Socially mediated issue ownership

Party competition and issue ownership

Socioeconomic cleavages no longer condition partisanship in ways that were once the norm. Some prominent scholars of party systems argue that parties have become disconnected from wider society and that the competition they pursue therefore lacks meaning. In Ruling the Void, Mair (2013) declares that the age of party democracy has passed and that parties are incapable of sustaining democracy in its present form. Several trends are identified that together strengthen the notion of a growing disenchantment with political parties: declining voter turnout, decreased party attachment, and increased electoral volatility. The parties themselves have reacted to these changes by dedicating more of their programs to a wider variety of issues unrelated to the traditional left-right dimension of party competition, such as the environment and immigration (Green-Pedersen, 2007). Parallel to the decline of established political parties and an overall decrease of the traditionally dominant economic cleavage in favor of newer salient issues, challenger parties have emerged. These parties reshape the political landscape by placing new issues on the agenda, such as immigration and European integration (Kriesi et al., 2008). Contemporary issue competition within a multiparty system is therefore multifaceted; it also plays out on a variety of media platforms.

Since socioeconomic cleavages have become less important in forming partisanship, the explanatory power of social–structural models of voting behavior have declined over the past few decades. Thus, scholars must now closely consider the role of issue preferences in party choice (Van der Brug, 2004). Issue ownership is a theoretical framework used for explaining both party and voter behavior. It implies that “parties and their candidates attempt to mobilize voters by emphasizing issues on which they hold a reputation of competence. Political parties in turn receive support on the basis of those issues that they are perceived to own at election time” (Bélanger and Meguid, 2008, p. 477). The effect of issue ownership is conditioned by the perceived salience of certain issues. A party’s competence in relation to a particular issue only influences voter behavior when the issue is also considered important (Bélanger and Meguid, 2008). The concept of issue ownership has been argued to involve both a competence and associative dimension (Walgrave et al., 2009). While the competence dimension reflects that voters consider some party or parties to be better able to deal with specific issues, associative issue ownership is defined as “the spontaneous identification between specific issues and specific parties in the minds of voters” (Walgrave, Lefevere, and Tresch, 2012, p. 771). Both dimensions are considered determinants of voting behavior when combined with the relative importance of issues. Voters may think that one party has the best policy, or they may associate a certain party with a specific issue. However, whether they cast their vote for this party also depends on the relative importance voters attribute to said issue.

Voters’ partisanship tends to dominate perceptions of issue ownership. However, their attitudes and performance evaluations also matter (Stubager and Slothuus, 2013). Perceptions of party reputation on a given issue are relatively stable, but the importance voters ascribe to certain issues is subject to more rapid changes. If parties strategically emphasize certain issues or successfully frame issues that have become salient during the campaign, they can influence perceptions of ownership in the short-term (Blomqvist and Green-Pedersen, 2004; Holian, 2004; Bélanger and Meguid, 2008; Walgrave et al., 2009). Mass media coverage determines perceptions of issue salience to a large extent (agenda setting), but also influences the perceived link between parties and certain issues, particularly in relation to the dynamics of short-term issue ownership (Page and Shapiro, 1992; Thesen, Green-Pedersen, and Mortensen, 2017; Walgrave and De Swert, 2007). Even if voters’ positions are relatively stable over time, the dynamic nature of issue salience and increased volatile voting behavior render it necessary for parties at each election to ensure they are associated with issues they are trusted to be capable of handling in order to make relevant policy changes.

Recent work has incorporated issue ownership as one component into larger frameworks such as the issue incentive model (Green-Pedersen, 2019) or the issue yield model (D’Alimonte, De Sio, and Franklin, 2020) in order to capture the more dynamic processes behind issue competition. Some of the major criticism of the issue-competition literature revolves around a partial neglect of the diverse nature of policy problems that drive attention to issues (Green-Pedersen, 2019), and to what extent parties’ selective issue emphasis implies that other issues are de-emphasized (Dolezal, Ennser-Jedenastik, Müller, and Winkler, 2014). While acknowledging that issue ownership cannot fully explain parties’ issue strategies or the issue content of party politics, it can provide an explanation of variation (or a lack thereof) in linkages between issues and parties on various media platforms. Instead of assessing voter evaluations or party strategy, a socially mediated issue-ownership dimension emphasizes interpersonal communication and inter-media agenda setting.

Issue ownership theory offers little insight into the role of media or interpersonal political communication in shaping which issues are perceived to be more important (Kiousis, Strömbäck, and McDevitt, 2015). Contemporary perspectives derived from social media analysis can therefore complement this traditional theoretical framework to create a more comprehensive understanding of issue dynamics around elections. In the next section, this paper will bring together the concept of issue ownership with the idea of a social media logic that intermediates parts of public opinion and traditional media content, which this article refers to as a socially mediated issue ownership dimension.

