The Mediatization of E-Campaigning: Evidence From German Party Websites in State, National, and European Parliamentary Elections 2002–2009



The rise of e-campaigning is often associated with its ability to circumvent journalistic principles of news selection and presentation. By this, parties and candidates are said to free themselves from the discretionary power of the mass media and to reach voters in an unfiltered way. This conventional wisdom is tested through a comparative content analysis of German party websites in state, national, and European parliamentary elections between 2002 and 2009. The results show that e-campaigns in all elections adhere in their messages to the media logic. Specifically, they replicate those patterns of offline coverage that have been held accountable for rising political alienation and civic apathy. Moreover, the mediatization of German e-campaigning grows over time.


The rise of e-campaigning is often associated with its ability to circumvent journalistic principles of news selection and presentation (e.g., Bimber & Davis, 2003). Due to the technical qualities of online communication (e.g., ubiquity, capacity, topicality), political actors are said to free themselves from the discretionary power of the mass media and to reach voters in an unfiltered way. This kind of “disintermediation”1—as described by Coleman (2005, p. 182)—is seen as both an opportunity and a challenge to democracy (cf. Coleman & Blumler, 2009):

From an optimistic point of view, the Internet might allow parties and candidates to provide citizens with more substantial information about the campaign (cf. e.g., Blumler & Gurevitch, 2001; Gibson & Ward, 2000). In particular, they could overcome those patterns of election coverage that have been held accountable for growing political alienation and civic apathy in the public (e.g., Selnow, 1998, p. 191), i.e. strategic horse-race depictions, shrinking sound bite news, and extensive negativism (cf. for an overview Kaid & Strömbäck, 2008). In this scenario, the World Wide Web would put an end to the mediatization of offline politics (cf. Schulz, 2004, p. 95) and would allow for an enriched campaign environment (see also Blumler, 2009). This could foster a strong democracy (cf. Barber, 1984).

From a pessimistic standpoint, however, political actors might use the Internet just as another means of self-communication (cf. Castells, 2009). In this way, they could continue to rely on traditional tactics in cyberspace that have been proven successful in real-world politics (cf. Margolis & Resnick, 2000). This includes a deliberate adaptation to journalistic needs and interests so as to secure public attention. As a consequence, the presentational opportunities on the web could be left untouched in favor of a traditional “politics as usual” approach (ibid.) that reflects the “impoverished mainstream patterns of political communication” (Blumler, 2009, p. 14) from the offline world. This so-called “normalization” of e-campaigning has indeed been observed for several web practices in the past (cf. e.g., Druckman, Kifer, & Parkin, 2010; Schweitzer, 2008, 2010; Xenos & Foot, 2005). It is an open empirical question, though, whether the convergence with the offline domain also encompasses a purposive mediatization of online political communication (cf. Blumler, 2009). This desideratum is due to the fact that most inquiries in this field concentrate on the functional or formal design of political homepages alone (e.g., Bimber & Davis, 2003; Foot & Schneider, 2006; Kluver et al., 2007) while content-specific aspects are seldom taken into account. Those few content analyses that do exist focus on selective indicators, specifically on the issues or the presence of negativity (e.g., Dolan, 2005; Benoit, 2007), and code these rather dichotomously across the whole website instead of using more fine-grained measures (e.g., Druckman, Kifer, & Parkin, 2010; Xenos & Foot, 2005). In addition, these studies are usually restricted to single-election periods, preferably on the U.S. senate or house level (e.g., Banwart, 2006; Latimer, 2007; Williams, Aylesworth, & Chapman, 2002) and thus do not allow for systematic generalizations.

In order to address these research deficits and to test whether e-campaigns basically suspend or adopt the media logic, this paper will present findings from a comprehensive quantitative content analysis of German party websites using a message-based coding across a range of mediatization indicators that are applied to a multi-election design. In this way, the study is able to track the essence, scope, and dynamics of mediatized politics on the web in comparison to traditional offline features of mediatization. The results provide empirical evidence that e-campaigns on all levels of the political system adhere in their contents to the media logic and thus replicate the structures of today's political journalism. This pertains to the selection of issues, the presentational style of the websites, their tone, and argumentative stance. The degree of mediatization increases over time and is slightly higher in first-order races. Other external variables, such as political ideology or electoral status, do not have a significant influence on the way parties incorporate the media logic in their online campaigns. Rather, mediatized politics on the Internet appears to be an all-encompassing phenomenon that affects various types of political actors and diverse categories of political races.

Applying the Mediatization Concept to Online Political Communication

The concept of mediatization has been receiving more attention in the academic literature since the mid-eighties (see e.g., Altheide & Snow, 1988). It seeks to describe the conditions, manifestations, and consequences that are associated with the diffusion and rising importance of the mass media in society (cf. Mazzoleni, 2008b; for a critique of the mediatization concept see Livingstone, 2009; Schrott, 2009). In the political arena, mediatization refers to the fact that people's opinion formation is based mainly on the respective accounts given by the mass media as the primary source of information. Journalists thus become central and independent players in political communication who shape the public perception of issues, actors, and events (see among others Page, 1996; Schudson, 2002). Parties and candidates try to exert an influence on these journalistic representations by adapting their behavior and public appearance to the so-called media logic. This encompasses the work routines, principles, and production techniques that determine today's news coverage (see Altheide, 2004, p. 294; Hjarvard, 2008, p. 113; Mazzoleni, 2008a, p. 2930; for a critique of the notion of a single media logic see Lundby, 2009). In the political context, mediatization can be understood as “the rising importance and the mass medial penetration of the political system and the displacement of political logic through media logic” (Schrott, 2009, p. 43f.; in similar ways see also Hjarvard, 2008, p. 113; Kepplinger, 2002, p. 973; Mazzoleni & Schulz, 1999, p. 250; Strömbäck & Esser, 2009, p. 216; on the different rationalities of politics and the media see Kepplinger, 2009a). In general, this displacement can be observed on four levels: in the decision-making of politicians (processes), in the organizational management of parties (structures), in their observation of the environment (awareness), and in their communication output (content) (cf. Donges, 2008, p. 31; Schulz, 2006, p. 42ff.).2

