Measuring party positions and issue salience with mass media and manifesto data: Comparing PolDem with Manifesto Project data

Abstract

To what extent do measures of the issue positions and salience of political parties differ when they are extracted from mass media debates and election manifestos? Answering this question, the paper serves a dual purpose within this special issue of Party Politics: (i) introducing the PolDem election dataset used throughout the issue, and (ii) analyzing its convergence with the widely used Manifesto Project data. The newly released PolDem dataset, based on a relational content analysis of newspaper articles published during national election campaigns, covers 15 European countries and 111 campaigns. Focusing on four broad issue domains (economic, cultural, European integration, and political issues) across countries from Northwestern, Southern, and Central-Eastern Europe, the results demonstrate strong convergence in issue positions, except for in political issues related to democracy and corruption. However, the datasets differ significantly in issue salience, especially for less salient, noneconomic issues, niche parties and in less polarized contexts.

Keywords

party competition cross-validation mass media manifestos

Introduction

Party positions and issue salience are frequently mentioned in scholarly conversation, though their precise meaning remains contested. Both concepts are fairly abstract and cannot be directly measured, requiring researchers to rely on observable manifestations in parties’ activities (Benoit and Laver 2009: 57). Political parties publish manifestos and press releases, their representatives give speeches or vote in parliament, and voters express perceptions of political parties. Another approach to studying party competition is through mass-media debates, which provide rich information on how political actors engage in public contestation and how they are portrayed to the electorate (Helbing and Tresch, 2011).

The choice of information source is both a methodological and theoretical decision. As Charlot (1989: 361) has observed, every party has two faces: “a public face turned towards the media, the voters and the rest of the world, and an inward-looking face reserved for the initiated, activists, elected representatives and leaders.” Harmel et al. (2018) describe this as the contrast between a party’s internal identity and external image. While party and electoral research are often interested in a party’s public face, scholars tend to use measurements of issue positions and salience that may represent a mix of both faces—by looking at electoral manifestos, for instance. Recent studies have highlighted the inward-looking components of electoral manifestos, noting the various functions manifestos fulfill for parties (e.g., Dolezal et al., 2016; Eder et al., 2017; Harmel et al., 2018; Kercher and Brettschneider, 2013).

This paper introduces the PolDem National Election Campaign Dataset, a publicly available and systematically updated dataset covering 111 national election campaigns across 15 European countries from the 1970s to 2022.¹ The dataset builds on previous work by Kriesi et al. (2008, 2012) and Hutter and Kriesi (2019), expanding its geographic and temporal scope. The PolDem dataset is the main workhorse used in this special issue of Party Politics on “Electoral Mobilization in Turbulent Times” (Wang et al., 2026). The countries covered include:

• Northwestern Europe (NWE): Austria, France, Germany, Ireland, the Netherlands, Switzerland, and the United Kingdom.

• Southern Europe (SE): Greece, Italy, Portugal, and Spain.

• Central-Eastern Europe (CEE): Hungary, Latvia, Poland, and Romania.

For most Northwestern European countries, the dataset covers a reference election in the 1970s and all national elections from the early 1990s up to 2022. For Southern and Central-Eastern Europe, the dataset includes one pre-2008 election (before the Great Recession) and all elections since 2008 up to 2022.

A defining feature of PolDem is its mass media-based approach to measuring party positions and issue salience. Unlike datasets that rely on self-reported party positions (e.g., manifestos) or expert judgments, PolDem tracks how political competition unfolds in the public arena through a relational content analysis of the coverage of two newspapers for each country in the two months before Election Day. The dataset thus captures parties’ interactions with competitors, media framing, and responses to external events—factors crucial for understanding contemporary party competition (Meyer et al., 2020; Van Aelst and Walgrave, 2017). Compared to other media-based datasets, such as the Comparative Campaign Dynamics (CCD) dataset (Debus et al., 2018; Somer-Topcu and Tavits 2023), PolDem offers a much broader temporal and geographical coverage, offering the opportunity to use measures from mass media in crossnational and longitudinal studies on party competition and electoral behavior (for the latter, see Kriesi et al., 2026). Specifically, the dataset facilitates scholarly debates about similarities and differences between the two “faces” of a party and how these might affect the dynamics of party competition and vote choice.

The present article takes an important step in this direction by comparing the updated PolDem election dataset with the well-established and widely used data from the Manifesto Project, which analyzes election manifestos in over 50 countries (Lehmann et al., 2024). It builds on recent work comparing how political parties communicate across different channels (e.g., Huber et al., 2022; Ivanusch, 2024; Peeters et al., 2021; Tresch et al., 2018; Schwarzbözl et al., 2020). We chose the Manifesto Project dataset to cross-validate the PolDem measures over available expert surveys (e.g., the Chapel Hill Expert Survey CHES, see Jolly et al., 2022)² because both datasets refer to specific election campaigns and because we can compare campaigns from the 1970s onward.

We proceed in four steps. First, we present the main features of the PolDem dataset, including the time period and countries covered, the selection of newspapers, and the coding strategy. Second, we compare the media-based PolDem measures with those from the Manifesto Project, drawing on previous studies that examine the relationship between media and manifesto data (e.g., Helbling and Tresch 2011; Merz 2017a, 2017b). Our study advances this debate by covering a larger set of elections and countries than the six Northwestern European cases originally analyzed by Kriesi et al. (2008) and by expanding the focus beyond European integration issues to assess party positions and issue salience more broadly. Helbling and Tresch (2011) found that party positions on European integration tend to converge across datasets, while issue salience measures diverge—a pattern that aligns with our findings. In contrast, Merz (2017b), while not conducting a direct cross-validation, examined how manifesto-based issue salience relates to party-issue linkages in media coverage and found a significant relationship, shaped by parties’ long- and short-term issue strategies. In the third step, we build on these insights and use regression analysis to examine how issue, party, and contextual factors shape differences in issue salience between the PolDem and Manifesto Project data. Our results show systematic variation: divergence is greater in non-economic issues, among niche parties, and in less polarized settings. Finally, we conclude and suggest avenues for further research.

