SISP Conference 2024

The spread of misinformation on topics like climate change underscores the need for scientific expertise to correct public misconceptions. Trust in scientists and their messages is vital for science's role in shaping public opinion and tackling societal challenges. However, the growing politicization of science poses barriers to effective science communication, highlighting the need for evidence on how scientists can effectively counter misinformation. Prior research indicates that media consumers are more inclined to accept scientific information that aligns with their existing beliefs, with partisans showing particular receptivity to messages that resonate with their priors. Beyond these ‘message effects’, the perception of the message's source can also play a role in its effectiveness. However, research on this issue remains scarce. We add to previous scholarship on misinformation and public perceptions of science by examining the interplay between source characteristics and message framing in shaping the credibility and effectiveness of scientists' corrections of misinformation, integrating insights from scholarship on political framing and demographic status cues. Specifically, we focus on the impact of the scientists’ race (source effect), the political framing of their corrections (message effect), and the combined influence of these factors, in a survey experiment with 4,000 US Republican voters, targeting misinformation about green energy. This focus is motivated by the known skepticism among Republican voters toward green energy, a divisive issue, that stalls the transition from fossil fuels away. Moreover, the reduced trust in climate science among Republicans, partly due to politicized climate discourse, highlights the need for targeted communication strategies effectively engaging this group. We use a between-subjects experimental design, randomly assigning participants across six conditions (full-factorial 2x3 design). Participants are presented with a news article containing misinformation consistent with Republican viewpoints on either wind energy or electric vehicles, along with a correction from a scientific expert. We vary the expert's race (Black or white) and the political resonance of the frame of their correction (Republican, Democrat, or neutral). Additionally, a control group receives an article with misinformation but no correction, allowing us to evaluate the uncorrected misinformation's perceived credibility. By focusing on the micro-level mechanisms shaping the perceived credibility of corrections, our study contributes to a broader understanding of how the spread of misinformation might be halted.

In this study, we focus on the relationship between emergency measures violating democratic norms, the effectiveness of said measures, and citizens’ perceptions of democratic quality under a state of emergency related to the COVID-19 pandemic. Employing a conjoint experiment in Hungary, we also examine whether citizens’ evaluation of democracy was polarized along partisan affiliations. We find that citizens expect that, even under a state of emergency, a democratic country will proceed with scheduled elections, have a parliament fulfilling its legislative role and also scrutinizing the government, not restrict freedom of association, expression, and movement, reduce death rates and at the same time, ensure economic growth. Unlike opposition voters and undecided electorates, however, government voters are amenable to compromises to democratic standards such as restrictions on the freedom of movement, and a reduced role of parliament in the enactment of emergency legislations.

Substance regulation is in a state of flux in the United States. Past policies led to drug-related epidemics and social inequalities. At the same time, there is a momentum for drug liberalization. These changes often embedded in grassroots or direct democratic processes, and their proponents often justify them with a higher responsiveness to the public opinion. However, exploring the mass opinion on drugs presents empirical challenges: it is difficult to attribute certain individual or social harms, or the lack of those, to the substances themselves, or the regulatory instruments surrounding them. As a result, the public opinion might be plagued by a tendency we refer to as reputation bias. We propose a conceptual framework that distinguishes between expressed -- routinely measured in surveys -- and latent preferences defined as support for policies to regulate hypothetical drugs that mirror the attributes of existing substances. Using an innovative experimental design, we generate these distinct measures of preferences for 14 common legal and illegal substances. In line with our argument, we find a strong evidence that citizens' support for currently legal substances is overly high compared to corresponding latent preferences, while the opposite is true for illegal substances. Our results showcase the challenges of studying public opinion on contested and complex policies with high societal impact, and its potential role in a regulatory failure.

Lack of redistribution in the American context is often attributed to Americans’ propensity to think of individual mobility out of poverty as determined by the presence of opportunity rather than welfare assistance (McCall, 2013). This belief, however, is at least partly informed by how parties talk about these issues. Despite deep divisions between parties, the rhetoric on meritocracy and equality of opportunity is consensual, so that voters are not presented with competing interpretations of equality. Does the success of meritocratic narratives depend on the absence of competitive alternatives or rather on them being read as more convincing and closer to one’s own experience? This study tests whether frames of equality of opportunity are preferred to sufficiency of outcomes through two survey experiments testing the differential persuasiveness of narratives, by themselves and coupled with policies. Results reveal that the description of meritocracy as the prevailing American ethos is unjustified, and that a small majority prefers egalitarian narratives.

Politicians are not equally responsive to all strata of the population, which results in a phenomenon known as unequal representation. One explanation is that the low-income population is not visible enough to political elites. We argue that higher visibility can have detrimental consequences as by increasing the attention to the low-income segments of the population, we run into the risk of also increasing the salience of deservingness beliefs. This could have counter effects and make politicians listen less to the poor given that poor citizens are often seen as undeserving. We present evidence from two experiments with Swiss local politicians where the first study focuses on the link between visibility (priming on low income citizens) and attitudes while the second explores the mechanisms of deservingness beliefs in more detail by showing them vignettes with deserving vs undeserving individuals and testing how this affects their willingness to represent the entire population.

