Romain Lachat and Pascal Sciarini
The stability of voting behaviour was a central finding of the pioneering studies of the 1950s and 1960s (Berelson et al. 1954; Campbell et al. 1960; Lazarsfeld et al. 1968 ). According to the classics, voting was to be explained on the basis of long-term factors, such as one's position in the social structure, or traditional loyalties acquired through socialization. In line with this, the electorate was said to be largely immune from the short-term influence of campaign activities, which could only contribute to a reinforcement of predetermined intentions.
This traditional view has not remained unchallenged, however, and the issue of campaign effects has attracted increased interest in the scholarly community. The main reasons for this renewed interest are twofold. First, the weakening of the traditional ties with parties that has occurred since the 1960s (see, for example, Dalton and Wattenberg 2000) has not only reduced the impact of party identification on electoral choice, but has also led citizens to rely more heavily on information delivered during the electoral campaigns when making up their minds. Second, the professionalization of campaign activities, together with the empowerment of the mass media, has created new opportunities for campaign influence (Flanagan and Dalton 1984; Mancini and Swanson 1996). As a result, individual voting choices are both less predictable and more prone to short-term changes.
Having said that, the issue of campaign effects remains, of course, hotly debated. Both the advocates of the 'minimal effects' model and those who claim that campaigns can have substantial effects on opinion formation have provided convincing evidence to support their view. In our opinion, this mixed picture-and the resulting controversy regarding campaign effects-is mainly due to two shortcomings of the literature in the field.
First, several studies are based on a homogeneous view of the electorate. Yet not all voters are likely to be influenced by campaign information, at least not to the same extent. A crucial difference regards whether a voter identifies with a party, or not. While traditional ties with parties have weakened over time, many voters still hold a party identification, and are likely to vote accordingly. Thus party identifiers are expected to be fairly immune from the influence of the electoral campaign. The same holds, more