Introducing the mass media-based PolDem election campaign dataset

The PolDem election dataset used in this special issue covers 15 European countries. Seven are from Northwestern Europe (NWE: Austria, France, Germany, Ireland, the Netherlands, Switzerland, and the United Kingdom); there are four each from Southern Europe (SE: Greece, Italy, Portugal, and Spain) and Central-Eastern Europe (CEE: Hungary, Latvia, Poland, and Romania). As Table 1 shows, the currently available version covers 111 national elections in the 15 countries. The number of campaigns covered differs across countries because the data were collected within several past research projects conducted by the PolDem principal investigators (see Hutter and Kriesi, 2019; Kriesi et al., 2008, 2012). For all Northwestern countries apart from Ireland, the dataset covers a reference election in the 1970s and all national election campaigns since the early 1990s up to 2022. For all the four Southern European countries, the dataset includes one election pre-2008 (the moment when the Great Recession hit the European continent) and all elections from 2008 up to 2022. For Central-Eastern Europe, PolDem covers one election from the early 1990s and one shortly before 2008 as benchmarks as well as all elections from 2008 up to 2022.³

Table 1.

List of countries, elections, and newspapers covered by PolDem.

Country	Election year	Newspapers	# Core sentences
Austria*	1975, 1994, 1999, 2002, 2006, 2008, 2013, 2017, 2019	Die Presse, Kronenzeitung	20,190
France*	1978, 1988, 1995, 2002, 2007, 2012, 2017, 2022	Le Monde, Le Parisien	18,156
Germany*	1976, 1994, 1998, 2002, 2005, 2009, 2013, 2017, 2021	Süddeutsche Zeitung, Bild	18,919
Greece	2007, 2009, 2012 May, 2012 June, 2015 January, 2015 September, 2019	Ta Nea, Kathimerini	13,312
Hungary	1994, 2002, 2006, 2010, 2014, 2018, 2022	Népszabadság, Magyar Nemzet	16,303
Ireland	2007, 2011, 2016, 2020	The Irish Times, The Sun (Irish ed.)	10,681
Italy	2006, 2008, 2013, 2018, 2022	La Repubblica, Corriere della Sera	10,842
Latvia	1993, 2002, 2006, 2010, 2011, 2014, 2018, 2022	Latvijas Avīze, Diena*	13,381
Netherlands	1972, 1994, 1998, 2002, 2003, 2006, 2010, 2012, 2017, 2021	NRC Handelsblad, Algemeen Dagblad	17,855
Poland	2005, 2007, 2011, 2015, 2019	Gazeta Wyborcza, Rzeczpospolita	17,974
Portugal	2005, 2009, 2011, 2015, 2019, 2022	Público, Diário de Notícias	13,481
Romania	1996, 2000, 2004, 2008, 2012, 2016, 2022	Jurnalul National, Adevarul	15,040
Spain	2000, 2004, 2008, 2011, 2015, 2016, April 2019, November 2019	El Mundo, El Pais	15,115
Switzerland*	1975, 1991, 1995, 1999, 2003, 2007, 2011, 2015, 2019	Neue Zürcher Zeitung, Blick	22,224
United Kingdom*	1974, 1992, 1997, 2001, 2005, 2010, 2015, 2017, 2019	The Times, The Sun	17,677

Note. The dataset is publicly available at https://poldem.eui.eu/; * Countries covered by Kriesi et al. (2008); ** For Latvia, a smaller sample of the Russian-language newspaper Вести сегодня has also been coded to cross-check the results given the strong ethnic divide in Latvia (see Eihmanis, 2019).

PolDem’s key methodological choice is its decision to explore party competition through the lens of mass media coverage. This decision is due to an interest in publicly visible conflicts, that is, in party contestation as it evolves throughout a campaign and as the electorate perceives it. Research on transformations in advanced democracies has routinely noted the importance of the mass media in political opinion-making and decision-making processes (for a classical example, see Swanson and Mancini, 1996). Virtually all political actors try to gain public support through the mass media, and political statements mediated by journalists receive much public attention (Meyer et al., 2020; Schmitt-Beck and Farrell, 2008; Van Aelst and Walgrave, 2017). While parties are in full control of their direct party communications—e.g., by means of electoral manifestos—they do not control the mass-mediated public debate and are more likely to be forced to respond to their competitors’ activities and issues. This is due to (a) the active role of the media in selecting the content of news, (b) the strategies of all parties (and not just one) in shaping the media agenda,⁴ and (c) the impact of exogenous shocks (such as economic crises or catastrophes) on news coverage. Thus, a party does not solely control its “face” in the mass media; it might change quickly as, to borrow a term from the agenda-setting literature, there is lower “friction” and less resistance to change in the media agenda (e.g., Baumgartner and Jones, 1993; Walgrave and Nuytemans, 2009).

The PolDem data are based on newspaper coverage, because newspapers tend to cover politics in great detail and have a role as agenda setters (e.g., Druckman, 2005; Walgrave et al., 2008). Like other media sources, newspapers select stories for their news value according to certain criteria, especially conflict (e.g., Galtung and Ruge, 1965; Harcup and O’Neill, 2017), and they may also be politically biased. As Haselmayer et al. (2017) have shown, news value interacts with partisan bias, as the Austrian newspapers in their study were more likely to cover a party press release if the statement had high news value and the newspapers’ readership favored the party in question. Relatedly, for the Netherlands, Van der Pas et al. (2017) showed stronger agenda-setting effects of the media on the parliamentary agenda (and vice versa) when newspapers had a readership tied to the party under study.

To avoid the limitations of a single-source study, the PolDem election dataset is based on coding two newspapers per country (see Table 1). In Northwestern Europe, Kriesi et al. (2008) opted for the most widely circulated quality and tabloid newspapers published throughout their research period. Including a tabloid newspaper aimed to provide a broader perspective on how parties were portrayed during the campaign. In contrast, for Southern and Central-Eastern Europe, the dataset includes the most widely read center-left and center-right newspapers. This decision was driven by two factors: first, the distinction between quality and tabloid newspapers does not translate well to these media environments, and second, media systems in these regions tend to be more polarized than in Northwestern Europe.

As a result, users of the dataset today are constrained by the project team’s earlier decisions. The selection of newspapers reflects choices made at the time based on data availability and research priorities, which inevitably shape subsequent analyses. This introduces some cross-country imbalances—for instance, in the UK, both selected newspapers (The Times and The Sun) belong to the Murdoch-owned global media conglomerate, while in Germany (SZ) and Switzerland (NZZ), the most widely read quality newspapers lean toward the center-left and center-right, respectively, and one has to also consider over-time changes. Importantly, the dataset enables an analysis of such variations. Moreover, at the highly aggregated level used in this article, the project team’s experience suggests that these constraints do not significantly affect the overall conclusions.

