European elections are often said to be second-order in character (Reif and Schmitt, 1980). This means that most electors consider the European political arena to be less important than the national one and that they, accordingly, use their votes in EP elections to express feelings of satisfaction or dissatisfaction with domestic parties or to bring about political change in their own country. Amongst the implications of this hypothesis are the following:
Important consequences for the party politics of the EP follow from the weakness of the relationship between what MEPs actually do there and their chances of reelection. Most obviously, it attenuates the political responsibility of MEPs to the electorate as a whole, and deepens their dependence on national parties as selectorates. On the other hand, the second-order pattern both facilitates the construction and operation of transnational party groups in the EP and detracts from any successes in this regard. Cohesion is easier to achieve where elite-level agreements in the EP are unlikely to affect the electoral chances of any national party, even in Euro-elections. Yet, in a sense, the capacity of the EP groups to build a transnational consensus -and sustain it across the several issues that make up a parliamentary term - will not really be tested until this involves the allocation of real electoral costs and risks between national parties.
One obvious question to ask is why political parties do not do more to break the second-order pattern of European elections, most obviously by organising themselves on a more European basis and making their transnational links more explicit. The probable answer is that the second-order pattern has a certain self-perpetuating logic. The irony of European elections is that while they have had only limited effects on the political development of the Union itself, they have had the unintended consequence of reducing predictability and increasing risk in the real heardand of political parties: the national arena. By agreeing to European elections - and by deciding to organise them around the member states - governments inadvertently gave their publics an added opportunity to comment on their mid-term performance. The fact that these are nationwide elections -in most cases the only nationwide polls between general elections -makes it impossible to dismiss popular verdicts too lighdy. On other hand, the absence of a direct threat of a change of power at national level means that governments submit themselves to public judgements that are more than usually volatile. In the four European elections since 1979, outcomes have affected changes of government, cabinets and party leaderships. They have also influenced the entry of new parties to the field of serious political contestation and even limited realignments of national parties. The 1984 election was crucial to the rise of the Front National in France, incidentally as much through the public platform provided by representation in the EP as through the momentum created in domestic politics (Marcus, 1995)
Against such a background, domestic parties have litde choice but to compete for as many votes as they can secure. This means that they must field candidates under national party labels that have 'name recognition' and benefit from high levels of inertial or habitual voting; and also that they have to campaign on the issues that are most likely to govern voter choice, which opinion surveys repeatedly show to be mainly national in character. Yet with each European election that is fought on domestic lines, the political parties miss opportunities to aid the development of a 'European public opinion' or a 'European public arena'. There is litde evidence, therefore, of what David Held calls a 'developmental democracy', in which each election improves the public's understanding of how the EU's political system works, so enrichening the possibilities for public debate over time (Held, 1987). This is a shame, because the creation of a demos is a far more demanding task in a transnational political system than in a national one.