minded oYceholders would overshoot in this ‘‘macro polity.’’ Two likely possibilities
are that the politicians are personally more extreme in their policy beliefs than the
public or that these politicians need resources from relative extreme partisan and
interests groups for renomination and reelection. Of course, since many politicians
were themselves once policy activists, both might be true. Further the public, in the
aggregate, is generally moderate (indeed may literally be the deWnition of moderate)
and so these activists may be only modestly ‘‘extreme.’’ But, to answer the question of
why party resources come from relative extremists is to ask a question about the
party organization, a subject we will touch on in the conclusion. For now, note that
Wndings that party activists are more extreme than the partisan identiWers in the
electorate is not unique to the USA. It holds in many nations, and is one basis to
begin to develop that general account that places the USA at one end and multiparty
parliaments at the other end of the same continuum.
Whereas the traditional question asked of American legislative parties is whether
they are ever united, the archetypical multiparty parliamentWnds the political
party almost invariably united. Indeed, parties are often the unit of analysis, in
virtual atom-as-billiard-ball fashion, rather than the American counterpart of
party as atom-as-mostly-empty-space. In this tradition, the primary question is
how parties form, maintain, or disband coalition governments, with the govern-
ment and its ministers choosing policies for the parliament to ratify with strict
party line voting. As Diermeier and Feddersen demonstrate ( 1998 ), the power of
the no-conWdence vote forces at least the parties in government into unity. It is
only recently that the atom has been broken open, as it were, and non-lock-step
unanimous behavior of party politicians considered.
The multiparty parliament inserts the extra step of government formation in the
democratic crossroads of going from citizen preferences to policy (even when one
party, majority or minority, see Strom, 1990 , ends up forming the government).
Technically, this is true in the US House, too, as itsWrst action is to select from its
own internal government by choosing a Speaker and a committee structure. All but
invariably, that vote is also a strictly party-line vote, just as in, say, Britain. Perhaps
the lack of a no-conWdence vote in the Speaker undermines primarily party-line
voting.
A substantial literature has sought to understand coalition formation in
multiparty parliaments based on policy preferences. Thus, one beginning point
would be with applications of Riker’s ( 1962 ) minimal winning coalition hypothesis,
with quite mixed empirical results. Axelrod ( 1970 ) added policy considerations
per se by modifying minimal winning to ‘‘minimal winning connected coalition,’’
and by ‘‘connected’’ he meant stand close or adjacent to each other on policy/
ideology. The empiricalWndings were improved but still mixed. Then, Laver and
SchoWeld ( 1990 ; see also Laver and Budge 1992 ) and Laver and Shepsle ( 1994 , 1996 )
applied insights from social choice theory. In theWrst, SchoWeld developed a multi-
dimensional analogue to the centrality of policy in the one-dimensional, median
political parties in and out of legislatures 569