To further analyze the impact of the selection of specific sources and certain coding decisions in PolDem, researchers can also draw on the Comparative Campaign Dynamics (CCD) Dataset (Debus et al., 2018), which covers 19 elections in nine countries—including 13 elections that overlap with PolDem. CCD similarly relies on newspaper content analysis to measure party competition, yet its focus is more limited in terms of temporal scope and methodological design. While CCD places emphasis on ideological cohesion and ambiguity by differentiating between how parties present themselves, how they discuss competitors, and how journalists frame party positions, PolDem aims for a broader longitudinal approach to study issue and actor-actor competition. However, CCD includes different newspaper selections for Germany, the Netherlands, and the UK, providing an additional reference point for cross-validation (see also Somer-Topcu and Tavits 2023).

Specifically, the PolDem dataset is based on sampling news articles published in the selected newspapers during the two months before national election days. The newspapers reported on the electoral contest and national party politics more generally.⁵ News articles in the sections on national politics and international politics were selected via an extensive keyword list (including party names, abbreviations and key politicians). Editorials and op-eds were excluded. Given the time-consuming coding procedure, Kriesi et al. (2020) relied on “chronological” sampling, which stratifies the sample throughout the campaign. The dataset covers a minimum of 2000 core sentences per campaign.⁶

Although the mass media is the data source, note that the PolDem dataset is not primarily used to study media attention or journalists’ perceptions. As previously stated, the focus is on party competition. For this reason, the dataset is based on so-called core-sentence analysis (CSA), a specific type of relational quantitative content analysis. CSA is based on the idea that every text can be represented as a network of relationships between “objects” or “nodes” (for the original formulation, see Kleinnijenhuis et al., 1997; Kleinnijenhuis and Pennings, 2001). Thus, the coding unit of CSA is not an entire article or a grammatical sentence but a relation between a subject and an object. Kriesi et al.’s (2020) adapted version of CSA is exclusively interested in the relationships between political actors and issues (i.e., so-called “actor-issue sentences”) or between two political actors (i.e., so-called “actor-actor sentences”) (for further details, see Dolezal, 2008; Dolezal et al., 2012; Gessler and Hutter, 2019). A grammatical sentence can contain none, one or several of these two types of core sentences.⁷ This means that each grammatical sentence in an article is reduced to its most basic “core sentence,” which contains only the subject (a political actor, in our case), the object (another actor or an issue), and the direction of the relationship between the two. The direction between actors and issues is quantified using a scale ranging from −1 to +1, with three intermediate positions indicating a “potential” or ambiguous relation.

As noted earlier, the PolDem data has been collected over multiple projects led by Hanspeter Kriesi and colleagues. To ensure reliability, at least two trained coders coded each campaign and most of them have been native speakers, highly familiar with politics in the respective country. All rounds of coding involved extensive coder training and supervision. In the latest update and expansion to Central-Eastern and Southern European democracies, for instance, coders participated in group and individual training sessions and had to accurately code ten English-language articles before proceeding. Early reliability tests showed that coders disagreed more on identifying core sentences than on coding specific variables—particularly at the higher aggregation levels of actors and issues used in this study. Consistent with findings from previous rounds of data collection (Dolezal, 2008; Dolezal et al., 2012), initial reliability tests showed an agreement of just under 80% for core sentence identification (Cohen’s Kappa = 0.76). To improve reliability, additional training and continuous monitoring were introduced, raising the coefficient above the typical 0.80 threshold. All analyzed variables at the aggregation level presented here exceeded this benchmark (>0.90 for the most aggregated issue domains and party affiliations). The same applied to the five-fold direction variable, where coders were instructed to use the three middle categories sparingly.⁸

Overall, the current version of the PolDem dataset covers more than 240,000 core sentences, with roughly two-thirds being so-called actor-issue sentences (again, see Table 1 for the number of observations by country). For all the observations, the dataset contains information about the type, the party affiliation, and (if available) the individual names of the actors (both subject and object actors). Moreover, the project has employed a very extensive list of issues and allowed coders to create new codes if necessary. After coding each campaign, the project team discusses new issues and decides how to aggregate them into upper-level categories for the analysis. The currently available dataset contains an issue variable with 17 substantive categories and a more detailed one with 105 substantive categories. This allows the user quite some flexibility in reaggregating issues.

Comparing party positions and issue salience based on mass-media and manifesto data

We compare the PolDem election data with the Manifesto Project data to cross-validate it and highlight its potential applications. The Manifesto Project serves as a benchmark because (a) it is the most widely used—and often the only available—dataset for studying party competition in a comparative and longitudinal perspective, (b) it is linked to specific national elections, like the PolDem data, and (c) it allows for the calculation of comparable positional and salience measures, at least for aggregate issue domains.

Following Bakker and Hobolt (2013: 30), we do not aim to determine which dataset serves as a “gold standard,” as both have limitations and serve different purposes—each capturing distinct aspects of party communication and competition. Consistent with prior research, we focus on convergent validity (Ray 2007: 12), assessing the extent to which measures of the same underlying construct align empirically. Given that media-based data primarily reflect the public face of party competition, whereas manifestos mix both outward- and inward-facing dimensions, we expect some divergence—both in general and depending on issue, party, and contextual factors (see below).

The test builds on Helbling and Tresch’s (2011) cross-validations, which provided a systematic comparison of manifesto, expert-survey, and media-based data (including the predecessor of the PolDem data as used by Kriesi et al., 2008). Yet, their analysis only covered the issue of European integration and the six original Northwestern European countries included in Kriesi et al.’s (2008, 2012) study. Their findings focusing on correlation coefficients indicated converging measures of party positions towards European integration but diverging salience measures across the datasets. In another study, Merz (2017a, 2017b) focused on salience only to better understand how the mass media link issues to parties in their news coverage. He merged the issue salience measures from the manifesto data with the PolDem data by Kriesi et al. (2008, 2012) for a broader range of topics but still for six of the Northwestern European countries only. With this different question in mind and adopting a regression analysis on the party-issue level, he came to a different conclusion than Helbing and Tresch (2011). He found a significant and positive relationship between the issues that parties emphasize in their election manifestos and the type of party-issue associations that were reported in the news. Moreover, Merz found that the strength of the relationship was conditioned by the long- and short-term issue emphasis strategies of parties as reflected in their manifestos.