For the content dimension, past research has applied six empirical indicators to measure the degree of mediatization in offline politics. This encompasses formal and argumentative aspects. The formal indicators include those features of political communication that reveal an adaptation to the mass media on a stylistic level, irrespective of the specific issues, actors, or arguments that are at the center of the message (see Kepplinger, 2002; Meyer, 2009; van Noije, Kleinnijenhuis, & Oegema, 2008; Vowe & Dohle, 2009). This means:

  • the extent to which parties or candidates adopt a journalistic news style to address the public, e.g. in the structure or language of their messages (format);
  • the amount by which political messages are triggered by mediatized or staged events (such as press conferences, interviews, or party conventions) in comparison to genuine events (like parliamentary decisions, international summits, or political negotiations) (inducement); and
  • the frequency with which politicians explicitly refer to the mass media in comparison to other societal groups to substantiate their argument, to comment on public discussions, or to announce their own actions and plans (references).

These indicators have been used in studies on parliamentary communication in Germany and other countries. The results show that the mass media have become a focal point in political debates: Politicians regularly refer to the mass media in their speeches (see Vowe & Dohle, 2009), they are responsive to the event-centeredness of journalistic reports (see van Noije, Kleinnijenhuis, & Oegema, 2008), increase the number of symbolic statements (see Kepplinger, 2002), and prefer language forms that are effective in the media (see Meyer, 2009). In this way, politicians have deliberately adapted their messages on a stylistic level to the requirements of today's political journalism (see also the survey results among MPs and other political spokespeople by Cohen, Tsfati, & Sheafer, 2008; Kepplinger, 2009b, and Maurer & Mayerhöffer, 2009).

Aside from these formal aspects, the mediatization of political communication has also been measured by three argumentative indicators (see e.g., Mazzoleni, 1987; Brants & van Praag, 2006; Strömbäck & Dimitrova, 2011; van Aelst et al., 2008). These are derived from the current rules of attention that guide the choice of issues, actors, and statements in the international coverage of elections (for an overview see Kaid & Strömbäck, 2008). In detail, this includes:

  • the way in which politicians focus in their messages on the election campaign itself and the horse-race aspect of the competition in contrast to substantial policy issues (metacommunication);
  • the extent to which parties' communication revolves around their top candidates, their personalities, and private lives at the expense of other political actors (personalization); and
  • the degree to which parties concentrate in their messages on conflict and criticism rather than on positive self-promotion (negativity).

These indicators have been used in content analyses of traditional campaign channels such as press releases, TV spots, posters, or newspaper ads. The results show that political actors adopt the aforementioned patterns of election coverage for their own self-presentation: For example, they focus in their press releases most of all on attacks (for Germany, see Donsbach & Jandura, 2005), put the candidates at the forefront of posters and newspaper ads (see Keil, 2003; Lessinger, Moke, & Holtz-Bacha, 2003), and deal primarily with the professionalism of their election campaign while policy issues like economy, education, or health care move into the background (see Holtz-Bacha, 2000, p. 176f.). In this way, political actors do not only try to use the limited advertising space of their offline tools (“paid media”) in a more effective way, e.g. by providing voters with short, simple, and emotional statements that can easily be recognized and remembered (see in brief Kaid, 2004). They also seek to enhance the publicity in the accompanying media coverage (“free media”). In fact, studies have shown that journalists are more likely to report about those campaign messages that adhere to traditional news values. In this way, the mass media increase the circulation of these messages and create additional opportunities for voter persuasion (see Jasperson & Fan, 2004; Ridout & Smith, 2008).

This multiplication effect should be particularly relevant for election campaigns on the Internet: In comparison to traditional modes of advertising, which can also be recognized inadvertently by voters (“accidental exposure”, e.g. by posters on the streets or spots on TV), political websites require user activity and interest. The audience of e-campaigning therefore remains rather limited to a small group of engaged people that are already committed to a party or candidate (cf. Bimber & Davis, 2003; for Germany see Wagner, 2004). In order to reach other target groups, political actors are dependent on the publicity that is created for their campaigns by traditional news reports. It can thus be assumed that parties and politicians seek to tailor their web campaigns to the informational needs and expectations of the mass media. This assumption is backed up by studies showing that the Internet has become an important means of journalistic research (see Machill & Beiler, 2009). Moreover, findings from the USA and other countries demonstrate that political websites indeed exert an independent influence on the news agenda (cf. Ku, Kaid, & Pfau, 2003; Meraz, 2004; see also Aldé & Borges, 2006; Lee, Lancendorfer, & Lee, 2005; Tedesco, 2005).