The mass-elite linkages are key to gauge the functioning of a polity, especially one that is in the process of forming like the European Union (Feld 1976; Bartolini 2005, chap. 6), and especially in times of crisis (Steenbergen, Edwards, and de Vries 2007). One of the ways in which elites (at different levels) can influence the public acceptance of policies is through cueing. First, the literature claims that such cueing works well in EU politics because people lack information on European policies and because these policies are less salient than national ones (Pannico 2020, Stoeckel and Kuhn 2018), but is cueing really stronger in EU politics than in domestic politics if compared directly and, if so, to what extent? Second, the literature emphasizes that national level actors do the cueing when it comes to EU politics (Ares, Ceka, and Kriesi 2017), but do EU level actors also have any cueing power in swaying the opinion of domestic publics and in what ways? This paper tackles these questions by examining the extent to which the public acceptance of new policies in the defence domain at the EU and Member States level is susceptible to cueing. To this end, I field a pre-registered factorial survey experiment in 16 EU countries (total sample of about 30,000 respondents) in May, 2023 varying three factors. The first factor introduces policy proposals which involve strengthening security and defence at either the European or the national level. The second and third factor vary the initiator of the cue (government/European Commission) and their (non-)support for these particular policies. Generally, I expect cues to work better for European policies given their low salience and cues from the government to have a stronger effect than cues from the European Commission following the idea of the second-orderness of EU politics (Gattermann et al 2021). With regard to the type of cue, I expect cues opposing a particular policy to have stronger effects than cues supporting a particular policy as they might trigger risk-aversion (in line with prospect theory Kahneman 2011, Kahneman and Tversky 1979). With regard to effect heterogeneity, I expect cueing effects to vary across socio-political groups in what regards previous attitudes on defence policy, trust in the EU and national governments, and country of origin. This paper speaks to the nature of mass-elite linkages in the EU polity as these have implications for both the legitimacy of the EU in times of crisis and the EU’s and domestic politicians’ capacity to build support around policies.

Climate change and the so-called green transition are increasingly moving from academic discussions to the political arena, sparking a growing number of conflicts of both ideological and distributive nature. This is not particularly surprising, as the urgency of slowing down global warming put governments under the pressure to reconsider their tax and spending priorities over time, presenting them with difficult trade-offs between environmental measures, macroeconomic strategies to sustain economic growth, and social policies that guarantee social cohesion. The Russian invasion of Ukraine, ensuing growing concerns about EU countries’ energy security and skyrocketing energy prices, dramatically renovated the urgency of addressing this multi-dimensional puzzle and related conflicts. The so-called ‘trilemma’ between fighting climate change, ensuring energy supply and economic growth, and protecting household disposable income is particularly suited to illustrate the complexity of and the interrelationships between these challenges. It also enables to better specify the emergence of different policy paradigms – the ‘growth first’, the ‘green growth’, the just transition’ and the ‘post-growth’ approach – that provide very different interpretations of the relationships between economic, social, and environmental goals by ranking differently the priorities that should be given among them. At the same time, our understanding of public opinion support for these paradigms remains limited, and we have even less insight into how different policy combinations contribute to increasing or decreasing consensus toward the green transition. Against such backdrop, this paper investigates the priorities and emerging divides within public opinion across EU countries around the ‘trilemma’ and related emerging approaches to solving it. To pursue this aim we employed an original conjoint experiment included in a cross-national survey fielded in 7 European countries in December 2022, when the energy crisis triggered by the Russian invasion of Ukraine reached its peak. The experiment is designed to allow respondents to evaluate a series of hypothetical policy packages that vary on three factors that represent the interrelated and potentially contrasting goals of the trilemma: the strategy adopted to ensure economic growth and energy security, the type of social compensation to face the negative externalities of the macroeconomic policy, and the source of financing. By randomly assigning respondents different attributes for each of the three factors, we can investigate whether there are specific combinations of measures that increase public support for the policy packages and whether respondents’ socio-economic differences and ideological leanings interact with the characteristics of the policy package in explaining public support for it.

The exponential evolution of Artificial Intelligence, has made this technology widely accessible and integral to many applications. LLM-based chatbots like ChatGP are now ubiquitous and likely to be highly consequential for content available online and how people interact with it. As such technology becomes an increasingly important source of information for voters, what effects are these developments likely to have on political attitudes and behaviour? Given the novelty of these models, their political effects are still unknown, though research on Voting Advice Applications (VAAs) and the effects of misinformation offer some clues. Though some voters are likely to be sceptical of the responses they receive from chatbots, or have more entrenched attitudes, younger or less highly educated voters may be more trusting and open to persuasion, from changing their mind about a given political issue to changing party support. We use an experimental research design to address these questions, taking attitudes and behaviour around the 2024 European Parliament Elections as our motivation. Respondents are assigned to one of three treatment groups designed to give them positive, negative, or neutral information about the work of the EU. Treatment is assigned via stratified randomisation to control for pre-existing EU attitudes and followed by a battery of political attitude and voting intention questions. We expect Euroscepticism and turnout to increase after interaction with the negative chatbot and decrease after interaction with the positive chatbot, particularly among younger and less highly educated voters. Our findings will demonstrate the level of threat or opportunity for democratic institutions which new AI technology creates.

Over the past three decades, Propensity to Vote (PTV) survey questions have been established as valid measures of electoral utility. Unlike traditional vote intentions, they provide perceived distances between parties and use a quasi-continuous scale. However, current research analyzes PTV with data reduction approaches, neglecting their full potential. This study proposes a belief system approach to PTV, conceptualizing them as behavioral intentions embedded in a belief system regarding vote choice. By analyzing PTV data from the European Election Studies (1989-2019; 28 countries, n = 121,959), we estimate network models for each national election. PTV items are rendered as nodes of a network whose weighted and signed edges indicate their unique shared variance. Findings reveal that characteristics of the national political field shape PTV network features. Party institutionalization is the best predictor of network connectivity. Affective polarization predicts the amount of negativity (i.e., the ratio of negative to positive edges) in the networks. Additionally, a higher ratio of negative edges predicts higher voter turnout for EU and national elections, demonstrating that PTV belief systems have downstream consequences on political behaviors. However, current research practices do not fully leverage these benefits, often simplifying PTV into binary constructs or applying data-reduction techniques such as multidimensional scaling. Therefore, the advances offered by the innovations in the measurement of electoral utilities were not accompanied by a throughout rethinking of the methodologies with which these new data ought to be analyzed. This contribution proposes to fill this gap by analyzing PTV with a novel network-based approach to multivariate data. Specifically, we posit PTV can be conceptualized as a network of interacting cognitive evaluations toward national parties. Adopting this strategy allows us to model these survey batteries as a weighted and signed network, where nodes represent variables, and edges their partial correlations. The higher-order construct emerging from these interactions is a political belief system embedding these behavioral vote intentions. Effectively, this modeling strategy shifts the focus from the levels of PTV (e.g.: the scores respondents assign to each party) to their relationships. Applying this methodology, we analyze the PTV of voters before each European election from 1989 to 2019, using data from the European Election Study. For each country-year unit in the database (142 cases circa) we estimate a partial correlation network between PTV items. Then, we craft a bipartite analytical strategy to fulfill two aims. First, we explore how the characteristics of each national political field shape the features of the PTV networks. In particular, we show that the connectivity of these belief structures is predicted by the levels of party polarization and party institutionalization of each country. Second, we test the associations between PTV’s network structures and turnout at the European elections. We find that a higher number of negative associations in a PTV network predicts greater EU and National turnout. These findings show that belief system structures are shaped by country-level political factors and that network characteristics, in turn, can have significant effects on voting behavior.