In the following, we present simple correlation coefficients for the relationship between the PolDem and Manifesto Project data like Helbling and Tresch (2011), but we do so for the broader set of European countries (including countries from Central-Eastern and Southern Europe) and for four issue domains: European integration, economic issues, new cultural issues, and political issues. These are also central for the various contributions in this special issue (for an overview, see Borbáth, 2026). We matched the issue categories from the two datasets as shown in Table 2. We followed the approach proposed by Merz (2017a: 448) but eliminated references to social groups and issues that we classified as “vague.” In a second step, we aggregated them into three broad issue domains or dimensions, taking up the suggestions by Bakker and Hobolt (2013: 33) in their comparison of manifesto and other positional measures. We additionally introduced a fourth issue domain relating to political issues in a narrower sense, i.e., issues referring to democratic quality and corruption. We did so because these topics are at the core of political competition in Central-Eastern Europe and have also become more important in structuring party competition in Southern Europe in the recent crisis (Borbáth, 2026; Freire et al., 2020; Hutter et al., 2018). Notably, it is more difficult to match positions for this category, since most Manifesto Project categories are unidirectional (e.g., there is no support for corruption or opposition to governmental efficiency), and some categories have an ambiguous meaning. We thus conducted extensive robustness checks with alternative matching schemes.⁹

Table 2.

Matching of PolDem and Manifesto Project issue categories.

Domain	PolDem	Manifesto project
Economic	Economy, Economic liberalism, welfare	Left: 404 Economic Planning: positive, 403 Market Regulation, 405 Corporatism: positive, 406 Protectionism: positive, 412 Controlled Economy, 413 Nationalization: positive, 409 Keynesian Demand Management: positive, 504 Welfare State Expansion Right: 401 Free Enterprise, 407: Protectionism: negative, 402: Incentives: positive, 414 Economic Orthodoxy, 505: Welfare State Limitation
“New” cultural	Immigration, nationalism, environment, cultural liberalism	GAL: 604 Traditional Morality: negative, 103 Anti-Imperialism: positive, 106 Peace: positive, 107 Internationalism: positive, 607 Multiculturalism: positive, 602 National Way of Life: negative, 501 Environmental Protection: positive TAN: 603 Traditional Morality: positive, 109 Internationalism: negative, 608 Multiculturalism: negative, 601 National Way of Life: positive
Europe	Europe	108: European Integ.: positive; 110: European Integ.: negative
Political	Corruption, Democratic Renewal, Democratic Reforms, Regionalism	Pro: 201: Freedom and Human Rights, 202: Democracy, 203: Constitutionalism: Positive, 301: Federalism, 303: Governmental and Administrative Efficiency, 304: Political Corruption, Contra: 204: Constitutionalism: Negative, 302: Centralization

Overall, the following analysis covers 2016 party/election/issue combinations for 149 different parties across 98 elections. For the comparison, we had to exclude political parties that were not represented in both datasets.¹⁰ In addition, we decided to drop cases with less than 30 observations in the datasets. In this way, we avoided comparing measures based on very few coded news items and extremely short manifestos. Note that these decisions did not affect the general conclusions. Moreover, we had to drop France from the analysis, as it was the only country for which the PolDem data covers the presidential elections and not the parliamentary elections, as does the Manifesto Project.

The empirical results indicated a fairly strong convergence between manifesto-based and mass media-based measures when we considered the campaigns in all 15 countries covered by the PolDem data and the four issue domains together. As Figure 1 shows, we observed moderate to high correlation coefficients of r = 0.57 respectively r = 0.6 for the two key features of party competition, party positions and issue salience. The finding is consistent with related research that shows a fairly high degree of convergence in party messages across communication channels when analyzing a broad range of issues (e.g., Schwarzbözl et al., 2020; Tresch et al., 2018).

Figure 1.

Issue position and salience measures from mass media and manifesto data across issue domains.

But would we observe the same kind of measurement convergence when considering the four issue domains separately? Starting with issue positions, we were able to generalize the results from Helbling and Tresch (2011) beyond the issue of European integration and Northwestern Europe: Mass media data converged with alternative measures of party positions. The only exception concerned what we called “political” issues, which covered statements about the quality of democracy and corruption. Relying on the data for all the countries, we observed a high correlation coefficient for economic (r = 0.58), cultural issues (0.63), and European integration (0.55). This situation was different for political issues referring to questions of democratic quality and corruption, where both measures were only weakly correlated and there was a resulting correlation coefficient of 0.18. As visible in Figure 2(a)), this was due to the limited variation in terms of positions regarding political issues in the Manifesto Project data. Parties tended to adopt very positive positions towards democracy and good government in their manifestos, but we observed significantly more variation in how their position was covered in the mass media. As Ivanusch (2024) showed in his recent comparative study of Austria, Germany, and Switzerland, political parties are also much less likely to cover such topics in their manifestos than in their press releases, parliamentary speeches, and social media activity. Finally, note that the observed correlation coefficients for the positional measures did not differ much when we excluded the countries in Central-Eastern Europe and also the ones from Southern Europe (for details, see Table 3). The only exception was that convergence for political issues was even weaker when we focused on the seven Northwestern European countries only.

Figure 2.

Issue position and salience measures from mass media and manifesto data by issue domain.

Table 3.

Comparison of PolDem and Manifesto Project measures—correlation coefficients.

	All countries	Without Central-Eastern Europe	Northwestern Europe only
Positions
cross-issue domains	0.57
Economic	0.58	0.64	0.69
European integration	0.54	0.52	0.54
New cultural	0.63	0.66	0.66
Political	0.18	0.16	0.07
Salience
cross-issue domains	0.60
Economic	0.18	0.23	0.28
European integration	0.35	0.35	0.29
New cultural	0.38	0.41	0.40
Political	0.33	0.48	0.34

Turning to the convergence of the salience measures by issue domain, our results help to reconcile the contrasting findings by Helbling and Tresch (2011) and Merz (2017a, 2017b). As shown above, when we considered all four issues simultaneously, we found a very high degree of convergence between the two types of measures. Issue domains that dominated in the manifestos were also more salient in our media-based data. Note again that the correlation coefficient based on the data for all four issue domains was 0.57 respectively 0.6 for direction and salience. However, these correlations became much weaker or almost nonexistent when we focused only on within-issue variation. As can be seen in Figure 2(b), the salience of “new cultural” issues was still moderately correlated across the two datasets with a correlation coefficient of 0.38 (all countries), while the two datasets converged much less for economic (r = 0.18) and European (0.35) issues. In other words, Merz’s results refer primarily to differences in salience across the four broad issue categories—economic issues are generally more salient than cultural and political issues defined in a narrower sense—while Helbling and Tresch refer to differences within a single category (in their case, European integration). Note that no clear pattern emerged when we excluded the countries from Central-Eastern Europe and Southern Europe, respectively. While the correlations in relation to cultural issues tended to remain similar, they became slightly stronger for political issues related to the quality of democracy and corruption when we excluded countries from Central-Eastern Europe and became even weaker for European integration when we focused on the countries from Northwestern Europe only (see Table 3).

