and new parties to achieve some policy objectives, as well as publicizing themselves.
This procedure also allows other groups and indeed spontaneous organizations of
electors themselves to exert inXuence 4.
However, we should avoid making too sharp a contrast between referendums and
initiatives, since there are important distinctions to be made within both categories
(cf. Bowler and Donovan 2001 , 128 – 9 ). Referendums mandated by the constitu-
tion—especially when this provision is interpreted by the courts—fall outwith the
control of government and may occur at very inconvenient times from their point of
view. Some initiatives on the other hand are not conclusive—like the Italian
‘‘referendum abrogativo’’ (see footnote 3 )—and theWnal decision has to be made
by parliament (even if it broadly conforms to majority preferences).
There is thus considerable variation in terms of government control over voting.
The ability to set the date of a referendum favors governing parties. Usually this
implies also that they have full control of the wording of the question to be put to
the vote and may also slant this to favor their side, as well as campaigning
vociferously for it. Referendums of this nature are most common under the
‘‘Westminster model’’ where constitutions commonly do not have any provision
for popular consultations other than general elections, so that the referendum takes
the form of a special dispensation by the government to allow a popular vote.
Where the constitution makes explicit reference to the need for a referendum on
certain policy matters (often constitutional change itself) the process can be
initiated by the government—in which case it is still largely in its discretion
whether it wants to pursue the question or not. Where however the process is
triggered automatically by constitutional provisions (usually in this case supervised
by courts) voting may occur at an awkward time for government parties and their
control is diminished. Again this is particularly true in Switzerland where practic-
ally all matters of foreign policy have to be put to a vote: Swiss cantons and certain
US states also have wide provisions forcing votes onWscal or revenue matters.
Such eVects are of course intensiWed where popular votes can be prompted by
popular petition. While this is inconvenient for governing parties in whatever form
it exists, it may be particularly so where by convention or rule the process can be
started by parties. Opposition parties generally use this opportunity to embarrass
governments while new parties use it to promote their own causes and themselves.
In any case the issues are likely either to be oVthe government agenda or to call
into question established policy.
4 Most empirical research is spurred by a general concern toWnd out whether the postulated
negative eVects of direct democracy actually appear when it is practiced. Generally they do not seem
to, though researchers have hedged this conclusion around with qualiWcations about what might
occur in the long term. They have also discerned some problematic eVects for established political
parties. The theoretical consequences of destructured forms of voting are covered in Arrow 1951 ,
McKelvey 1979 , McLean 1989 , and—from the opposed point of view that structure is rarely absent—in
Niemi 1969. Grofman and Feld 1988 provide a fascinating account of a mathematical theorem which
might lead us to the belief that unstructured popular majority voting on policy produces the best
results.
direct democracy 603