La relazione è volta a presentare e discutere l’applicazione di due approcci di ricerca qualitativi, come quello biografico e quello prosopografico, allo studio delle reti di attori delle politiche pubbliche. L’attenzione ai dati biografici non è una novità nell’ambito delle scienze politiche: al contrario, solo recentemente anche la prosopografia, che inizialmente si è sviluppata nell’ambito delle scienze storiche, ha iniziato ad attirare un qualche interesse. La prosopografia è un approccio di ricerca che punta sulla ricostruzione delle caratteristiche comuni di un determinato gruppo sociale. Da un punto di vista metodologico l’uso dell’approccio prosopografico presuppone l’identificazione di un gruppo target quanto più omogeneo possibile e la scelta di una serie di domande (una sorta di questionario) alle quali sottoporre le fonti per far emergere i dati di interesse. Una volta definito il campione e gli interessi di ricerca il lavoro si articola essenzialmente in due stadi, il primo dei quali corrisponde alla fase di raccolta dei dati relativi a ciascun individuo membro del gruppo selezionato mentre il secondo coincide con l’analisi delle informazioni aggregate. La prosopografia si avvale dell’approccio biografico ma non è sovrapponibile a esso, differenziandosi tanto per oggetto di interesse (caratteri comuni e non specificità) quanto per procedimento: la ricostruzione biografica è infatti solo il primo passaggio dell’approccio prosopografico, uno step propedeutico alla realizzazione di un database e alla lettura delle informazioni estratte. Dopo aver discusso i rapporti di contiguità e/o di interazione degli approcci biografici e prosopografici con alcune metodologie impiegate nell’ambito dello studio delle politiche pubbliche, come ad esempio la Network Analysis, la Sequence Analysis e il Programmatic Action Framework, la relazione fornisce alcuni esempi di applicazione empirica dei due approcci: nel far ciò dà conto di attività di ricerca che sono ancora in corso.

Political behavior scholars extensively explored how voters cast their votes during elections. However, the decision to vote for one party among the parties available during an election has always been studied as independent from the decision of not voting for another. Still, the preferences of each voter are not independent of each other. The relationship between the preferences that each voter has for the parties in a political system can be explored using social network analysis. This study considers the propensity to vote questions from the British Elections Survey (BES) and explores the relationship between the preferences of each voter employing network autocorrelations models. Network analysis can provide valuable new insights into how voters perceive parties and how the set of party preferences of each voter determines collective voting behavior. Theories explaining electoral behavior have been developed in isolation, reaching conclusions with limited validity. A network approach allows for testing concurring theories at the same time. Our findings show that these theories co-exist and that a network approach offers a comprehensive picture of people’s electoral behavior.

Once offered as a synthesis of the variable-oriented and case-oriented research traditions in comparative social science, QCA has been recently portrayed as harboring different rationales and called for a consistent diversification of its protocol. This contribution explores the diversification thesis to clarify if QCA research still appeals to a single standard or variations in the practice exist that align the analysis with some specific orientation.

Different disciplines ask why public sector corruption occurs, addressing diverse phenomena without engaging much with each other. Beyond checklists of variables, how different approaches and micro, meso, or macro levels relate to each other is seldom theorized. This article develops an integrated “Corruption Hexagon” model with six dimensions. The analytically relevant context provides a not directly causal background that influences the interplay of the pressure to act corruptly, the opportunity to benefit from corruption, the capability to exploit the opportunity, the supply of corruption, and the rationalization of one’s corrupt behavior. Using secondary data from 23 European countries, we operationalize the Hexagon to explain differences in procurement-related corruption risks. Results show that whereas the circumstances enable corruption, their interplay with personal characteristics or rationalization triggers corruption. The Hexagon offers a context-dependent, complexity-informed model for cumulative research applying different methods and theoretical assumptions about the agency underlying corruption.

Argument and content of the paper: Originally started in 1989 in Porto Alegre (Brazil), Participatory Budgeting (PB) has been spreading all over the world. In 2008 it had already reached hundreds of implementations in Europe (Sintomer et at. 2008). Since then, the number of European cities which implemented PB at least once has been growing. However, literature proved that on many occasions PBs are interrupted, and referred to instability as «a key element to understand participatory budgeting» (Alves and Allegretti 2012). This raises questions about the conditions for the irreversibility of the process. Indeed, one of the problems nowadays, with all participatory procedures, is that they tend to be ad hoc projects (Chwalisz 2019) which rarely manage to become part of communities’regular political cycles (Chwalisz 2019; Dryzek 2017). Or, as other authors put it, to become «ordinary government tools» (Allulli 2011, 444). That is, participatory policies are only rarely «institutionalized». This paper will explore under which conditions Participatory Budgetings manage to become institutionalized policies. The work will focus on the set of the 17 biggest Italian cities, for historical and theoretical reasons which will be fully explained in the final version of the paper. After determining the degree of institutionalization of PB in these cities in 2023, the paper will aim at answering the following research question: why is PB institutionalized in certain cities and not in others? Research methods and empirical material used: To anwer the above-mentioned research question, a Qualitative Comparative Analysis (QCA) will be carried out to detect the (combinations of) conditions necessary and/or sufficient for the presence (and absence) of the outcome. Conditions will be selected based on previous literature (both theoretical and empirical) on the topic and on the knowledge of the specific administrative-political situation in Italy. Processes of measurement and calibration for these conditions will be explained too. Raw data for the measurement and calibration processes will be collected through a number of different sources (official statistical institutions, local governments, research institutions).