Why do issue salience measures differ between PolDem and the Manifesto Project?

Building on the previous analysis, we now go one step further and examine the reasons for the divergence between the salience measures from election manifestos and mass media coverage. To reiterate, these differences are likely due to the fact that, while parties do have control over direct party communication, they do not have complete control over the public debate and are more likely to be forced to respond to the activities of their competitors and the key issues of the day following the media’s attention logics. Importantly, we also acknowledge that the media are not neutral transmitters of information but actors in their own right, guided by editorial selection logics, commercial interests, and sometimes partisan preferences, which may further amplify or distort certain party priorities. To systematically explore the divergence, we focus on issue, party, and contextual factors.

Issue domains: First, we expected divergence to vary across issue domains. To begin with, we expected that the divergence would be larger for issues that were less salient or visible because their salience in public debates might depend even more on contingent campaign dynamics and external events. Relatedly, the literature on party competition has highlighted that parties not only compete with each other on the basis of their positions on established dimensions of party conflict but also struggle to shift competition to new, previously underrepresented issues (e.g. De Vries and Hobolt 2012; Kriesi et al., 2008; Rovny and Edwards 2012). We expected to find a larger divergence in the salience measures regarding nonprimary dimensions of political conflict, as political parties’ success in shifting public debate may have been more visible in the mass-media public debate, which may serve as a seismograph of emerging trends.

Party types: Beyond issue-specific dynamics, previous research suggested that party characteristics affect how parties are able to translate their priorities into mass media debates (e.g., Helfer and Van Aelst 2015; Merz, 2017a, 2017b; Schwarzbözl et al., 2020; Vos 2014). Specifically, we expected to observe a higher degree of convergence for parties that were able to set the election campaign agenda. Again, the key was the visibility in public debates—not of the issue domain but of the party under scrutiny. A party’s capacity to translate its priorities into public debate relates to several factors, the most important being size and incumbency (on the incumbency bonus, see, e.g., Hopmann et al., 2012; Green-Pedersen et al., 2017). In other words, large parties in general and incumbents in particular have an advantage over the competition in shaping the public discourse. By contrast, we expected smaller opposition parties to be more likely to be associated in the public debate with the issues they emphasize or even “own.” Relatedly, Ennser-Jedenastik et al. (2022) have argued that political parties with a high volume of communication (often well-resourced large parties) are more likely to be able to deal with new issues in addition to other topics (i.e., volume expansion), while parties with a lower volume of communication (often less-well-resourced small parties) have to give up “old” issues for “new” ones (i.e., issue substitution). Thus, reporting the position of a political party that is generally associated with the topic provides an important anchor for readers to interpret other actors’ positions and might also reflect the party’s campaign strategy. Hence, we expected the media-based measures to overemphasize a party’s specialization in a specific policy niche compared to the manifesto-based measures.

Divergence across contexts: Finally, we were also interested in the extent to which manifesto-based and mass-media-based measures of issue salience diverged across contexts. Specifically, we focused on two main dimensions of party systems: fragmentation and polarization. With respect to fragmentation, we built on the aforementioned size and niche arguments, because multiparty systems with many competitors are more likely to have such dynamics at play. This, in turn, should have led to greater convergence between what political parties emphasize in their manifestos and how they are covered in the mass media. By contrast, we expected to find greater degrees of convergence in party systems that were more programmatically polarized, because political parties and journalists alike would aim to highlight the differences given the news value of conflict and the clearer patterns of opposition in the party system.

For the following regression analysis, we constructed a divergence measure for each party-issue domain combination as our dependent variable. We opted for the absolute value of the natural logarithm of the Manifesto Project salience measure divided by the PolDem ones: divergence = abs(ln(salience_Manifesto_p,i/salience_PolDem_p,i)).11 Apart from focusing on relative differences across the two datasets, our logarithmic measure had the advantage that positive and negative deviations were of the same size.¹² We used a simple linear regression.

For the issue characteristics, we used the four domains introduced above—i.e., economic, new cultural, European integration, and political issues—as independent variables. We introduced the PolDem salience measure in the models as indicating the salience of the issue in the public sphere. For the party-level characteristics, we relied on the vote share of the party (as provided by the Manifesto Project dataset), the visibility of an individual party in the public debate (indicated by the number of coded statements in the PolDem dataset), incumbency status (based on PARLGOV and manual coding) and the squared Manifesto Project RILE measure (divided by 100), and the nicheness variable provided by Meyer and Miller (2015) on the basis of the Manifesto Project data. The contextual measures we used are the effective number of electoral parties from Emanuele et al. (2020) for fragmentation and the divergence based on the RILE index from the Manifesto Project for polarization (Lehmann et al., 2024).

Table 4 shows eight models that test the expectations we outlined in this section. Starting with those models that tested our expectations regarding divergence across issue dimensions (M1, M2), we found robust and statistically significant differences across issue dimensions. As we expected, in the first model, the divergence between PolDem and the Manifesto Project was smallest for economic issues. It was largest for European integration issues, followed by political issues related to democracy and corruption. Note that when we controlled for the salience of the issue in the public debate (M2), we found a decrease in the substantive effect size for the issue dimensions. The effect of the salience of an issue in the public debate was also as expected, with more salient issues showing a smaller degree of divergence.

Table 4.

Impact of issue and party characteristics on the divergence between mass media and manifesto-based issue salience measures.

