single dimension of a potentially N-dimensional spatial model. Legislators’ esti-
mated ideal points, moreover, tend to be extremely stable over time (Poole 1998 ).
Because parties so consistently dominate legislative organization, it is diYcult to
test the extent to which they account for the orderliness of voting patterns. In a pair
of ingenious studies, however, Jenkins ( 1999 , 2000 ) compares voting in the Con-
federate Congress of 1861 – 5 with that in the US Congress during the same era. The
legislatures were similar in formal structure, in membership (many legislators
served in both chambers), and even in the issues on which they voted, but the
Confederate Congress was not organized along party lines, and the voting patterns
of Confederate legislators were far less stable in important ways. First, spatial
models correctly classify fewer votes in the Confederate than the US Congress
(Jenkins 1999 ). Second, Confederate legislators, operating in a party-less environ-
ment, are less stable in their ideological positions over time (Jenkins 2000 ). Overall,
the results suggest that political parties impose order on voting in ways that make
legislative decisions predictable and stable.
Political parties may play this role in general, but even casual observers will
note that not all parties are equivalent. Comparative legislative scholarship has
long made much of the diVerence between strong and weak political parties in
controlling legislative outcomes. Scholarship on the US Congress has been largely
occupied for over a decade with the extent to which the levels of party voting we
observe are due to like-mindedness among co-partisans (cohesiveness) or pressure
from party leaders (discipline) (Krehbiel 1998 ; Cox and Poole 2004 ). Much of the rest
of the legislative world, however, has yet to be mapped at all in terms of party unity in
voting. Factors that may account for relative levels of party unity can be divided
between those that operate at the system level, and are constant across all par-
ties within an assembly (e.g. regime type, federalism, electoral system, regime age),
and those that vary across parties within assemblies (e.g. government vs. opposition,
seat share, ideological composition, party age). Hix ( 2004 ) takes advantage of
the European Parliament’s multinational structure to gain analytical leverage on
the eVect of the electoral system—normally constant within a given legislature—
showing higher voting unity in parties with centralized control over legislators’
election (and reelection) prospects than in those where legislators cultivate personal
support among voters to secure election. Carey ( 2005 a) draws on voting data
from eighteen legislatures to conWrm the conventional distinction between highly
uniWed parties in parliamentary systems and less uniWed ones in presidential
regimes, and shows that governing parties are more uniWed than opposition parties
in the former regime type, but indistinguishable in terms of unity in the latter. Cross-
national analyses of legislative voting remain relatively rare, and mostly limited in
their scope (Morgenstern 2003 ; Noury 2005 ). Recent eVorts by comparative legisla-
tive scholars to archive voting data from across many legislatures in a standard
format will facilitate cross-national research, however, and can be expected to
446 john m. carey