The relevance of imagery in political communication has traditionally been a significant topic in academic studies. However, the rise of social media platforms has greatly amplified interest in this area in recent decades. In particular, the visual aspects of populist political communication online, and how visual communication relates to traditional textual communication also represent an underexplored area of this literature. This study employs multilevel regression models to analyze a dataset containing all Instagram posts published by the leaders of political parties represented in the ninth European Parliament -- from March 1st to August 31st 2019. We explore how populist and mainstream political actors communicate, and whether there is emotional consistency between their textual and visual communication. To detect these visual features, all images have been analyzed using automated face and emotion recognition-algorithms. These algorithms allowed us to quantify whether a picture contains a positive emotionalization (e.g., through smiles), or if instead it expresses a negative/neutral emotionalization. At the same time, we employ large language models (LLMs) to quantify the emotional content of the text accompanying each Instagram post. Our results show significant differences between populist and mainstream political actors in terms of their use of consistent (vs inconsistent) emotional communication. These findings contribute to the literature on emotional online communication by political actors and on populism as a political communication style.

Populists are known for their polarizing, aggressive, and uncompromising political style, and for rewarding a more combative behavior in their followers. Yet, virtually no evidence exists that aggressive populist rhetoric has an inciting effect among their followers towards greater support for political violence against political opponents. In this article, we present novel experimental evidence from representative surveys run in four countries with different history, political culture, and types of party systems (Argentina, Czechia, Germany, Italy), assessing whether exposure to populist negative campaigning on social media triggers greater support for political violence among their supporters. While we find only marginal evidence when looking at attacks in general, uncivil attacks from populists seem associated with a clearer uptick in support for violence among their followers – especially more severe forms of it (e.g., shooting political opponents). These results shed a new light on the intersection between populism, aggressive rhetoric, and radical partisanship, and have profound implications for a more comprehensive understanding of the roots of political violence in the public in times of surge of populism and populist political parties across the world.

This paper aims to analyze the evolution of party salience and positions on various thematic issues—namely illiberalism, nativism, populism, and sovereignism—in the communications of Brothers of Italy (FdI). Utilizing a novel dataset of FdI's official press releases from 2013 to 2023 and employing supervised machine learning tools, we detect changes and continuities over time. Specifically, we examine whether and how these dimensions were recalibrated as the party's prominence in the political system increased, culminating in Giorgia Meloni's victory in the 2022 national elections. Additionally, we investigate whether these fluctuations correlate with shifts in public opinion, measured using Google Trends data. The reasons for selecting FdI as our focus are twofold. Firstly, FdI presents a classification challenge: while its right-wing stance is clear, there is debate on whether the party led by Giorgia Meloni should be considered populist radical-right or nationalist/conservative. Secondly, the party has transitioned from a relatively marginal position within the Italian political landscape to a dominant one, securing a leadership role within the center-right coalition and a leading position in government. This shift offers a unique opportunity to examine how the party's transition from opposition to government has influenced its communication strategies.

This paper proposes a comprehensive assessment of content analysis performed by different Large Language Models (LLMs), in order to develop a gold standard protocol for using LLMs in content analysis research. The study uses a dataset comprising around 4,000 Facebook posts from political leaders in six countries (Italy, Spain, France, Greece, Germany, and the Netherlands), which have been human-coded for 18 variables related to populism content and communication style. The same corpus of Facebook posts and the identical codebook used by human coders—containing variable descriptions and instructions—were applied to four distinct LLMs: GPT-o, Gemini Ultra, LLama 3, and Claude Sonnet. The coding has been compared among the LLMs, as well as between the LLMs and human coders. To assess the performance of the LLMs, the consistency between LLMs and human coders was quantified using inter-coder reliability measures. In cases of discrepancies between human and LLM coding, the research team implemented a supervised assessment procedure to identify the correct response, simultaneously verifying the reliability of both human and LLM content analysis. The significance of this research lies in its potential to evaluate the capabilities of various LLMs, particularly in non-English languages, and to establish a standardized protocol for utilizing LLMs in content analysis within political and communication studies. This study aims to advance our understanding of LLMs’ efficacy in replicating human coding processes and their application in multilingual contexts, ultimately contributing to the methodological rigor in the field of computational content analysis.

The lively literature on audience costs (ACT) has recently enlarged its scope, considering how norm violations during war, historical rivalries, personalities of leaders and other factors influence trend in inflicted costs. Here, we stress the need for more realistic experiments on ACT, and we focus on a scenario in which Russia invades an Eastern European country. This is fundamental, we claim, to understand public opinion ‘here and now’ and to more effectively predict trends in public support for war in future crises. We run a survey experiment in six European countries and in Turkey, focusing on how NATO membership – whether the country invaded by Russia is a member – influences how citizens in NATO countries think their leaders should react. We control for additional factors such as the motivation of the Russian invasion, and anti-Russia reactions (e.g. sanctions, military aid, military intervention), and we explore as well sub-group heterogeneity on aspects such as political ideology, nationalism and cosmopolitanism, completing our investigation with a thorough analysis of open-ended answers. Our results provide engaging implications for studies on audience costs, public opinion on war, experimental approaches in Political Science, as well as relevant policy implications about public support for war.