	Models (Dependent variable: Divergence)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
European integration	0.537^***	0.433^***	0.438^***	0.434^***	0.435^***	0.374^***	0.376^***	0.359^***
European integration	(0.043)	(0.060)	(0.060)	(0.060)	(0.060)	(0.061)	(0.061)	(0.062)
New cultural	0.192^***	0.134^***	0.133^***	0.131^***	0.133^***	0.112^**	0.110^**	0.101^**
New cultural	(0.040)	(0.046)	(0.046)	(0.046)	(0.046)	(0.046)	(0.046)	(0.047)
Political (ref.=economic)	0.384^***	0.312^***	0.311^***	0.310^***	0.311^***	0.299^***	0.298^***	0.282^***
Political (ref.=economic)	(0.040)	(0.050)	(0.049)	(0.049)	(0.050)	(0.049)	(0.050)	(0.050)
Issue salience		−0.004^**	−0.004^**	−0.004^***	−0.004^**	−0.004^***	−0.004^***	−0.005^***
Issue salience		(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
Visibility			−0.0002^***
Visibility			(0.0001)
Incumbency				−0.068^**			−0.046	−0.056^*
Incumbency				(0.030)			(0.032)	(0.032)
RILE²				0.849^***			0.773^***	0.493
RILE²				(0.264)			(0.290)	(0.315)
Vote share					−0.003^***		−0.001	−0.0002
Vote share					(0.001)		(0.001)	(0.001)
Nicheness						0.261^***	0.193^***	0.204^***
Nicheness						(0.061)	(0.064)	(0.064)
Polarization								0.043^**
Polarization								(0.019)
Effective # Parties								0.014
Effective # Parties								(0.009)
Constant	0.467^***	0.594^***	0.656^***	0.591^***	0.650^***	0.624^***	0.637^***	0.502^***
Constant	(0.028)	(0.059)	(0.061)	(0.061)	(0.062)	(0.060)	(0.065)	(0.085)
Observations	1825	1825	1825	1825	1816	1695	1686	1669
R²	0.091	0.094	0.101	0.102	0.097	0.101	0.106	0.111
Adjusted R²	0.090	0.092	0.098	0.099	0.095	0.098	0.102	0.106
Residual Std. Error	0.625 (df = 1821)	0.624 (df = 1820)	0.622 (df = 1819)	0.622 (df = 1818)	0.624 (df = 1810)	0.601 (df = 1689)	0.601 (df = 1677)	0.599 (df = 1658)
F Statistic	60.866^*** (df = 3; 1821)	47.278^*** (df = 4; 1820)	40.662^*** (df = 5; 1819)	34.429^*** (df = 6; 1818)	39.081^*** (df = 5; 1810)	37.979^*** (df = 5; 1689)	24.976^*** (df = 8; 1677)	20.749^*** (df = 10; 1658)

Note: p < .1; p < .05; p < .01.

The next set of models (M3–M7) sequentially tested our expectations regarding party characteristics. Note that we first tested each expectation separately before assessing the joint model (M7), as many of these characteristics are in fact closely related. We first test party visibility in the media (M4) which significantly reduced the divergence between the two data sets. However, visibility can often be explained by factors such as incumbency and extremeness for which we observe a significant effect in the next model (M4). A larger vote share (M5) similarly reduces the divergence. In contrast, the nicheness of a party increased the mismatch between the mass media and manifesto-based measures (M6). In the joint model, M7, only the extremeness and nicheness measures remained statistically significant.

The final model, M8, includes all the issue and party characteristics plus the two variables on the contextual level—that is, the effective number of parties and the degree of polarization. The results show that there was only a statistically significant and positive association between polarization and divergence, suggesting that the more polarized the party system, the more convergent the salience measures tended to be. This ran counter to our expectation that, in a more polarized setting, manifestos and mass media debates would converge more, not less. Compared to our previous model M7, extremeness loses significance but incumbency is significant again (albeit at the p < .1 level only).

Conclusion

The article serves a dual purpose within the special issue of Party Politics on “Electoral Mobilization in Turbulent Times” (Wang et al., 2026). First, we have presented the main features of the PolDem election campaign dataset, which is based on a relational content analysis of newspaper articles. This dataset serves as the primary source for the various contributions in this special issue, allowing the authors to assess the changing structure and dynamics of party competition in contemporary European democracies. Second, we have compared the PolDem measures of party positions and issue salience extracted from mass media content with those derived from the Manifesto Project, a well-established and widely used data sources based on the coding of electoral manifestos. Both datasets refer to specific election campaigns and allow us to cover the 15 countries, in some cases from the 1970s up to 2022.

Our empirical analysis, as well as the broader applications of the PolDem dataset in this special issue, yield three main conclusions:

First, the empirical analysis demonstrates that the PolDem election campaign dataset captures a distinct and valuable dimension of party competition in Europe—the public face of party communication as reflected in mass media coverage. By using mass media coverage as a primary data source, the PolDem dataset allows capturing the dynamic interactions and contests that unfold during electoral campaigns in front of a wider public. As the contributions in the special issue show, the dataset allows researchers to trace long- and short-term changes in the structure and dynamics of party competition across the different European regions, especially in moments of accelerated political change such as the recent European polycrisis (Wang et al., 2026; Borbáth, 2026). Moreover, the dataset offers new ways to compare public debates—as represented in the mass media—with direct party communications, i.e., with election manifestos, as we do in this paper, or with press releases or social media content. Furthermore, the mass-mediated campaign can be linked to electoral outcomes (Kriesi et al., 2026).

Second, our comparison between PolDem and the Manifesto Project highlights both convergence and divergence in measuring party competition: There is a high degree of convergence in the measures of party positions for the issue domains of economic, cultural, and European integration. We only find strongly diverging position measures in the case of so-called political issues that refer to questions of democratic quality and corruption. In contrast, for all issue domains, the salience measures diverge strongly when based on mass media and electoral manifesto data. Our analysis thus allows us to generalize the results from Helbling and Tresch (2011) beyond the issue of European integration and beyond the space of Northwestern Europe. This suggests that while parties may converge in their positions on certain issues across communication channels, the relative salience of these issues in public discourse may vary significantly from what parties want to emphasize in their election manifestos. This offers important theoretical and empirical avenues for future research.

Third, our analysis identifies issue, party, and contextual factors driving differences in issue salience between the PolDem and Manifesto Project data. The divergence is larger for less salient, non-economic issues, while high-profile issues tend to align more closely. Moreover, party characteristics—especially vote share and factors related to visibility in the public debate—play a role, with larger and more visible parties showing less divergence across datasets. Notably, niche parties tend to experience greater discrepancies, as their media coverage is more focused on their key policy areas than their election manifestos. Finally, contrary to expectations, we find stronger divergence between PolDem and the Manifesto Project in more polarized settings. This might be due to the fact that many of the campaigns in our dataset took place during the Great Recession, when public debate was highly focused on a few dominant issues—sometimes even resembling referendum campaigns, as in the case of Greece’s 2012 and 2015 elections.

