the vote. We expect members who are in an electorally precarious situation to be less likely to
The models also include a variable to control for
the member’s revealed preferences using 1st dimension DW-Nominate coordinates, which range
from -1 (liberal) to 1 (conservative) (McCarty, Poole, and Rosenthal 1997). Finally, each model
controls for the type of vote (amendment, final passage, special rule); Cox and Poole (2002)
demonstrate that parties have their greatest sway on procedural votes, while Roberts and Smith
(2003) show that party defections most often occur on amendment votes. Each model further
controls for Congress-specific fixed effects.
Empirical Results
Although we are primarily interested in understanding the individual-level manifestations
of cross-pressuring, we turn first to a discussion of the relationship between majority party rolls
and the votes that interest groups choose to score. As Table 1 demonstrates, the roll rate for the
majority party in the House of Representatives averaged 9.22 percent across the entirety of our
period of study (1991-2000).
regularly get rolled (although we think that nearly once out of every ten votes is a non-trivial
number of instances). However, a look at Table 1 reveals that when even a single group scores a
vote, the majority party’s roll rate increases to 22.5 percent—more than once every five votes—
on average. When two or more groups scored a vote, the majority party’s roll rate increased to
one time in every four.
– Table 1 about here –
The results reported in Table 1 also point out the differences in roll rates across the two
parties. The baseline roll rate for the Democrats and Republicans are similar, but rolls on scored
19
However, we do not expect all groups to have the same effect on all members. For example, a Democrat from a
competitive district is unlikely to find a perfect score from the ADA to be an electoral asset.
20
If we exclude votes that were not scored by groups in our study the roll rate drops to 8.10 percent.
17
Convention