Deliberative democracy promotes more inclusive procedures that improve democratic representation. Advocates of these tools, which range from deliberative polling to public assemblies, from online communities to citizen juries, argue that when citizens with different cultural backgrounds and viewpoints are invited to discuss political issues, this generates a number of desirable outcomes, such as recognition of others’ views, knowledge of issues, critical reflection on the complexity of the political process, and consequently democratic legitimacy and trust. While much of the research has focused on the relevance and impact of deliberation as far as the supply side is concerned, analyzing how to involve citizens, we know very little about the demand for deliberation, in particular from the standpoint of decision makers, and, specifically, about elected politicians’ interest or willingness in participating in time-consuming and potentially-high-cost deliberative meeting processes with constituents. While such deliberation may be normatively desirable, are politicians interested in engaging in deliberative meetings? Can political elites be encouraged to deliberate with their constituents? We argue that political elites can be incentivized to engage in deliberation. Reducing budgetary and other costs to organizing deliberative mini-publics would increase the likelihood that political elites are willing to engage in direct deliberation with constituents. To address these questions, we present the results of a field experiment conducted in Italy in 2023-2024 including a randomized intervention to encourage municipalities in organizing a deliberative mini-public to help in the policy-making process. We randomly assign Italian municipalities to treatment conditions and control conditions. Municipalities were randomized to treatment conditions encouraging deliberative assemblies with financial support and with technical assistance. Preliminary findings suggest that encouragement leads municipalities to engage in deliberative assemblies. The analysis presented in the paper will explore the effect of our encouragement stimuli in municipalities’ availability in involving citizens in the decision-making process. The study speaks to the literature on deliberation studying politicians’ availability to deliberate with their constituents by using innovative experimental designs, and contribute to the policy debate on how to encourage policy makers to improve representation.

n the past decade, scholars have conducted a number of survey experiments with the aim of assessing how public opinion about military interventions varies depending on a number of factors, most of which are associated with the intervention itself. However, the majority of these studies have focused on the public opinion of developed and democratic countries regarding interventions that their country is supposed to carry out. In contrast, our study adopts an alternative perspective by investigating support for military interventions in a developing country that has hosted such interventions by diverse actors. In particular, we examine whether the local perception of the military interventions depends on the identity of the external actor (Western/non-Western actors, individual states/international organizations). To test our argument, we plan to field a survey experiment in Mali, a country plagued by political instability and violence provoked by armed conflict, that has experienced a number of interventions in the past from different actors, including the United Nations, France, and the Wagner Group. The objective of this research is to contribute to the ongoing debate on the factors influencing the perception of military interventions and the role of Western and non-Western actors in global peacekeeping.

The relationship between human costs resulting from the use of force and public support for military intervention has been extensively explored in scholarly research. Whether characterised as a negative relationship, with human costs related to a monotonic decline in public support, particularly at the beginning of a conflict, or as a spurious relationship affected by substantive expectations and evaluations of a war, previous studies have predominantly focused on individuals’ responses to their own government’s use of force and military casualties, overlooking the analysis of approval or disapproval of military initiatives by third countries and the impact of civilian victims. Indeed, very little is known about civilian casualty sensitivity and the little we know is exclusively focused on the US case. To address this gap, we investigate in a comparative perspective how American and European citizens responded to the ongoing military crisis between Israel and Hamas in the Gaza Strip, based on available information about Israeli and/or Palestinian casualties during the conflict. By embedding manipulated versions of the Wikipedia page dedicated to the conflict in a survey administered to representative samples in France, Germany, Italy, the United Kingdom, and the United States between February 21 and March 1, 2024, we examine how public opinion support for the Operation Swords of Iron is influenced by the priming of civilian casualties suffered by Israel before its commencement and the number of Palestinians killed during the Israeli counteroffensive. By controlling for certain predispositions and ideological traits, we assess how support for military initiatives undertaken by foreign countries may vary depending on information priming effects among specific groups of citizens.

In the recent years, the International Relations (IR) scholarship has increased its efforts to collect relevant corpora of texts large enough to be analyzed through text as data methods, with the aim of addressing fundamental research questions in the field. Most of these corpora contain documents regarding the United Nations, including country statements at the general assembly and Security Council resolutions. In this context, the North Atlantic Treaty Organization (NATO), the most institutionalized alliance alive and a fundamental actor in global security, has been completely ignored. We address this literature gap by presenting a corpus containing all NATO press releases, the most recurrent form of communication performed by this organization available. We describe the main features of this dataset and then employ it to explore changes in how NATO articulated key concepts of international security across time. In doing so, we aim to provide a comprehensive and quantitative assessment of existing arguments about the evolution of NATO that thus far has often been tested in a qualitative way.

This paper explores the contention and agreement within the universe of Russian pro-war Telegram channels, demonstrating the extent of consolidation and dissent on issues related to the regim’s stability. We study the 20 most popular channels according to the number of subscribers. We collected all the posts from February 24, 2022, until October 14, 2023. First, to provide the overview of the discourse produced, we conducted structural topic modeling (Roberts et al., 2019) of the posts. Next, we use Latent Semantic Scaling (Watanabe, 2021) to measure the sentiment change of the pro-war channels towards main military actors from the Russian side. An essential aspect of our work involves the analysis of Yevgeny Prigozhin’s role on the agenda of pro-war channels. We employ a difference-in-differences methodology to explore the changes in content before and after his failed mutiny and subsequent demise by analyzing the forwards from Prigozhin-associated Telegram channels before and after his mutiny. The findings are expected to contribute to a deeper understanding of digital discourse in the Russian autocracy, concerning consolidation and conflict among the supporters of the decision to start the war in Ukraine, thereby offering insights into the regime's internal coherence and potential fault lines. Keywords: Latent Semantic Scaling; Topic Modeling; Network Analysis; difference-in-differences Russia; Ukraine; Telegram

