Circumstances dictate the kinds of power the majority will wield. When there were
fewer collective interests to serve, the party was consequently less important to
citizens, and it was better to be discrete in its use of power, typically by the use of
negative agenda control. As polarization has led to more common interests and the
party has thus become more important to the public, then it is increasingly
satisfactory to exert more positive control over the agenda, to pass majority-
preferred policies.
The third view is what Aldrich and Rohde call ‘‘conditional party government’’
( 1997 – 98 , 2000 ). If, they assert, there is variation in the inXuence of party in
Congress, then we should investigate conditions under which it is at a higher and
at a lower level. They argue that it is at a higher level when the electorate has
selected members of the majority party with more homogenous preferences than at
other times, and with preferences that are more clearly diVerentiated from the
(typically also more homogenously distributed) preferences of the minority party.
There is more for the majority party to win by acting together. And, when party
preferences are more homogenous, there is less risk for the individual member of
ceding authority to the party leadership than when their party is more heteroge-
neous. This view diVers from that of, say, Krehbiel by virtue of its conditionality.
That is, they agree that the electorate is the driving force. They diVer in the role of
the party in Congress. According to Krehbiel, it is epiphenomenal. In conditional
party government, it magniWes the eVect of the constituency at high levels, but not
at low levels. It diVers from the ‘‘party cartel’’ argument of Cox and McCubbins, if
at all, by virtue of the latter’s emphasis on the importance of negative agenda
control (that is, blocking legislation from coming to a vote) when the majority
party is heterogeneous, and emphasis on positive agenda control (that is seeking to
pass legislation favored by the majority) otherwise. The conditional party govern-
ment argument is simply that, instead of negative agenda control, a divided
majority party exerts little control at all.
All three accounts argue that the driving force for explaining the observation of
variation in levels of party voting in Congress are due to changes in the preferences of
the electorate. Missing from all three accounts is an explanation of how and why
those preferences change. Erikson, MacKuen, and Stimson ( 2002 ) argue that there is
a thermostatic relationship or feedback between what the government does and how
the electorate’s partisan preferences and voting choices react. In particular, theyWnd
that the majority party tends to overshoot what the public wants, the Democratic
Party acts too liberally, and a Republican Party too conservatively when in the
majority. As a result, the public shifts back in the direction favored by the minority
party, helping them work toward achieving majority status. Like a pendulum, parties
in Congress sweep left and right farther than voters prefer and the public serves as
counterweight, pulling the overly extreme policy choices back toward what the
public as a whole desires. These propositions are new and still only lightly tested
but seem both plausibly descriptive and enticing. The question then is why reelection
568 john h. aldrich