This influence can take place in two ways: on the one hand structurally through the provision of materials and services on the websites that facilitate journalistic work (e.g., through the use of newsletters, picture galleries, or audio archives). On the other hand, journalists can be influenced through the very content of the homepages which is adapted to those principles of news selection and presentation that support follow-up media coverage. For the first route of influence, there are already sufficient findings from various elections that prove a structural adaptation of e-campaigning to the production techniques of journalists (see Jackson & Lilleker, 2004; Lipinski & Neddenriep, 2004; Tedesco, 2008). For the content level, however, there is still a need for studies that explicitly and comprehensively examine the presence and scope of the media logic on political websites. To address this gap, the following analysis applies the six mediatization indicators of offline research to German e-campaigns.


Sample and Research Design. The study content analyzed the websites of all German parliamentary parties that took part in six elections between 2002 and 2009. This includes the homepages of the Conservatives (CDU/CSU), the Social Democrats (SPD), the Liberal Democrats (FDP), the Greens (Bündnis 90/Die Grünen), and the Left Party (Die Linke). These organizations participated in three national elections (2002, 2005, 2009), one European parliamentary (EP) election (2004), and two state elections (Brandenburg 2004 and North Rhine-Westphalia 2005).3 Using the same political organizations on all levels of analysis allows both a cross-sectional comparison for the three types of races in 2004/2005 (state, national, European) and a longitudinal observation between 2002 and 2009 for the last three national elections (see Figure 1).

Figure 1.

Research Design

Moreover, the study reflects the diversity of the German federal system in that it includes two states that differ considerably from each other with regard to their geographic location (far-Western vs. far-Eastern state), their political history (former state of the FRG vs. former state of the GDR), their government constellation before the election (coalition of Social Democrats and Green Party vs. grand coalition of Conservatives and Social Democrats), the election outcome (change of government vs. continuance of government), and the degree of Internet usage among the population (higher in the Western and lower in the Eastern state). In addition, both state elections took place at almost the same intervals between the 2004 EP election and the 2005 national election and were not influenced by simultaneous campaigns on the next higher election level (first-order races).4

The units of analysis were all text-based messages that were published on parties' homepages.5 These were saved on a daily basis for the last four weeks before Election Day and coded manually on the article and statement level.6 The focus on parties' websites and their front page messages is due to four considerations: First, websites are the oldest and most common tool in e-campaigning. Hence, they are particularly suitable for multi-level and longitudinal comparisons. Second, websites are the most widely used format among German journalists and voters (see Machill & Beiler, 2009; von Pape & Quandt, 2010). Their contents are thus most salient on the Internet. This is particularly true for those text-based messages that are published in a prominent position on the homepage. In fact, these messages are also reproduced on other platforms, such as parties' newsletters, blogs, social networking sites, or discussion forums. Consequently, they have the widest audience reach and the potentially largest exposure effect among Internet users. Analyzing these units thus allows conclusions about the way in which the media logic shapes particularly those parts of the websites that have the greatest significance for parties' digital self-presentation. Third, in contrast to other text-based elements (such as party manifestos, issue sections, and “about us” features), these digital messages are neither borrowed from other offline paraphernalia nor pre-produced by the parties before the election campaign starts. Rather, they are written and delivered instantly during the “hot campaign phase” in reaction to current developments, events, and happenings. In this way, they provide a more valid and context-sensitive insight into parties' factual communication behavior during the election. Finally, the text messages were used in the very same manner on all party websites. This secures methodological equivalence for the later comparison.

Variables and Operationalization.

The study coded for several formal and argumentative variables that correspond to the mediatization indicators that were used in past offline studies (see above). Specifically, the factors format, inducement, references, and metacommunication were operationalized on the article level, while the variables personalization and negativity were measured on the statement level. For each article, eight statements could be coded including the respective source.

The format assessed the stylistic editing of the messages. This includes both the text genre and the formal presentation. With regard to the genre, differences were made between a speech that directly addressed voters (direct voter address), a message based on a sequence of questions and answers (interview), and articles that appeared as journalistic news reports (journalistic news style). The latter is characterized by the imitation of classical modes of presentation in the regular media coverage (see also Hjarvard, 2008, p. 113; Mazzoleni, 2008a, p. 2930). This includes a chronologically reversed structure with the most important information at the beginning (inverted pyramid principle); an adherence to the journalistic W-questions (where?, who?, what?, why?, etc.); the embedding of quotes instead of direct speech; the use of the third person; a distanced, matter-of-fact language; and a recognizable author who is not identical with the political actors themselves. The copresence of these features is regarded as a manifestation of the media logic (see also Strömbäck & Dimitrova, 2011). Furthermore, the study coded the number of words, images (photos and graphics), hyperlinks, the add-on materials (such as embedded audio/video streams or PDFs), and the days that the messages stayed online on the homepages. In this way, conclusions can be drawn about the formal sophistication of the website content in comparison to the traditional media coverage.