Social scientists are increasingly focusing on the factors that explain the development, organisation, and enforcement of cybersecurity capabilities. However, a systematic analysis of these capabilities in terms of the policy instruments used has not been attempted so far. Which policy instruments can actors use? How do they vary between the institutions? How can the variation in the availability and use of capabilities be explained? This paper aims to answer these research questions using text-as-data techniques to analyse policy documents of EU Member States and European institutions. Reviewing the literature concerning the different phases of cybersecurity policy design, I develop a set of hypotheses within a general theoretical framework. I use mixed methods to cross-validate hypotheses and refine the theoretical framework. The data used are policy documents from 1990 to 2023, sourced from the online libraries of the UN, NATO, EU, and Member States, focusing on the cybersecurity policy sub-areas of Defence, Crime, Diplomacy, and Resilience. The identification of the policy instruments and the actors involved is conducted using Named Entity Recognition (NER) in the documents. Furthermore, I combine NER with a machine-learning method for estimating bureaucratic constraint variation to consider potential effects on implementation. Upon obtaining numerical values that represent the variety and frequency of policy instruments, I assess the consistency of the models corresponding to hypotheses through regression analysis. The findings offer a comparative analysis of state policy instruments within the EU’s supranational framework. The paper’s contributions are threefold: it provides a method for analysing the policies of individual states, considering the influence of supra-national frameworks, replicable in other policy areas; this methodological contribution follows the development of a theoretical framework that combines the literature of IR, policy analysis, IPE, and organisational studies in the field of cybersecurity; and it provides a systematisation of European cybersecurity policy documents.

International Relations theory argues that International Organizations (IOs) promote cooperation among their members. Practically, however, states do not engage with all peers to the same extent. On what basis are partners selected? Although interstate cooperation is mainly assessed through co-voting patterns in the case of the United Nations Security Council (UNSC), such cohesion is often accidental and requires little interaction. Contrarily, this article argues that co-sponsorship of draft resolution offers a more reliable measure to assess cooperation between two states when steering UNSC decision-making. Contradicting alliance theory, it posits that delegations looking to author a draft resolution tend to seek partners with similar characteristics. Cooperation of like-minded states limits coordination costs, reduces the need for concessions and eventually enhances the prospect of mutual benefit compared to collaboration in asymmetrical relations. I test the argument based on 16,0000 co-sponsorship dyads on complex humanitarian emergencies for 1990-2019. States contributing to the drafting of resolutions were retrieved through the automated text scraping of each individual document in combination with qualitative coding. Measuring similarity on four dimensions (regime type, wealth, regional affiliation and membership in political-economic groups) in a regression series, I find supportive evidence for the hypothesis. Moreover, distinct dynamics of in-group cohesion exist across regional or political-economic clusters. This analysis contributes to the current literature on the inner life of the UNSC, evidencing how global networks of state cooperation are shaped by domestic factors.

Turkey's foreign policy has undergone a remarkable transformation under the two-decade rule of President Recep Tayyip Erdoğan. Studies often provide distinct periodizations and foreign policy categorical postures adopted by Turkey during these years. This paper delves into this dynamic by employing text-as-data and computational methods. First, we construct a novel dataset of speeches by President Erdogan between 2014 and 2024. Secondly, we trace the evolution of Turkey's foreign policy under 20 years of rule by using a semi-supervised topic modeling and identify relevant trends in Turkish foreign policy. Third, we assess whether there is a congruence with previous analyses and periodicizations provided by scholars on Turkish Foreign Policy and provide a more nuanced account. The paper makes several contributions by proposing an original dataset that is absent in the literature and which can advance further analyses of Turkish foreign policy, leadership traits and geopolitical postures. Furthermore, we provide a novel methodological approach, namely using text-as-data in foreign policy and security studies.

Contemporary political communication is increasingly characterized by the integration of textual, visual, and many other modalities. In this study, we present a methodology for computational social sciences to understand multimodal political communication, specifically communication employing both textual and visual elements. We propose an approach to simultaneously represent and analyze these components, thus improving the measurement of political phenomena, and offering deeper insight into political communication strategies. We use multimodal embeddings as a unified distributed representation capable of capturing the intertwined relationship between linguistic and visual elements in political messages. To achieve this, we leverage computational text analytic techniques to capture the semantics embedded within post captions. Simultaneously, we employ computer vision models to capture visual elements from accompanying images, embedding the visual narrative that supplements political discourse. We suggest that such a multimodal representation can provide a more comprehensive understanding of the communication strategies employed by political actors than unimodal models. We illustrate the utility of this approach by analyzing the communication adopted by political party leaders on Instagram in the context of the EP 2024 campaign.

This paper purports to enrich the burgeoning field of research on the content of people's beliefs about inequality by studying the structure of these beliefs. We develop a theoretical and methodological framework that combines Correlational Class Analysis and Exploratory Graph Analysis, and we test it empirically with original survey data collected in the United States and the Netherlands (n = 2,501 and 1,618). Using CCA, we identify groups of individuals who share construals of inequality, while EGA allows us to model these structures as inequality belief systems, which are networks of perceptions, explanations and attitudes about inequality. Results reveal the presence of two distinct belief systems in each country. These systems exhibit structural differences and are related to different sociodemographic factors in the U.S. and the Netherlands. Moreover, we show that inequality belief systems are more socially patterned in the former country. Finally, we demonstrate that belief systems, in both countries, are associated with different levels of support for redistribution. We discuss the significance of our findings for the politics of inequality and stress that overlooking attitudinal structures impedes a full understanding of people’s views on inequality and their support for redistribution. This contribution paves the way for future inquiries into other attitudinal structures and provides valuable insights into studying and measuring belief systems. The paper has been recently accepted for publication in Social Indicators Research.

