Parties are objects about which beliefs and loyalties, preferences and assessments,
are formed and used. They help lead the voters in making choices rather than being
the objects of choice themselves. And, all of these are particularly American
conceptions.
The authors of the Michigan model, to be sure, sought to develop the compara-
tive extension of their ideas from the beginning. Perhaps the most extensive
example is by Butler and Stokes ( 1974 ), in which they sought to use the ideas of
The American Voter(Campbell, Converse, Miller, and Stokes 1960 ), including
partisanship, to understand British politics. This was, of course, the obvious
natural extension, given its similar continuity of an essentially two-party system
with comparable continuity in stances of the major parties. Of course, the problem
was that Britain diVered from America in being a unitary parliamentary govern-
ment with strict party loyalty, so that voters decided which party to vote for, rather
than which nominee of a major party to support in their riding. To put it
otherwise, voters typically said they voted for the party and not the person, the
exact reverse of the claims of the American voter.
The great theorist of partisanship, Phil Converse, made a strong case for the
general, comparative utility of partisanship, perhaps especially in his classic article,
‘‘Of Time and Partisan Stability’’ ( 1969 ). There, he demonstrated that the concep-
tion of partisanship was helpful for understanding major properties of party
systems, and one might infer back that partisanship in the electorate is a function
of the party system and not ‘‘just’’ of the properties of the parties themselves. He
and Dupeux ( 1962 ) saw a surrogate identiWcation to partisanship, what they called
an ideologically based ‘‘tendence’’ in the then current French system with its diverse
and highly variable cast of political parties contending for votes (see Converse and
Pierce 1986 for a more modern view of French partisanship and voting). This
‘‘stand in’’ for partisanship suggested that voters thrived when they couldWnd
ways to hold matters suYciently constant to provide structure to their conception
of politics. Retrospective voting, for its part, is one of the most migratory of
American-originated conceptions for understanding electoral politics. Fiorina’s
notion of party identiWcation as a running tally has been applied metaphorically,
although rarely in precise ways. The result, often, is a use of the term party
identiWcation or partisanship in comparative contexts, which lack precise and
theoretical speciWcation. Finally, it seems evident that if voters in stable two-party
systems need heuristics to guide them through electoral decision-making, voters in
less stable and/or in multiparty systems would be in far greater need of such
informational short-cuts.
And yet, the concept of party identiWcation did not travel particularly well as,
say, retrospective voting did. The question for here is why? The answer I propose is
not that there is no general value in these ideas. Rather, it is that the American
electoral landscape has a unique conWguration of attributes that highlights
‘‘parties-as-assessments,’’ while in virtually all other systems, political parties are
564 john h. aldrich