The inducement of the messages was operationalized according to the categories that were used in a study by Kepplinger (2002). He differentiated between statements, issues, and events as triggering moments. A message is induced by a statement when it refers back to an oral or written declaration of another politician that is then discussed on the party website under various points of view. An issue is the cause of an article when the message does not explicitly comment on an individual statement, but addresses an ongoing, general matter of concern that is debated in the public, such as questions of health care, security, or unemployment. In contrast, events are coded as triggers, when a message reacts to a single, self-contained incident that is clearly defined in space and time. Here, differences are made between genuine and mediatized events. Genuine events happen without any influence of the mass media. This includes, for example, natural disasters, parliamentary decisions, or political negotiations. Mediatized events, on the other hand, indicate public happenings that are either changed in their course due to the presence of media representatives (partly mediatized, e.g. canvassing tours, party conventions) or that are made possible only because of the media presence (fully mediatized, e.g. press conferences, media appearances, televised debates). An adaptation to the media logic is given when parties' website messages are published primarily in reaction to mediatized events.

The references of an article deal with the extent to which parties are concerned with representatives of the media system in comparison to other societal groups. For each message on the homepage, six references could be coded at maximum. These were categorized according to their name, their organizational or institutional affiliations, or their professional activity in various sectors of society, such as politics, economy, science, etc. A mediatization of e-campaigning is seen when the website messages are directed more towards members of the media system (e.g., to journalists, single media outlets, or broadcasting corporations, etc.) than to other civil actors (on this operationalization see also Vowe & Dohle, 2009).

The indicator metacommunication was operationalized on the issue level of the articles. Based on an extensive review of the research material, the author created an inductive list of single-subject matters for all election campaigns. These were individually coded and later summarized into common categories (such as foreign affairs, economy, education, etc.). For each report, the main subject was coded that covered more than 80% of the message. A mediatization of e-campaigning is visible when the articles deal more with the election campaign itself (e.g., parties' canvassing tours, their advertising strategies, or recent public opinion polls) than with substantial policy issues (see also Brants & van Praag, 2006; Mazzoleni, 1987; Strömbäck & Dimitrova, 2011).

Connected to this operationalization is the variable personalization. It was measured on the article and statement level. Personalization refers to the extent to which parties concentrate on the top candidates as main issues, main targets, and main sources of the statements. For all election campaigns, name lists of the top candidates and other party members were generated. These actors could then be coded as main focus of the messages (as part of the general issue coding), as statement object, or originator. An adoption of the media logic is given when other agents are neglected on all three levels of the websites in favor of the respective top candidates (see also Strömbäck & Dimitrova, 2011; van Aelst et al., 2008).

The last mediatization indicator deals with the news factor negativity (cf. Strömbäck & Kaid, 2008). In the online messages, this was coded on the statement level. Specifically, all statements were categorized into positive comments that explicitly praised the own party or candidate (self-promotion) and in negative comments that questioned or attacked the opponent (negative campaigning) (see also Benoit, 2007; Druckman, Kifer, & Parkin, 2010). For the data analysis, all positive and negative statements were classified according to their content. The basis for this were the five image dimensions that are regularly used in the international research literature to analyze the portrayals of political actors (i.e., competence, leadership abilities, integrity, empathy, and charisma; cf. Hellweg et al., 1989). An adaptation to the media logic can be seen when the website messages concentrate mostly on conflict and criticism, i.e. on attacks rather than on acclaims (see also Schweitzer, 2010).

Results: Formal Mediatization

In the six election campaigns that took place between 2002 and 2009 altogether 2,000 messages were coded on parties' homepages. The majority stemmed from the national and EP elections. These races offered the most comprehensive (between 280 and 349 words) and most illustrated text messages and updated them also more regularly (= shortest length of persistence) than on the state level (see Table 1).

Table 1.  Formal Mediatization
 NationalEuropean 2004State BB 2004State NRW 2005
  1. BB = Brandenburg; NRW = North Rhine-Westphalia

  2. n/a = The categories were not coded in this election.

  3. *= Differences to 100% are due to rounding errors.

Articles (N)350451370464186179
References (N)499558354553166163
Words (Ø)286349280318238242
Persistence (Ø days)
Articles (in %) with…      
 additional materialsn/a23.925.5n/a1.620.1
Format (in %):      
 news style77.066.761.981.062.454.2
 direct voter address13.416.924.18.024.715.6
Inducement (in %):      
 mediatized eventsn/a33.341.9n/a33.956.4
 genuine eventsn/a7.325.7n/a34.414.5
 issue discussionsn/a12.20.8n/a4.33.9
 othern/a00n/a0.5* 0
References (in %):      
 mass media25.324.033.310.330.739.3
 political institutions29.334.829.161.128.911.0
 other11.* 3.610.4

Furthermore, parties in the European and national elections provided more add-on materials in the articles (including PDFs, embedded audio and video streams, e-cards) and offered a greater number of hyperlinks. German e-campaigns are thus marked by a digital divide between first- and second-order races. This parallels the traditional differences in campaign professionalism that have been found for German parties' offline advertising in national races on the one hand and European parliamentary or state elections on the other (cf. Tenscher, 2007, 2008). For example, they spend three times more money in races for the national parliament (the Bundestag) than in European elections and have twice as many co-workers in the headquarters (cf. Tenscher, 2007, p. 72ff.). This divide is due to the political significance and reach of the first-order races and the correspondingly greater attention that they receive in the public and among journalists (cf. Tenscher & Schmid, 2009; Wilke & Reinemann, 2007). Both factors motivate parties to be also more proactive on the Internet: They publish a larger amount of sophisticated online messages to feed the extensive news interest of the media and to influence their overall campaign coverage. The formal presentation of these website messages has even gained importance over time: Between the 2002 and 2009 national elections, the number of illustrations, multimedia add-ons, and hyperlinks has increased across all parties (see Table 1). These developments correspond to findings on traditional election coverage that prove a more comprehensive and elaborate visualization of the candidates and the campaign in the press and on television since the end of the nineties (cf. Esser & Hemmer, 2008). German online campaigns thus seem to reflect the presentational trends that shape political journalism. This is particularly evident when looking at the format, inducements, and references of parties' website messages.