A socially mediated issue-ownership dimension on Twitter

Voters are more likely to operate on the basis of short-term considerations and influences when disengaging from the arena of conventional politics. When the electorate becomes progressively destructured, it affords more scope to the media to play the role of agenda-setter and requires a much greater campaign effort from the parties (Green-Pedersen, 2019; Mair, 2013). This points to an increased importance of media in shaping issue-ownership perceptions among the population. The extent to which social media reinforce these processes has yet to be explored as is what role social media fill in contemporary policy agenda setting. In policy agenda setting, the allocation of attention to issues is understood as an agenda-setting process, since “attention is important for politics because it is consequential and, at the same time, scarce” (Green-Pedersen, 2019, p. 25). Parties therefore compete by presenting their views on the issues dominating the political agenda, which can be understood as a hierarchy of attention to issues existing at a given time (Green-Pedersen, 2019, p. 28). While parties try to influence the issue agenda, they are at the same time swayed by it. When an issue climbs up the media agenda or becomes prominent on social media, parties need to be responsive. Policy agenda-setting dynamics on Twitter are, however, formed by different mechanisms than the traditional news media agenda.

Issue dynamics on Twitter are shaped by this platform’s specific digital architecture (Bossetta, 2018), be that technical, spatial (for example, limitations in the number of characters), or participatory (those who involves themselves in political discussions). This might then result in an aggregated issue agenda in which some issues become systematically marginalized, whereas others appear more salient. Previous research has shown that public, political, and media agendas mutually influence each other (Soroka, 2002; Vliegenthart and Walgrave, 2011; Walgrave, Soroka, and Nuytemans, 2008). These mutual influences are particularly visible on Twitter, as prominent political topics in the public’s mind, in the news, and on Twitter often correspond (Jungherr, Schoen, and Jürgens, 2015). Thus, Twitter offers a mediated image of political reality that is highly interconnected with traditional media and public opinion on salient issues.

Due to Twitter’s speed and flexibility, discussions can be expected to be responsive both to changes in the media agenda and public opinion. Issue ownership measurements on Twitter may therefore reflect short-term changes in issue salience in the media or for the public in general, more noticeably, since those discussing politics on Twitter react to real-time events (Jungherr et al., 2015; Pasek et al., 2018). Political events and topics covered by mass media are more likely to catch the attention of many users simultaneously and thus be reflected in pattern shifts in the aggregates of Twitter messages (Jang and Pasek, 2015; Jungherr et al., 2015). Many of the issues covered on Twitter derive from the mainstream media agenda, and therefore Twitter would amplify certain issues depending on the media’s reporting rather than put forth an alternative issue agenda. That is, when some issues are salient in the news media, they will evoke discussions on Twitter. Similarly, priming effects of events reported in the news media influence survey responses (Iyengar and Simon, 1993). Traditionally, agenda-setting influence descends from the media or politicians to citizens. On Twitter, however, there is a reciprocal relationship between news outlets and Twitter content as well as between the whole spectra of parties, politicians, journalists, and regular social media users that is likely to be reflected.

The mediation of issue ownership through social media therefore captures both aspects of interpersonal political communication that occur on Twitter as well as the reciprocal agenda-setting relationship with the mass media. This measurement is more closely related to an associative dimension of issue ownership than a competence dimension. What is being measured, however, reflects what issues are discussed and addressed in relation to which parties on social media; hence, it involves the logic pertaining to political discussions on Twitter. It is therefore more accurate to define measurements of issue ownership on Twitter as socially mediated rather than merely relating to an association. Issue ownership has been identified as a determinant of political parties’ media coverage (Van Camp, 2017; Van der Brug and Berkhout, 2015). On social media, there is no indication as to why this media logic would not resonate similarly. If anything, the social media logic would contribute to a reinforced mainstream agenda and create an accentuated association between issues and parties. In considering how sensitive Twitter discussions are to the news media agenda and newsworthy events, the following expectation has been formulated:

Hypothesis 1: Parties are addressed more frequently in relation to issues they are perceived to own, compared to other issues on Twitter.

