judgments and condition the impact of other deciding factors, such as partisanship,
ideology and demographics.
To explore this hypothesis, this paper examines the impact of political information
on patterns of voting behavior in referendums on European integration. It is generally
recognized that individual voters differ greatly in their ability and incentives to gather and
understand political information and that this affects opinion formation processes (Zaller
1992; Converse 2000). Moreover, the opportunity for political learning also depends on the
type of information available in the political environment. Following on from this, in this
paper variation in political information will be examined at two levels: contextual and
individual. At the contextual level, the impact of the campaign setting is examined. This
paper suggests that intensive referendum campaigns provide a favorable informational
environment that encourages citizens to absorb and process more information and
consequently rely on more sophisticated decision criteria when deciding on the ballot
proposal. This implies that in high salience campaigns voters are more likely to rely on
their attitudes towards the EU than in a lower intensity campaign environment. At the
individual level, it is argued that voters’ level of political awareness will act as an
intervening factor in determining voting behavior. This paper does not contend that
differences in political awareness as a rule make voters more (or less) likely to vote yes (or
no) in referendums, but rather that political awareness acts as a mediating factor that has an
impact on the relative importance of other factors determining voting behavior
To begin with, this paper presents a brief overview of the scholarly debate of voting
behavior in EU referendums. Thereafter, the impact of political information on opinion
formation is discussed, and I propose a model of how campaign salience and political
information affect voting behavior in EU referendums. The theoretical propositions are
tested in an analysis of voting behavior in referendums in Denmark, Ireland and Norway.
By estimating a logit model with interaction terms using data from post-referendum
surveys, the relative importance of EU attitudes, partisanship and government satisfaction
and the conditioning impact of information are evaluated. The impact of political
awareness on patterns of voting behavior is also assessed by subdividing the samples into
three strata of political awareness (low, medium, high) and comparing the expected
changes in voting behavior given simulated changes in each of the explanatory variables
across these strata. Finally, the variation in aggregate voting patterns is correlated to
3
Convention