Format. On all election levels and over time, a factual news style dominated as text genre on the homepages. In contrast, direct voter addresses by politicians were rarely used (see Table 1). Specifically, the reports were written by members of the online campaign staff and appeared with individual headlines and teasers. They provided information according to the inverted pyramid principle, followed the journalistic W-questions in the course of the article, addressed the top candidates only in the third person, and referred to the latter mainly through brief and commented quotes (see also the results on the statement sources). In these ways, parties seek to evoke a sense of seriousness and credibility on their websites. Moreover, they also hope to increase their chances of publicity in the traditional mass media if they adhere to the regular news format. In fact, past input-output analyses from public relations research have shown that press releases are particularly likely to be covered in the news if they imitate the journalistic style of presentation (see in brief Fröhlich, 2008).

Inducement. The majority of the online messages was not caused by genuine events in society or by ongoing issue discussions. With the exception of the state election in Brandenburg, individual statements of politicians and mediatized events were the most common triggers for parties' online news (see Table 1). Political organizations thus react to those offline incidents on the Internet that promise intensive journalistic research and follow-up media coverage, for example in relation to the televised debates, the party conventions, or the canvassing tours. In this way, parties increase their likelihood of being covered in the news. Moreover, responding to the media agenda allows them to establish alternative interpretations for those campaign situations that are currently discussed in the public. In contrast, substantial policy debates or in-depth comments on social problems were hardly found on parties' websites (see also the findings on metacommunication). Hence, they concentrate more on tactical spin-doctoring than on civic education. This applies to both first- and second-order races and can be observed for all political organizations. Particularly strong is the focus on mediatized events among the major (55.4% vs. 35.9%) and Conservative parties (46.3% vs. 35.8%). Moreover, the reliance on mediatized events has grown over time: In national elections, the share of messages induced by these occasions has risen from an initial 33.3% in 2005 to 41.9% in 2009, while the number of issue-induced website messages has dramatically declined in the same period (from 12.2% to only 0.8%). Parties thus become even more responsive in their online communication to the offline happenings that are partly or fully controlled by the mass media. This indicates a strong mediatization in German e-campaigning. The same is true for the references in the website messages.

References. Across all political organizations and on all election levels, members of the media system were the most common (see 2009 national elections; 2004 state elections in Brandenburg) or the second most common actors (national elections 2002 and 2005; EP election 2004) that were named on parties' homepages (see Table 1). In particular, parties referred to the public broadcasting stations and the national quality newspapers (Süddeutsche Zeitung, Frankfurter Allgemeine Zeitung, Welt; 21.7%). Political representatives from national or supranational institutions dominated only in the 2004 EP election, while representatives from economy, science, or civil society (NGOs) were hardly mentioned. References to the population or individual voters were even completely missing in all elections. The website messages therefore focus primarily on the interaction between politics and journalism. Specifically, references to the mass media were used (a) to discuss journalistic statements; (b) to comment on statements of opponents that were covered in the news; (c) to announce politicians' appearances in the media; or—on rare occasions—(d) to substantiate personal arguments by citing news sources. The mass media thus advance to the focal point of parties' online messages. This can be seen for all political organizations, even though to a greater extent among the major parties (31.8% vs. 23.5%), for Conservatives (30.6% vs. 20.5%), and among incumbents (30.2% vs. 24.6%). The latter are, due to their political influence, a common subject in the media coverage. Hence, they have more opportunities to refer to news articles or journalistic commentaries that deal with their position. In addition, the number of media references has increased for all parties over time: Between 2002 and 2009 the respective share on the national level rose from an initial 25.3% to 33.3% and thus surpassed for the first time the number of references to political institutions (see also Table 1). Consequently, parties do not only adapt stylistically to the requirements of the mass media. They also point regularly and explicitly to journalistic resources in order to link their self-presentation strategically to the media discourse. This provides strong evidence for a formal mediatization in German e-campaigning. Whether this can also be substantiated for the argumentative dimension shall be discussed next.

Results: Argumentative Mediatization

The argumentative indicators include those aspects of parties' website messages that suggest an adoption of the content-specific patterns of election coverage, i.e. a preference for campaign rather than policy issues (metacommunication); a focus on the candidates as main issues, statement targets, and sources (personalization); and a conflict-oriented news selection (negativity). For metacommunication and negativity, the results confirm the assumed mediatization of German e-campaigning (see Table 2).

Table 2.  Argumentative Mediatization
 NationalEuropean 2004State BB 2004State NRW 2005
  1. BB = Brandenburg; NRW = North Rhine-Westphalia

  2. += Based on the total number of statements by challengers or incumbents, respectively.

  3. *= Differences to 100% are due to rounding errors.