Due to the logic of political coverage on Twitter and the above-mentioned characteristics of Twitter with a discursive component, a fluxional news component, and a political content component, it is expected that Twitter mediates a dimension of issue ownership in a way that is sensitive to short-term changes during election campaigns. Thus, it might reveal issue competition dynamics or variations in the overall policy agenda during election campaigns. It can therefore be expected that parties’ perceived ownership does not reflect similarly for all the issues analyzed. In line with the hypotheses above, the linkages between parties and issues are likely to be accentuated on Twitter for those areas where some parties dominate issue-ownership perceptions, compared with issue areas characterized by weaker ownership perceptions. In situations with weak associations between parties and issues, Twitter discussions might reproduce a more ambiguous association in how parties are addressed together with these issues on Twitter. Some deviations might therefore be expected:

Hypothesis 2: Political topics characterized by strong issue-ownership perceptions are more likely to be discussed on Twitter, together with the mention of parties “owning” these issues, than issues that are less linked to specific parties.

The issue-competition literature contributes valuable insights on the importance of the specific type of policy problem (Green-Pedersen, 2019) as well as the distinction between positional and valence issues (D’Alimonte et al., 2020). However, instead of offering a potential explanation of parties’ issue strategies, this second hypothesis is formulated based on expectations as to what extent the Twitter discussions are more likely to be in flux when it comes to issue areas with weaker ownership perceptions. Similar to the mainstream news media agenda, it is likely that the attention is directed towards bigger mainstream parties in economic issues, for example, when these parties make statements or present new initiatives, whereas issues characterized by weaker ownership perceptions are addressed in relation to several parties on Twitter.

Case selection: Sweden as a two-dimensional multiparty system

In multiparty systems, voters can have confidence in more than one party to handle a specific issue. Issue ownership might therefore accumulate over several issues, meaning the more issues a party is trusted to handle within a comprehensive policy set, the more likely that party is to receive votes (Karlsen and Aardal, 2016). In multiparty systems, it is also possible for several parties to be internally united on a similar issue goal, which could mean multiple parties pursuing ownership of a particular issue (D’Alimonte et al., 2020). Sweden has historically been seen as one of the most unidimensional political systems, since the traditional left–right dimension strongly structured party competition and voting behavior. In this party-centered system with its highly ideologically motivated voters, parties’ policy positions and voters’ issue standpoints have had large explanatory power in models of voting behavior (Oscarsson and Holmberg, 2015). Left–right ideological predispositions are still strong determinants of voters’ party choice, even though new conflicting issue dimensions, such as immigration, the environment, and gender equality, have gained importance. Judging by its recent elections, Sweden can therefore be characterized by a multidimensional cleavage structure that forms a two-dimensional party system (Berg and Oscarsson, 2015). An overview of how the parties are placed within these two dimensions is presented in Figure 1. There is a clear divide between parties’ positions on the left–right political scale. On a libertarian–authoritarian scale, the Sweden Democrats (SD) are distinguished by their high placement in terms of authoritarian values, followed at some distance by the Christian Democrats (KD). Issue ownership in Sweden thus relates to issues on the left–right cleavage structure alongside a second cleavage revolving around issues such as the environment and immigration.

Figure 1:

Party positions in Sweden on a two-dimensional, left–right, libertarian– authoritarian scale.

Note: The data presented in this figure are taken from the Chapel Hill Expert Survey (Polk et al., 2017) using the variables galtan and lrgen.

In Sweden, few regular citizens use Twitter and even fewer tend to actively participate in discussions.^[1] Political discussions in particular tend to be dominated by a smaller group of active users that make up a substantial part of the activity; these are often politicians themselves, established journalists or opinion-makers (Larsson and Moe, 2015). The possibility to reach out to influential actors such as journalists provides Swedish parties with strong incentives to use Twitter strategically during election campaigns.

Natural language processing

A dictionary-based automated lexical approach was used to structure data (i. e., the textual content of individual posts) from Twitter. This is a common method within the semantic and computational linguistic to analyze the content of large corpora of text, using either data-driven or bottom-up lexical approaches (Schober, Pasek, Guggenheim, Lampe, and Conrad, 2016). This approach allows for an analysis of the frequency of keywords from a particular semantic category for each post in the corpus (Schober et al., 2016). As part of the data preprocessing stage, the stem and lemma of each word were detected using stemming and lemmatization libraries specific to the Swedish language, separating text into separate words and removing inflectional endings only. A lemma, the base dictionary form of a word, can be used in different ways in a sentence (e. g., as a noun or verb), thus providing different meanings to a word. As a part of speech tagging, each word in a sentence is determined, making it then possible to analyze tweets mentioning a word as a political topic and not a verb. For example, in the Swedish language, “job” can have several inflections and be used as the verb “to work”. In addition, stop words were removed from these tweets (dropping common terms).

