Lost in Technology? Political Parties and the Online Campaigns of Constituency Candidates in Germany's Mixed Member Electoral System

ABSTRACT

The Internet is changing the technological context of election campaigns in dramatic ways. The question of the larger impact of this development on the structures and strategies of election campaigns has stimulated two competing hypotheses. The orthodox view perceives the new medium as facilitating centralized election campaigns that allow party headquarters to target and mobilize groups of voters in more efficient and direct ways. A revisionist view stresses the Internet as a means for individual constituency candidates running candidate-centered campaigns at the local level. From this perspective, the Internet has a decentralizing effect on the structure of election campaigns. This article tests both hypotheses on the basis of the German Candidate Study 2005. It looks in particular at the impact of the electoral context on the style of constituency candidates' online campaigns in the German Federal Elections of 2005.

KEYWORDS:

• Candidates
• election campaigns
• electoral systems
• Internet
• political parties
• political representation

This analysis is based on data from the German Candidate Study 2005. I am grateful to the German National Science Foundation (DFG) for funding the field phase of this study (ZI 608/4-1) and to the Mannheim Centre for European Social Research (MZES) for providing institutional context and support. Thomas Gschwend, three anonymous reviewers, and the participants in a workshop at Cornell University on October 24, 2008 provided most helpful comments. My thanks go to Christopher Anderson for convening this workshop.

Notes

1. These concepts differ in certain respects. However, in this article I treat them as synonyms. My analysis focuses on their commonalities rather than their differences, and it does not seek to provide a comprehensive definition and clarification of these terms.

2. A replication data set is available from the author upon request.

3. The realized sample largely represents the population. The evidence for this is as follows: The distribution of candidates by party does not systematically deviate from a theoretically expected uniform distribution (SPD, 18 percent; CDU/CSU, 21 percent; FDP, 20 percent; Greens, 20 percent; Left.PDS, 21 percent). The mean age of the candidates in the realized sample is identical with the mean age in the population (46 years), and even when considering the mode of candidacy, the distribution in the realized sample (35 percent party-list-only candidates; 20 percent district-only candidates) is essentially the same as in the population (37 percent party-list-only candidates; 19 percent district-only candidates). The share of double candidacies in the realized sample is the same as in the population (45 percent). Moreover, the realized sample reflects the ratio of incumbents to non-incumbents in the population in satisfactory ways (7:93 in the sample; 11:39 in the population).

4. I assume that candidates estimate their likelihood of winning on the basis of the results of the previous elections and in light of the margin between the first and the second winner in the previous district vote. I further assume that candidates will calculate this estimate on the basis of a threshold rather than working on the basis of continuous increments. If a candidate won the previous election or lost by a narrow margin, the present candidate is more likely to assume a successful outcome and the incentives for an individualized campaign will increase. If a candidate lost the previous election by a large margin, the present candidate is less likely to assume a successful outcome, and there will be fewer incentives to run an individualized campaign. I define the threshold marking the distinction between winnable and non-winnable districts with a margin of <10 percent and >10 percent. This is a frequently used assumption in the literature. It is used, for example, by Turner (1953) or, for the German case, by Schmitt and Wüst (2004). The New York Times uses this criterion, too, for electoral district predictions.

5. I should add that very few candidates conducted chats or kept a blog without using a personal Web site. These candidates were excluded from the analysis.