Articles (N)350451370464186179
Statements (N)6891,2891,067755407531
Metacommunication (in %):      
 campaign as issue49.059.659.828.945.768.7
 policies as issue51.040.440.271.154.331.3
Personalization (in %): candidates…      
 …as issues3.
 …as targets24.018.915.8015.721.8
 …as sources26.728.316.07.246.725.4
Negativity (in %):      
 total amount56.656.651.055.056.347.1
 source: party60.858.381.490.836.252.4
 target: party69.272.485.390.073.469.2
 other0* 6.4010.0* 11.414.4
 dimension: political98.570.981.886.083.478.8

Metacommunication. Longitudinal content analyses of election coverage on television and in newspapers have found that German journalists increasingly concentrate in their reports on the strategic horse-race aspects of the campaign while substantial policy issues move into the background (see Esser & Hemmer, 2008). This trend is reflected in parties' e-campaigning: With the exception of the 2004 EP election, the amount of metacommunication varied on the websites between one half and two thirds of all examined online messages (see Table 2). On their homepages, political organizations thus concentrate on aspects such as their campaign activities, recent opinion polls, or the performance of the candidates in the last televised debate. Controversial questions from the fields of economy, finances, or foreign affairs, however, were hardly discussed. This campaign-centeredness on parties' websites is strongest for the state election in North Rhine-Westphalia in 2005 (68.7%) as well as for the past national elections in 2005 (59.6%) and 2009 (59.8%). Moreover, the amount of metacommunication has continuously increased over time from 49% in 2002 to 59.8% in 2009, which exactly mirrors the development that has been observed for German offline campaign coverage. For parties, this concentration on the horse-race fulfills several functions: First, political actors seek to feed journalists with those information on the websites in which the latter are most interested in. This secures follow-up coverage. Second, by focusing on the campaign itself, parties avoid polarizing or unfavorable policy discussions that could appall potential voters. Third, by talking about the own campaign performances, parties create an impression of modernity and professionalism in the public. This is seen as a functional equivalent to political competence (see also Parry-Giles & Parry-Giles, 1999). Members of the ruling government coalition (= incumbents) were particularly likely to center their website messages on metacommunication (59.2% vs. 49.3%). In this way, they tried to divert attention from controversial political decisions that were made in the past legislation period (see also Holtz-Bacha, 2007, p. 71). The only exception from this pattern was the EP election: Here, political organizations focused on their homepages throughout on EU policy debates, e.g. on economy, agriculture, and security. Campaign-related articles, on the other hand, accounted for only a quarter of all messages (see Table 2). This was true for all parties irrespective of their size, their parliamentary status (government/opposition), or their ideological orientation. This finding can be explained by both the scarce EU knowledge in the population (see the findings of the latest Eurobarometer: and by the very nature of the EP election: It is a second-order race that is marked by low degrees of campaign professionalism (see Tenscher, 2007) and limited media attention (see Wilke & Reinemann, 2007). To mobilize voters, parties are therefore forced to convey even the most basic political information about the EU on the Internet so as to explain the meaning, purpose, and relevance of the EP election. This motivation is not given in online campaigns at the national or state level, where parties rely instead on the horse-race aspects. This finding defies initial hopes that political actors could provide citizens with more extensive, fact-based, and non-mediatized information on the web. Rather, they adhere in their messages to the issue priorities of traditional political journalism. This confirms a strong adaptation to the media logic on the content level of e-campaigning. For the next argumentative indicator, though, this does not hold true.

Personalization. In communication research, personalization has been found to be a central criterion in the selection of news. The focus on individual actors allows journalists to explain complex political situations in a simple, straight-forward, and engaging manner that finds its way to the audience. In Germany, longitudinal content analyses have shown that the focus on the top candidates in TV and newspaper coverage has dramatically increased in national elections since the introduction of the televised debates in 2002 (cf. Esser & Hemmer, 2008). In this context, even politicians' private lives receive more attention in the news (see Rohowski, 2009). Similar trends towards personalization have also been found for traditional campaign channels like parties' TV spots, ads, or posters (see above). Here, the candidates are placed at the center of the messages to emotionally connect with voters and to contribute to parties' image-building. For German e-campaigning, however, this strategy is not confirmed. Rather, the study finds a clear trend towards depersonalization on parties' websites (see Table 2). This becomes evident in all three dimensions of analysis; i.e. on the issue, statement, and source level. In all elections and across all political organizations, less than 4% of the online messages were primarily concerned with the candidates as main subjects. This proportion has also dropped significantly between 2002 and 2009. References to the private lives of politicians, i.e. to their families, homes, or personal interests, were almost absent (only 2.7% of the text messages in the 2009 national election). In addition, the top candidates were addressed as main targets in less than a quarter of all statements in the website messages. These concentrate instead on other parties, specifically on CDU/CSU and SPD, that are named and attacked as collective entities (see also below). This organizational focus was particularly prominent in the website reports that were launched during the 2004 EP election (see Table 2). Due to the low name recognition of EU politicians, the messages dealt primarily with national parliamentary parties and their European coalition partners. The major candidates, on the other hand, were barely mentioned on the issue, statement, or source level. With the exception of the 2004 state election in Brandenburg, a low citation rate was found in all races for the respective top politicians. Generally, the candidates accounted for only a quarter of all statements. On the national level, the number of direct quotations has also decreased significantly between 2002 (26.7%) and 2009 (16%). This development corresponds to the shrinking sound bite phenomenon that has been observed in Germany for both television and newspapers for the past national election cycles (cf. Esser & Hemmer, 2008). In contrast to the regular news coverage, though, this trend towards de-authenticization is not connected to a parallel rise of journalistic comments. Across all organizations, other party members remain the most common sources that are cited on the websites. In this way, a diversification of voices can be seen in that second-rank politicians get a chance to speak for the interests and concerns of the organization. Comments of the top candidates, in contrast, are increasingly transferred to more specialized platforms for personal self-promotion on the Internet, such as candidate homepages and profiles on social networking sites. This unburdens the party websites and allows political organizations to counter the general focus on the candidates in the mass media. These findings illustrate that political actors are indeed able to partially suspend the media logic in their online campaigns. However, with the exception of the detected depersonalization trend, this option is hardly used. This also holds true for the last indicator.