Overall, our study contributes to a deeper understanding of the complexities involved in measuring party positions and issue salience in contemporary democracies. By integrating mass media-based data with traditional manifesto data or expert surveys, researchers can gain more nuanced insights into the interplay between parties, media, and public opinion during election campaigns. Future research could further expand this approach by integrating additional mass media datasets, such as the Comparative Campaign Dynamics (CCD) dataset (Debus et al., 2018; Somer-Topcu and Tavits 2023), which offers alternative newspaper selections for Germany, the Netherlands, and the UK. Such comparative efforts would help refine our understanding of how media selection shapes measures of party competition and issue salience. More broadly, combining PolDem with other sources of party communication—including social media and television news—could further enhance our knowledge of how party strategies evolve across different media environments. These findings underscore the importance of analyzing both the public face and internal identity of political parties when studying party competition and electoral behavior.

Footnotes

Acknowledgement

The authors thank the guest editors and the other authors of this special issue as well as Christoph Ivanusch and the participants at the Manifesto User conference for their helpful feedback. Furthermore, they thank all team members involved in creating and updating the PolDem election campaign dataset hosted at .

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Theresa Gessler

Swen Hutter

Notes

Author biographies

Theresa Gessler is professor of political science with a focus on democratic political decision-making at the Department of Social Sciences, University of Hamburg. theresa.gessler@uni-hamburg.de

Swen Hutter is professor of political sociology and director of the Center for Civil Society Research at Freie Universität Berlin and the WZB Berlin Social Science Center. swen.hutter@wzb.eu

References

Bakker

Hobolt

(2013) Measuring party positions. In: Evans

de Graaf

(eds) Political Choice Matters: Explaining the Strength of Class and Religious Cleavages in Cross-National Perspective. Oxford: Oxford University Press, 27–45.

Baumgartner

Jones

(1993) Agendas and Instability in American Politics. Chicago: University of Chicago Press.

Benoit

Laver

(2009) Party Policy in Modern Democracies. Transferred to Digital Print. Routledge Research in Comparative Politics, 19 [i.e. 18]. London: Routledge.

Borbáth

(2026) Cleavages in party competition in Central and Eastern Europe. Party Politics, In this issue.

Charlot

(1989) Political parties: towards a new theoretical synthesis. Political Studies 37(3): 352–361.

De Vries

Hobolt

(2012) When dimensions collide: the electoral success of issue entrepreneurs. European Union Politics 13(2): 246–268.

Debus

Somer-Topcu

Tavits

(2018) Comparative Campaign Dynamics Dataset. Mannheim: Mannheim Centre for European Social Research, University of Mannheim. https://www.mzes.uni-mannheim.de/d7/en/datasets/comparative-campaign-dynamics-dataset

Dolezal

(2008) The design of the study: the distinguishing characteristics of our approach. In: Kriesi

, et al. (ed) West European Politics in the Age of Globalization. Cambridge University Press: Cambridge, 53–74.

Dolezal

Ennser-Jedenastik

Müller

Winkler

(2016) Analyzing Manifestos in Their Electoral Context A New Approach Applied to Austria, 2002–2008. Political Science Research and Methods 4: 641–650. Available at: https://doi.org/10.1017/psrm.2015.38.

10.

Dolezal

Hutter

Wüest

(2012) Exploring the new cleavage across arenas and public debates: Design and methods. In: Kriesi

, et al. (ed) Political Conflict in Western Europe. Cambridge University Press: Cambridge, 36–64.

11.

Druckman

(2005) Media matter: how newspapers and television news cover campaigns and influence voters. Political Communication 22(4): 463–481.

12.

Ecker

Jenny

Müller

, et al. (2022) How and why party position estimates from manifestos, expert, and party elite surveys diverge: a comparative analysis of the ‘left-right’ and the ‘European integration’ dimensions. Party Politics 28(3): 528–540.

13.

Eder

Jenny

Müller

(2017) Manifesto functions: how party candidates view and use their party’s central policy document. Electoral Studies 45: 75–87.

14.

Eihmanis

(2019) Latvia–An Ever-Wider Gap: The Ethnic Divide in Latvian Party Politics. In: Hutter

Kriesi

(eds). European Party Politics in Times of Crisis. Cambridge University Press: Cambridge, 236–258.

15.

Emanuele

Chiaramonte

Soare

(2020) Does the iron curtain still exist? The convergence in electoral volatility between Eastern and Western Europe. Government and Opposition 55(2): 308–326.

16.

Ennser-Jedenastik

Haselmayer

Huber

, et al. (2022) Issue substitution or volume expansion? How parties accommodate agenda change. Electoral Studies 76: 102437.

17.

Freire

Barragán

Coller

, et al. (eds) (2020) Political Representation in Southern Europe and Latin America: Before and After the Great Recession and the Commodity Crisis. London: Routledge.

18.

Galtung

Ruge

(1965) The structure of foreign news: the presentation of the Congo, Cuba and Cyprus crises in four Norwegian newspapers. Journal of International Peace Research 2: 64–90.

19.

Gessler

Hutter

(2019) The Media Content Analysis and Cross-Validation. European Party Politics in Times of Crisis. Cambridge University Press: Cambridge, 53–72.

20.

Green-Pedersen

Mortensen

Thesen

(2017) The incumbency bonus revisited: causes and consequences of media dominance. British Journal of Political Science 47(1): 131–148.

21.

Harcup

O’Neill

(2017) What is news? News values revisited (again). Journalism Studies 18(12): 1470–1488.

22.

Harmel

Tan

Janda

, et al. (2018) Manifestos and the ‘two faces’ of parties: addressing both members and voters with one document. Party Politics 24(3): 278–288.

23.

Haselmayer

Wagner

Meyer

(2017) Partisan Bias in Message Selection: Media Gatekeeping of Party Press Releases. Political Communication 34(3): 367–384. Available at: https://doi.org/10.1080/10584609.2016.1265619

24.

Helbling

Tresch

(2011) Measuring party positions and issue salience from media coverage: discussing and cross-validating new indicators. Electoral Studies 30(1): 174–183.