A manual classification of the preprocessed words in the whole subsample was then conducted. Based on the phrases and words used in the coding scheme of the SNES^[2] for the nine policy areas, the base dictionary form of the words in the total corpus of tweets was manually matched to these predefined categories of political issues to allow for comparison. In the coding scheme of the national election survey, the economy, for instance, has nine sub-issues for coding the answer of survey respondents such as economic growth or banks. For example, the occurrence of “banks” in the Twitter data set corresponds to one of the subcategories of economy. Any preprocessed word that did not correspond to any of the subcategories used to code survey answers was left out^[3].

Comparing issues between surveys and tweets captures different aspects of issue ownership. Words on Twitter do not equate to survey respondents’ answers concerning the question of which party has the best policies in a particular area. The competence dimension, as measured in the survey, demonstrates that a positive relationship between party and issue is present. On Twitter, the dimension of issue ownership is more closely related to an associative connection, since the mention of an issue in relation to a party on Twitter can be positive, negative, or neutral. Nonetheless, the assumption is that voters’ perceptions of issue ownership will be reflected in the differences between what party is discussed in connection to which issue. In order to analyze how similarly voter’s issue ownership perceptions are mediated on Twitter, the Pearson correlation coefficient was used to compare the two different measurements of issue ownership. Since this approach cannot capture the differences in the two data sets, the results should be interpreted with this in mind. However, the correlation coefficient does signal whether a similar ratio in the relationship between issues and parties on Twitter and in surveys can be found, which is helpful for interpreting these results.

Issue ownership in the 2014 Swedish general election

Table 1 shows the electorate’s perceptions of issue ownership in the 2014 Swedish election. The Social Democratic Party (S) had the best reputation on most issues and is currently and historically the biggest party. The Conservative Party (M) became the second biggest party after the 2014 election and had the strongest issue ownership on the economy and taxes. The anti-immigration party, Sweden Democrats (SD), became the third biggest party after the election and had the strongest ownership on immigration/integration. The Sweden Democrats’ electoral gains in the 2010 and 2014 elections are likely explained by the increased salience of the immigration issue. The Green Party (MP) was regarded as owning the environmental issue, and almost half of the voters perceived the party to have the best policy in this area. Voters thought that the Centre Party (C) had the second-best policies for environmental issues. The Liberal Party (L) has since the 1998 election maintained issue ownership over education and schools (Martinsson et al., 2013). In the 2014 election, however, they lost part of their hold on this main profiling issue, with more than a quarter of voters instead thinking that the Social Democrats had the best policies in that domain. The Left Party (V) had the strongest ownership on social welfare (by 20 % of the voters). During the 2014 election, the major issue that the Left Party campaigned on was limiting profits for private companies in the welfare sector. The majority of voters were in favor of the Left Party’s position (Nilsson, 2017). The main profiling issue of the Christian Democrats (KD) was health care. However, three other parties experienced higher levels of trust from voters in dealing with this issue.

In relative terms, a stronger issue-ownership dimension characterizes some issue areas over others. Measured as the standard deviation of each issue as it relates to the parties, issues with weaker ownership perceptions are immigration, education, social welfare, pensions, and health care. For example, ownership over immigration-related issues is distributed across more parties and therefore has a lower standard deviation.^[4] Policy areas with relatively strong perceptions of ownership are taxes, environment, labor market, and economy.

Issue-ownership dimension on Twitter

The analysis revealed a strong issue-ownership dimension on Twitter, thus supporting Hypothesis 1. When the political issues were addressed and coupled with a party, this resulted in 0–47 % issue association for a particular party (see Table 2). The Social Democrats and the Conservative Party dominated on Twitter in discussions of taxes, the labor market, the economy, and pensions. These are issues that they are considered to also “own” in the representative survey. When correlating the two measures, all of these issues had a strong, positive correlation.^[5] The traditional economic dimension has strongly structured party competition and voting behavior in Sweden and elsewhere, issues usually owned by large mainstream parties (Green-Pedersen, 2019), which is consistent with the pattern on Twitter. Social welfare, also belonging to a traditional left–right dimension, had the strongest association with the Left Party and Social Democrats on Twitter (Table 2), which was positively correlated with the survey results.^[6] On Twitter, the Left Party was mentioned most often when issues concerning social welfare were addressed, particularly in relation to discussions of limiting profits in the welfare sector. Searching tweets for the whole sentence “profits in welfare” gave 139 results in tweets also mentioning the Left Party but only between 0 and 17 for all other parties. The word “welfare” alone accounted for most mentions of the Left Party with this issue.