Negativity. The focus on conflict and attacks is another key factor in the journalistic news selection. In the German press, the number of critical reports has significantly increased since the 1998 national elections (cf. Esser & Hemmer, 2008). Moreover, negative information are more easily recognized and remembered by recipients, irrespective of their political predispositions (see Lau, 1982). Parties and candidates thus try to make use of this quality by focusing in their campaigns on assaults on the opponent. In this way, they hope to yield additional publicity in the mass media and to cast doubt on the competitors' suitability for office. In German e-campaigning, this strategy is visible in all elections and across all parties. With the exception of the 2005 state elections in North Rhine-Westphalia (47.1% negativity), the statement-based criticism on the opponent dominated over positive self-promotion. As to that, challengers were generally more likely to go negative than incumbents (see Table 2; for the exception in the early national election in 2005, cf. in detail Schweitzer, 2010). A significant difference between the ideological camps, however, was not apparent. Over time, the number of assaults has also slightly declined at the national level (from 56.6% to 51%). This can be explained by the shared government responsibility of CDU/CSU and SPD since 2005 that made it difficult for both partners of the grand coalition to attack each other in the campaign. Despite this special circumstance, however, a conflict-oriented argumentative behavior prevailed in the website messages even in the 2009 national election. The attacks were usually formulated by second-rank party members or by third sources (e.g., supporters, journalists) and were targeted in all elections at the parties as collective entities (see Table 2). The latter were mainly criticized in political terms, i.e. with regard to their competence (especially in economic matters) or their leadership abilities. In contrast, personal denigrations referring to a politicians' appearance, popularity (charisma), or empathy were rather scarce (see also Schweitzer, 2010). This underscores the trend towards depersonalization in German e-campaigning. Overall, these results indicate another orientation on the media logic. Parties do not make use of the freedom of self-presentation on the Internet to break away from the conflict-centered style of news reporting. Rather, they continue to focus on their websites on a provocative and headline-grabbing attack behavior that fosters journalists' attention (cf. also Ridout & Smith, 2008). The assumed mediatization of online political communication thus receives further support. In a final conclusion, the main findings of the analysis will be summarized and discussed with regard to their theoretical implications.

Summary and Conclusion

The present study sought to challenge the conventional wisdom that e-campaigns are basically non-mediatized and thus enable political actors to bypass journalistic principles of news selection and presentation. The inquiry was based on a quantitative content analysis that applied six mediatization indicators from past offline research to German party websites in two state elections, three national elections, and one European parliamentary election between 2002 and 2009. This design allowed cross-sectional comparisons of three types of races in the years 2004/2005 and a longitudinal observation for the national elections in the years 2002, 2005, and 2009. In this way, the study offers a comprehensive account of the extent, scope, conditions, and trends of mediatization in online campaigning. The results provide empirical evidence that parties adopt rather than suspend the media logic on their websites. This was true for five of the six indicators.

In detail, parties' online messages were presented in all elections in a journalistic news style as regards the sequence of information, the preferred language, and the formal appearance (format). The articles were not caused by genuine happenings (such as parliamentary decisions) or ongoing public controversies but were rather published in reaction to mediatized, i.e. staged events such as press conferences, party conventions, or canvassing tours (inducement). Moreover, the messages referred frequently to representatives of the media system as the most or second most common societal group, while other actors from politics, economy, or science moved into the background (references). Substantial policy discussions were scarce as well: The website messages concentrated instead on the strategic horse-race aspects of the campaign (metacommunication) and thus mirrored the patterns of traditional election coverage. This was also true for the statement level of parties' websites: Their releases focused most of all on attacks on the political opponent to match the conflict-centered news selection in political journalism (negativity). The only deviation from the media logic was visible for the indicator personalization. This was evident for all political organizations and all races: On the issue, statement, and source level, the top candidates were not at the forefront of parties' online messages. Rather, other party members dominated as actors in the website reports. This led to a diversification of voices that overcomes the focus on the party leaders which has been observed for both German offline election coverage and parties' traditional campaign channels. For all other indicators of the study, however, the reliance on the media logic continued. In fact, the mediatization of German e-campaigning has grown over time as the number of media references, the amount of metacommunication, and the visual adaptation to current journalistic trends has increased in the national elections from 2002 to 2009.

With regard to the basic research question, these findings therefore lead to the conclusion that e-campaigns—contrary to the conventional wisdom—are not free of the media logic. Rather, the mediatization of online political communication appears to be an all-encompassing phenomenon that affects all parties and all election levels in Germany. Slight variances appeared only in the degree of mediatization with the latter being a bit more prominent in those instances where extensive offline coverage and journalistic interest is almost guaranteed, i.e. in first-order races as well as for major parties and incumbents. The discretionary power of the mass media thus goes well beyond the realm of traditional news reporting. The basic elements of election coverage are adopted by parties in their online political communication to improve their chances of publicity. In this way, the mediatization of e-campaigning can be understood as a strategic response to the limited audience reach on the Internet. The latter shall be overcome by a homepage that is specifically tailored to journalistic needs so as to ensure a wide and favorable follow-up coverage in the news media (see section 2).