Whether experiences of violence can affect political behavior in post-war period has been extensively investigated within the literature on legacies of political violence. Scholars have generally focused on estimating political outcome via electoral preferences, measured as the vote share for left/right parties in post-war elections. Yet, while literature has produced significant theoretical and methodological contributions to measuring violence effects on electoral behavior, little attention has been given to whether experiences of violence can affect political polarization in post-conflict period. Measuring political polarization and its variation over time -- i.e., before and after the violent event -- presents several challenges and pitfalls for electoral research due to the availability and reliability of data. This paper aims to fill this gap by assessing whether and how episodes of war-related violence during the Italian civil war (1943 – 1945) can affect political polarization in the 1948 Italian national elections. Moreover, this research attempts to push the agenda further by measuring the effect of local experiences of violence on local variations of political polarization, providing diverse way to measure polarization, focusing on the Italian national elections held in 1921 and 1948. Therefore, this paper presents descriptive and correlational analyses of variations of political polarization in the 1948 national elections at the provincial level, thereby illustrating the historical data and methodological choices. It then employs comparative method to analyze variations in political polarization between provinces, testing the external validity of findings. Additionally, the paper examines within-province variations by localizing episodes of violence and their effects at the municipal level. By examining political polarization as an effect of wartime on peacetime politics, this paper shows that measuring pre- and post-conflict political polarization, along with its local variations, can significantly enhance the methodological toolbox of legacy studies, thereby improving our ability to analyse and use historical data.

For decades now, Western development agencies and donors have been castigated for their colonial biases (Easterly 2006) in foreign aid allocation to Africa. As literature extensively shows, donors provide considerably more foreign aid to their former colonies than to other countries (Chiba and Heinrich 2019). Extending this logic, we should expect foreign aid in aid recipient countries to be largely concentrated in favor of the former colonial powers over time. However, as development practice would suggest, we instead observe a rapidly growing number of new donors providing aid to African countries. This puzzle points to an important question: What is happening to colonial legacies in foreign aid over time and are former colonizers increasing or declining in importance across the continent? Furthermore, what variations can be observed amongst these former-colonizers-turned-donors? Political science literature has been unable to answer this question owing to a crucial issue: afterall how does one ‘measure’ persistence of colonial legacies? In my paper, I develop the concept of a ‘donor concentration index’ to measure how concentrated aid is in each aid recipient country, and what this means for the extent of influence that the former colonizer can exert. I find that persistence of colonial legacies, measured this way, points to a secular decline over time, with legacies in former French colonies declining faster relative to countries formerly colonized by other colonial powers. The concept of donor concentration has been applied in some papers in development economics, but not to the question of colonial legacies. Donor concentration serves as a suitable proxy for measuring persistence in this context because one of the most straightforward ways in which donors exert influence is through funding volume. If the aid volume received by a country in a given year is solely driven by 1 or a handful of donors, especially the former colonial power, it speaks to the donor’s ability to exert control and influence. However, if the total aid received by a country is provided by a range of donors, no single donor can exert complete control. In assuming that the former colonial power will typically be one of the largest (if not the largest) donor in any aid recipient country, I conceptualize an increase in donor concentration as an increase in the persistence of colonial legacy and decrease in donor concentration as a fall in colonial legacy. I measure donor concentration through the Herfindahl–Hirschman index (also called HH Index or HHI). The HHI is a commonly used measure of market concentration and is calculated by squaring the market share of each donor in the donor ‘market’ and then summing the resulting numbers. The index approaches 0 when a market is occupied by several donors of relatively equal size and a maximum of 10,000 when the market is controlled by a single large donor. The ‘market’ here is conflated to the ‘total volume of aid received by an aid recipient country’. I calculate each donor’s market share within every aid recipient country by year, and then calculate the HHI for every country/year combination, and run regression analyses alongside other covariates. This index is developed using data from the OECD-CRS dataset. The index underwent several iterations owing to challenges including poor quality of historical observations, specificities of how aid projects are coded, limited reporting by certain donors, and the restrictive scope in terms of not capturing subcontracting information. Despite these issues, I argue that it is possible to develop such proxies, which can serve as starting points to systematically analyze fundamental, yet understudied, questions in foreign aid research. Having said that, I also flag the limits of such an index and qualify how the findings are critically driven by decisions made during the construction and analysis of the index.

Despite substantial evidence showing that increasing spatial inequality explains geographical political polarization, we still know little about how local economic shocks travel and translate into perceptions of the economy and the labor market. Theories linking local economic shocks with anti-globalization responses often do not provide evidence on a key mechanism: when and how individuals perceive these economic shocks and develop economic concerns. Moreover, existing evidence favors the view that people form political views through sociotropic lenses, while little evidence contrasts egotropic and sociotropic labor market concerns simultaneously. We theorize that individuals' spatial and social 'proximity' to local economic shocks heavily influences their information and concerns. To test this hypothesis, we conceptualize and test measures considering various demographic and spatial factors. Specifically, we match geolocated panel survey respondents with geolocated economic shocks and operationalize 'proximity' by various weighting schemes, such as using average commuting distances and population density in the local area. The preliminary results show that the translation from local economic shocks to concerns about the economy and political views might be more complex than expected and lead to potential spatial and social issues not accounted for in existing research. At the same time, we offer new approaches to understand properly the spatial and social effects of local economic shocks.

