohde (1991; see also Aldrich, 1995; Aldrich et al, 2007). Rohde argues that the effects of the majority party leadership on voting behavior in Congress—and ultimately party unity—is dependent upon the degree of heterogeneity of members’ personal and constituency preferences (see also Brady, 1972; Sinclair Deckard, 1976).…”

“…One problem with many studies examining party unity is they do not examine the two parties separately. Instead, they focus either on one party (Rohde, 1991) or examine party unity using a majority and minority divide (Aldrich et al, 2007). Neither of these approaches allow us to determine if theories better explain one party's level of unity.…”

“…While several theories have been proposed to explain party unity in Congress, most studies of party unity assume at both the theoretical and analysis level that the same factors affect each party’s level of unity. Many studies of party unity do not even look at the parties separately, but either only focuses on one party (Rohode, 1991) or examines them using a majority and minority party divide (Aldrich et al, 2007). However, this assumption limits our ability to understand the sources of party unity.…”

“…5.As will be discussed later, the fact the issues debated changes is why it is important to account for both DW-NOMINATE dimensions (see Aldrich et al 2007). …”