the laws and practices that excluded blacks and poor whites from politics). The Jim
Crow South was, however, also set within a functioning democracy with an
established, durable two-party system at the national level. While this ‘‘natural
experiment’’ happened to be found in the USA, he oVers no reason, nor can I think
of oneWfty years later (Aldrich 2000 ), that makes his contrast less than fully general.
The result of the experiment was clear, clean, and simple to convey. Politics was a
perfectly reasonable real-world approximation of democracy as imagined in
theory when found within an established and durable party system. Politics was
extraordinarily undemocratic in the South, that is, it was undemocratic when not
embedded in a competitive and durable party system, and Key was scathing in his
description of the choices, such as they were, confronting voters.
The question of this section, then, is what role does the party play in furthering
electoral democracy? Of the myriad aspects of parties-in-the-electorate, the core
questions are ‘‘What does the party mean to potential and actual voters?’’ and
‘‘How does that meaning help shape their political decisions?’’ Here, I therefore
address that core pair of questions.
TheWrst question opens an apparent case of American exceptionalism, in that
the theoretical understandings of party identiWcation developed in the context of
American survey research, are distinct and possibly theoretically unique to the
USA. I suggest here that such a conclusion may be premature. The claim is that, if
we can parse out the contemporaneous context ofvotingfor, rather than assessing
of, political parties, we mayWnd beliefs akin to American party identiWcation.
Campbell, Converse, Miller, and Stokes’ classic accounts ( 1960 , 1966 ) conceived
party identiWcation as an early-formed, durable, aVectively-based loyalty to a
political party. Their data showed that this conception was consistent with the
beliefs and attitudes of a substantial majority in the American electorate, both in
the 1950 s and 1960 s as they developed their theory, and again in recent years, as the
(actually rather modest) attenuation of partisanship in the 1970 s resurged to
roughly the earlier levels (Bartels 2000 ). The key point was that this notion of
partisan identiWcation was relevant for understanding how ordinary citizens, with
typically marginal interest in politics, were able to negotiate the complicated
political world. This aVect-centered view held that most people began with a bias
in favor of their favored party (childhood socialization), they tend to hear things in
a way biased toward their party (selective perception), and they are likely to further
that bias even more by consuming information from sources that are themselves in
favor of the citizen’s preferred party (selective attention). Thus reinforced, partisan
loyalty means that it is hard to change the minds of supporters of the opposing
party, more so than it is to win over independents and apolitical citizens. In turn, it
is harder to woo the uncommitted than to cement those already predisposed in
one’s favor.
An alternative view is due to Downs ( 1957 ), Key ( 1966 ), and Fiorina ( 1981 ). This
view is of a more cognitively-based assessment. It assumes that voting and the
562 john h. aldrich