A territory’s level of exposure to challenges and the features that can heighten the impact of crises is commonly referred to as Vulnerability. Measuring vulnerability requires disentangling the concept across several dimensions, which usually are the socio-economic and the environmental ones. Nonetheless, the political-institutional dimension of vulnerability stands out, since it determines a territory’s ability to effectively deliver policies and provide timely responses to crises through effective and democratic institutions (EC, SFR 2020). And it is instead, more often than not, neglected. This article aims to fill this gap by constructing a novel composite index of territorial vulnerability based on all three dimensions (environmental, socio-economic, and political-institutional). To this end, we conducted a systematic literature review (SLR) focused on empirical studies on the concept of vulnerability – and also on the complementary concept of resilience - when directly linked to a territorial dimension. The SLR covers different areas of study and applies the PRISMA approach to a selection of 261 academic articles. A quantitative and qualitative analysis of these papers is conducted to identify definitions of vulnerability and resilience, and highlight some key features of the scientific production on these concepts - including the types of crisis/external event they are usually linked with, the territorial level more often analyzed, the method employed, etc. Furthermore, we aim to detect some possible gaps that could be filled to include the political-institutional dimension in a more comprehensive understanding of the concept of territorial vulnerability. Further, one of the main goals of this SLR is to identify effective indicators used by previous studies to build a novel compounded index that encompasses the socio-economic, environmental, and primarily the political-institutional dimensions of the vulnerability of territories. Our systematic review revealed that these indicators must be directly related to the event or the crisis the region, the municipality, or the country is vulnerable to. Accordingly, in this paper, we present different indices to measure the vulnerability of Italian municipalities to current and future challenges (such as health emergencies, green transition, and environmental hazards, with a specific focus on hydrogeological risk). These indices are composed of both general, broad indicators of social, economic, ecological, and political vulnerability, and other socio-economic, environmental, and political indicators, which are instead ‘specific’ to the particular hazard or risk. Thus, these indices - as well as the methodology to build them, discussed in this article - can be applied and adjusted to assess a territory’s vulnerability to several other challenges and countries, thus being highly up-scalable and replicable.

Given the challenges to contemporary democracies from populism, illiberalism and authoritarianism, it seems crucial to understand the sources of general attitudes towards democracy. We understand these attitudes broadly here. They are normative views on democracy, satisfaction with democracy, political efficacy or trust in political institutions. Intergenerational differences are perceived as one of the factors explaining attitudes towards democracy. Theoretical considerations indicate that intergenerational differences result from early political socialisation and its period (Mishler and Rose 2007). In particular, it may be relevant whether the first political experience occurred when the country had a democratic or an authoritarian regime (Neundorf 2010). Theoretical considerations also consider "the living through model" according to which attitudes towards democracy depend on the time of exposure to a democratic regime - living more years in the democratic regime should induce stronger support for attitudes consistent with the regime ideology (Tucker & Pop-Eleches 2017). This paper will present empirical analyses to test hypotheses about the influence of the period of early socialisation and life experiences on broader attitudes about democracy. To date, analyses of intergenerational differences in attitudes towards democracy have been conducted with reference to a single country or group of countries. Our study will take advantage of the analytical possibilities offered by the ESS data and carry out analyses in a broad comparative framework covering European countries. Our analyses will focus on the attitudes of groups whose early political socialisation occurred in the period of authoritarian regimes, i.e. the older generation in the post-communist countries and some southern European countries (Spain, Portugal). We will also look at the attitudes of those whose beginnings of political consciousness were at a time of democratic backsliding in their countries, i.e. the young generations of Hungarians and Poles. We will contrast the attitudes of these groups with the views of other generations in their countries and countries with a long tradition of stable democracy. We will analyse data collected within the ESS rotating module "Understandings and evaluations of democracy" (2012, 2020) and data from the ESS core questionnaire on issues such as political interest, political efficacy, political trust, political participation, and satisfaction with democracy. The analytical technique used in the paper will mainly be Age-Period-Cohort Analysis.

The relationship between religion, religiosity and party politics has always been a central topic for both political sociology and electoral studies. Despite many studies claiming that the link between religious identities and party choice has become increasingly weaker in the last decades, the role of religious beliefs, practice and identity for voting behavior remains important for at least two main reasons. First, for a contextual issue. If secularization processes triggered a deep evolution in how people deal with religion and develop their own beliefs and practice, such processes advanced at a different pace in different institutional and political contexts, especially in the West. Therefore, we should expect the relation between religion and party choice to work differently in different contexts. Second, for a party competition reason. The relationship between individual religiosity and voting choice has increasingly gained salience with the growth of radical right parties (RRPs) throughout Europe. Such electoral successes have progressively called for a reconceptualization of the effect of religiosity on voting behavior. On the one hand, the radical right family is tightly connected to conservative and traditionalist issues. On the other hand, religious voters tend not to massively support them, or at least most regular churchgoers seem to prefer more moderate alternatives. Indeed, more recent research shows an increase of religious mentions within conservative and RRPs’ political programs, as they use them in a traditionalist frame or by shaping an “us vs them” rhetoric against immigration, especially against Muslims. Using the latest ESS wave dataset, our research aims to investigate the link between religiosity and support for RRPs across European countries by proposing a typology which combines both religious denomination and various other religious dimensions. Specifically, we focus on the role of “Cultural Christianity” in shaping support for RRPs. By cultural Christians we refer to people who identify as Christians (being them Catholic, Protestant or Orthodox) even though they do not practice their faith regularly nor they coherently follow religious commands; still they cling to religion as a repository of traditions and values. In doing so, we aim to bridge a gap in the electoral behavior literature, which perhaps prematurely deemed religiosity as irrelevant, by shedding light on the category of people who identify as Christians albeit not practicing nor explicitly adhering to religious dictates. Additionally, capitalizing on the heterogeneous contexts offered by ESS dataset, we explore cross-national variations especially between European countries to reveal the presence of cultural differences.

Multi-purpose international surveys are generally conducted in different national contexts simultaneously, ensuring comparability in a global context. Nonetheless, this is a sub-optimal situation for political research, especially when focusing on party preferences. These surveys often occur at different points in the national electoral cycle, which can significantly influence respondents' answers. The validity of the measures can vary depending on the proximity to the last national election, and the number of non-responses can also fluctuate. Adding to this complexity, different surveys employ various methods to gauge party preferences. For instance, the European Values Survey 2017 uses party identification, the World Values Survey includes a question on prospective voting, and the European Social Survey focuses on vote recall. This paper describes the outcomes of these different approaches and compares them with the actual results of the previous and following elections. Our expectation is that the distance from the previous or next election decreases the validity of the measures, while the number of non-responses increases significantly among respondents who are not interested in politics. We conclude with an appeal to use different measures to detect the so-called party utility, in order to de-seasonalize the measurement of party preferences.