25.

Helfer

Van Aelst

(2015) What makes party messages fit for reporting? An experimental study of journalistic news selection. Political Communication 33(1): 59–77.

26.

Hopmann

Van Aelst

Legnante

(2012) Political balance in the news: a review of concepts, operationalizations and key findings. Journalism 13(2): 240–257.

27.

Huber

Bodlos

Graf

, et al. (2022) Disseminating legislative debates: how legislators communicate the parliamentary agenda. Party Politics 28(2): 365–376.

28.

Hutter

Kriesi

(eds) (2019) European Party Politics in Times of Crises. Cambridge: Cambridge University Press.

29.

Hutter

Kriesi

Vidal

(2018) Old versus new politics: the political spaces in southern Europe in times of crises. Party Politics 24(1): 10–22.

30.

Ivanusch

(2024) Where do parties talk about what? Party issue salience across communication channels. West European Politics 48(3): 618–644.

31.

Jolly

Bakker

Hooghe

, et al. (2022) Chapel Hill Expert Survey trend file, 1999-2019. Electoral Studies 75: 102420.

32.

Kercher

Brettschneider

(2013) Wahlprogramme als Pflichtübung? Typen, Funktionen und Verständlichkeit der Bundestagswahlprogramme 1994–2009. In: Wahlen und Wähler. Wiesbaden: Springer VS, 269–290.

33.

Kleinnijenhuis

Pennings

(2001) Measurement of party positions on the basis of party programmes, media coverage and voter perceptions. In: Laver

(ed) Estimating the Policy Positions of Political Actors. London: Routledge, 162–182.

34.

Kleinnijenhuis

de Ridder

Rietberg

(1997) Reasoning in economic discourse: an application of the network approach to the Dutch press. In: Roberts

(ed) Text Analysis for the Social Sciences: Methods for Drawing Statistical Inferences from Texts and Transcripts. Routledge, 191–207.

35.

Kriesi

Grande

Lachat

, et al. (eds) (2008) West European Politics in the Age of Globalization. Cambridge and New York: Cambridge University Press.

36.

Kriesi

Grande

Dolezal

, et al. (eds) (2012) Political Conflict in Western Europe. Cambridge and New York: Cambridge University Press.

37.

Kriesi

Grande

Hutter

, et al. (2020) PolDem-National election campaign dataset. Version 1. https://poldem.eui.eu/

38.

Kriesi

Wang

Altiparmakis

(2026) Electoral campaign effects: an aggregate analysis of electoral issue competition. Party Politics, In this issue.

39.

Lehmann

Franzmann

Al-Gaddooa

, et al. (2024) The manifesto data collection. Manifesto Project (MRG/CMP/MARPOR). Version 2024a. Berlin: Wissenschaftszentrum Berlin für Sozialforschung (WZB) / Göttingen: Institut für Demokratieforschung (IfDem). https://doi.org/10.25522/manifesto.mpds.2024a

40.

Manin

(1995) Principes du Gouvernement Représentatif. Paris: Flammarion.

41.

Merz

(2017a) Gaining voice in the mass media: the effect of parties’ strategies on party–issue linkages in election news coverage. Acta Politica 52(4): 436–460.

42.

Merz

(2017b) The Manifesto-Media Link: How Mass Media Mediate Manifesto Messages. Berlin: Humboldt Universität.

43.

Meyer

Miller

(2015) The niche party concept and its measurement. Party Politics 21(2): 259–271.

44.

Meyer

Haselmayer

Wagner

(2020) Who gets into the papers? Party campaign messages and the media. British Journal of Political Science 50(1): 281–302.

45.

Michel

(this issue). War in the campaign? Issue emphasis and issue salience during the 2022 French presidential campaign. Party Politics.

46.

Peeters

Van Aelst

Praet

(2021) Party ownership or individual specialization? A comparison of politicians’ individual issue attention across three different agendas. Party Politics 27(4): 692–703.

47.

Ray

(2007) Validity of measured party positions on European integration: assumptions, approaches, and a comparison of alternative measures. Electoral Studies 26(1): 11–22.

48.

Rovny

Edwards

(2012) Struggle over dimensionality: party competition in western and eastern Europe. East European Politics & Societies: And Cultures 26(1): 56–74.

49.

Schmitt-Beck

Farrell

(2008) Introduction: the age of non-party actors? In: Farrell

Schmitt-Beck

(eds) Non-Party Actors in Electoral Politics. the Role of Interest Groups and Independent Citizens in Contemporary Election Campaigns. Baden-Baden: Nomos, 13–24.

50.

Schwarzbözl

Fatke

Hutter

(2020) How party-issue linkages vary between election manifestos and media debates. West European Politics 43(4): 795–818.

51.

Somer-Topcu

Tavits

(2023) Message distortion as a campaign strategy: does rival party distortion of focal party position affect voters? The Journal of Politics 85(3): 892–904.

52.

Swanson

Mancini

(eds) (1996) Politics, Media, and Modern Democracy: An International Study of Innovations in Electoral Campaigning and their Consequences. Westport: Praeger.

53.

Tresch

Lefevere

Walgrave

(2018) How parties’ issue emphasis strategies vary across communication channels: the 2009 regional election campaign in Belgium. Acta Politica 53: 25–47.

54.

Van Aelst

Walgrave

(eds) (2017) How Political Actors Use the Media: A Functional Analysis of the Media’s Role in Politics. London: Palgrave Macmillan.

55.

Vos

(2014) Which politicians pass the news gates and why? Explaining inconsistencies in research on news coverage of individual politicians. International Journal of Communication 8: 2438–2461.

56.

Walgrave

Van Aelst

(2006) The contingency of the mass media’s political agenda setting power: toward a preliminary theory. Journal of Communication 56: 88–109.

57.

Walgrave

Nuytemans

(2009) Friction and Party Manifesto Change in 25 Countries, 1945–98. American Journal of Political Science 53(1): 190–206. Available at: https://doi.org/10.1111/j.1540-5907.2008.00365.x

58.

Walgrave

Soroka

Nuytemans

(2008) The mass media’s political agenda-setting power: a longitudinal analysis of media, parliament, and government in Belgium (1993 to 2000). Comparative Political Studies 41(6): 814–836.

59.

Wang

Borbáth

Altiparmakis

(2026) Electoral mobilisation in turbulent times: Party Politics.