Issue ownership over education also displayed a similar pattern on Twitter as in the survey.^[7] On Twitter, this issue was most strongly associated with the Social Democrats, followed by the Liberal Party and the Greens. As discussed previously, the Social Democrats took over the “ownership” of education from the Liberal Party, which had had ownership over the issue following the 1998 election. This shift in ownership was reflected on Twitter, where the Social Democrats were clearly associated with the issue. In terms of party mentions and topic discussion, the Social Democrats gained 25 % compared with 17 % for the Liberal Party and the Greens.

Health care did not generate as clear differences between the parties as those mentioned above did.^[8] The Social Democratic Party is considered to have the best policies regarding health care by survey respondents (38 %). On Twitter, this association was weaker. Health care was discussed equally in relation to the Conservative Party, the Social Democrats, and the Christian Democrats.

Both environment and immigration have emerged as important issues for the electorate, introducing another policy dimension in Swedish politics. The parties who are mobilizing voters on primarily one of these two issues are the Green Party and the Sweden Democrats, which is reflected in their ownership of these issues. In environmental politics, the Green Party is considered by 48 % of survey respondents to have the best policies. On Twitter, this association was present, and the results show a strong correlation between issue ownership in the survey and on Twitter regarding environmental issues. Both in the survey and on Twitter, the Centre Party came second to owning this issue, with a big gap between them and all remaining parties.^[9] Immigration/integration is a polarizing topic in Sweden. Not surprisingly, the issue ownership over this issue is also divided: 25 % of survey respondents consider the Social Democratic Party to have the best policy, whereas 21 % consider the Sweden Democrats to have the best policy. On Twitter, the Green Party shared ownership with the Sweden Democrats in terms of receiving the most mentions with regard to discussions of this issue, which caused a weaker correlation.^[10]

Overall, there is a strong resemblance between a socially mediated issue-ownership dimension on twitter and ownership perceptions measured in the survey. In seven of nine political areas, the correlation between the two came out as strong and significant. Two topics, however, showed a clear deviation. These were immigration/integration and health care, where one or two parties were mentioned more on Twitter than what would be expected from the survey results. In the national election study, areas with relatively weaker ownership perceptions were immigration/integration, education, social welfare, pensions, and health care (compared to taxes, the environment, the labor market, and the economy). Areas with the weakest correlations between the survey and Twitter were immigration/integration and health care, followed by education. This result renders some support for Hypothesis 2 but cannot fully explain the differences in issue ownership between the survey and Twitter.

In understanding why these differences emerge on Twitter, it is relevant to examine these deviations a little closer. In the following, the two issue areas displaying the largest differences in ownership will be discussed in light of the Swedish 2014 election.

Issue ownership deviations between the survey and Twitter

Debates revolving around immigration/integration played a major role in the 2014 election. In the news media, topics concerning immigration and refugees featured most prominently during the election campaign for the first time ever (Johansson, 2017). A cordon sanitaire practice of isolating the anti-immigration party, the Sweden Democrats, was also upheld (Rydgren and van der Meiden, 2018). In terms of the liberal–authoritarian dimension, the Greens and the Left Party were most strongly opposed to the Sweden Democrats. Before the 2014 election, the Green Party campaigned on an anti-racist agenda. At the time, the Green Party’s secretary said the party was active in analyzing which issues resonated best on social media; the party noticed that they generated the most engagement on social media when they opposed the Sweden Democrats. This response then inspired the design of their broader campaign (Stockholm University, 2014). The engagement on social media did not translate into votes, however, and when seen from the perspective of issue ownership, a stronger focus on environmental issues in the election campaign might have garnered better results. Therefore, how the Green party themselves acted to put emphasis on integration/immigration and oppose the Sweden Democrats, in particular on social media, might explain why the party was mentioned frequently with this issue, which resulted in a weaker correlation between the survey and Twitter measurement.

In total, 20 % of the electorate viewed health care as one of the most important issues: It placed sixth on the scale of the most salient issues (Statistics Sweden, 2015). Party system attention to health care has increased in several countries, and parties tend to focus on this issue regardless of ownership perceptions, which make ownership less clear (Green-Pedersen, 2019). The high proportion of tweets mentioning the Christian Democrats in relation to this issue could therefore be explained by the party’s effort to compete over this issue during the campaign. Examples of this can be found in the data: When the Christian Democrats were mentioned together with health care, a large share referred to nationalizing health care, which was the policy position the Christian Democrats took to differentiate themselves from other parties’ position on the issue. The party’s own campaign efforts might therefore explain why they were discussed more in relation to health care on Twitter than would be expected given the ownership perceptions of this issue.