This does not preclude, however, that other, less prominent parts of the website do provide some basic information for citizens, e.g. on the election manifestos, the party history and structure, or the candidates (cf. Schweitzer, 2008). In Germany, though, these data are usually borrowed from offline material and are not specifically produced for the online campaign. Moreover, they are placed in subordinate sections of the website which makes them harder to find for voters. This impairs their effectiveness as a means of civic education (cf. Lupia & Philpot, 2005). In contrast, the salient front pages could be used to invite citizens to learn more about the election and the policy issues at hand. Yet German parties do not make use of this opportunity: Rather, they deliberately preserve the most central parts of their Internet presences to reflect the media logic. In this way, the homepages become strong indicators of parties' strategic choices and priorities: Instead of fostering civic education, they concentrate here on an adaptation to journalistic needs and requirements. This has the side-effect that also those patterns of election coverage are prominently replicated on the web that have been held accountable for rising political alienation and civic apathy in the public, i.e. extensive negativism and horse-race depictions (cf. Cappella & Jamieson, 1997; Patterson, 1993). German parties thus do not only make it difficult for voters to find substantial policy discussions on their websites. They also thwart the potential effects of online civic education through focusing on detrimental campaign portrayals in the first place. In this way, the results of this study cast doubt on the democratic hopes that have been placed on the Internet as an information basis for a strong democracy (cf. Barber, 1984).

In fact, the technical potentials of the World Wide Web do not guarantee an online campaign style that compensates for the deficiencies of offline political communication. Rather, typical patterns, considerations, and strategies of real-world politics are likely to be transferred to cyberspace when parties and candidates perceive them to be beneficial for their respective campaign goals. Research therefore needs to ascertain in each case whether the theoretical hopes that surround the diffusion of e-campaigning are realistic in practice. For the aspect of mediatization, future studies should clarify in particular whether the adoption of the media logic can be found for other parts of the websites and for other forms of e-campaigning as well (e.g., blogs, candidate homepages, or online videos7 ). These inquiries can be combined with additional surveys among campaign consultants, politicians, and party members to learn more about the reasons and motives that lead political actors to adopt the media logic on the Internet. In the present study, this multi-method approach could not be applied due to economic restrictions. Furthermore, longitudinal and comparative analyses are needed to test for mediatization in different temporal, spatial, and social settings, such as in political routine periods (legislative period), in other countries, and among other types of political actors. This allows insights into the generalizability of the present findings in relation to past evidence on the variation of offline mediatization (see e.g., Esser, 2008).


  • 1

    Originally, the term “disintermediation” emerged in the field of finances at the end of the 1960s (cf. e.g., Hester, 1969). It denotes the removal of intermediaries (such as brokers or agents) in a supply chain. The term has been popularized in economics by Hawken (1983) before it became widespread in online communication. I am grateful to one of two anonymous reviewers who pointed this out.

  • 2

    This understanding of mediatization has to be distinguished from the overall concept of “mediation” (cf. also Hjarvard, 2008, p. 114; Strömbäck & Esser, 2009, p. 207). The latter denotes “the rather neutral act of communication through different media, as opposed to reconfiguring the whole of political life around media practices, technologies, and institutions” (Strömbäck, 2011, p. 368). The mediation in society—as shown in the growing importance of various media channels for human interaction—is thus a logical prerequisite of the mediatization of politics that denotes a deliberate adaptation of parties and candidates to journalistic needs and requirements within these channels.

  • 3

    In the 2002 national elections, a pilot study was carried out that concentrated only on the websites of CDU, SPD, FDP, and the Greens.

  • 4

    On the classical distinction between first- and second-order elections, cf. Reif & Schmitt, 1980; Reif, 1984; see also recently Carrubba & Timpone, 2005; Weber, 2007.

  • 5

    The total data collection also comprised 33 online videos that were shown on the front pages in the hot campaign phase of the 2009 national election (= 8.9% of all messages in this election; 1.6% of all messages in all elections). In earlier races, audiovisual content was not at all used by German parties. To allow for a consistent longitudinal and multi-level comparison, the online videos for the 2009 election were therefore removed from the analysis. Future studies should include this format as well (see also the remarks in the conclusion).

  • 6

    To determine the intracoder reliability, the author recoded a random sample of 5% of the research material in each race. For all election campaigns, the respective coefficients varied between .83 and 1.00 according to Holsti. This formula has been used in other analyses of political websites as well (e.g., Banwart, 2006; Bimber & Davis, 2003; Trammell et al., 2006). In the methodological literature, it is deemed appropriate for coding decisions on a nominal level (cf. Stempel, 2003, p. 216; Watt & van den Berg, 1995, p. 375).

  • 7

    In a recent study on political TV news, Strömbäck & Dimitrova (2011, p. 35ff.) have developed several indicators to measure mediatization in audiovisual content. These could be applied and extended for an analysis of online videos.

About the Author

Eva Johanna Schweitzer is a PhD candidate in the Department of Communication at the University of Mainz, Germany. Her research interests include online political communication, e-campaigning, and comparative media studies.

Address: Department of Communication, University of Mainz, Colonel-Kleinmann-Weg 2, 55099 Mainz, Germany. Email: