objects of actual choices, not just the basis for making assessments. Thus,
the continuity of parties combined with a lack of rigorous party discipline in the
legislature means thatchoicesare and must be over candidates and not over parties.
This is narrowly so, as in many systems votes are cast for political parties and not
individual candidates, but it is also true more metaphorically. In Britain or other
Anglo-American two-party systems, votes can (oftenmust) be cast for individual
candidates, but high party discipline dilutes the personal name-brand value any
candidate may have, something of high value in the USA, and accentuates the value
of the name brand of the party. As a result, vote decisions made in the name of a
party naturally trump assessments of individual candidates, and that is reXected in
responses to party-identiWcation-like questions on an election survey. It does not
follow from a concept being hard to measure that the concept is not relevant in
those systems. It only follows that the concept is obscured—explaining, perhaps,
why Converse couldWnd the very abstract patterns so striking in the very same
political systems where the micro-measures were diYcult to observe.
The above is inferential. Historical evidence in America seems consistent with
this set of claims. Voting in eighteenth century America was highly partisan, indeed
as strongly so as in contemporary parliamentary systems. Historians of American
elections naturally and correctly point to the form of ballot—non-secret voting,
ballots made by the separate parties, etc.—and their interaction with institutions,
notably partisan machines, to explain highly partisan elections (e.g. Hays 1980 ).
While the move from open to secret balloting and other technical features of the
voting process are important parts of the explanation of the decline of partisan
elections in the USA, it is by now well understood that intervening between ballot
reform and candidate-centered elections was the development of the individual
oYce seeking motivation that these reforms and others made possible (Katz and
Sala 1996 ; Price 1975 ). Thus, it was the increasingly candidate-centered campaigns
of the late nineteenth and early twentieth centuries that generated theWrst level of
decline in partisan elections, followed by the new technology of mid- to late-
twentieth century politics thatWnalized the candidate-centered campaign as all
but fully replacing the party-centered contests of the earlier era (Aldrich 1995 ). In
short, the voters were responding to the possibilities of the electoral setting and
especially to the nature of the campaigns they observed in generating Wrst
highly partisan and then highly candidate centered voting. Perhaps were party
identiWcation questions asked in nineteenth-century America, they would
have been understood as asking vote intention.
In a comparative context, the above argument is also inferential. Several empir-
ical observations might test the notion. For example, as party discipline is tending
to erode in many nations’ parliaments, those with single-member districts com-
bined with durable parties dominating the system, or other systems (e.g. Japan)
where candidate names have some value, the importance of party-as-assessment
should be increasing, while as party discipline increases in the USA,
political parties in and out of legislatures 565