Menu
WORDPRESS USER REQUESTED

Convegno Sisp 2017

Sections and Panels

Section 11. Metodologia della ricerca (Research Methodology)

Coordinators: Luigi Curini (luigi.curini@unimi.it), Pierangelo Isernia (isernia@unisi.it)

L’attuale dibattito metodologico in letteratura vive una stagione di straordinaria vivacità, testimoniata dagli sviluppi di tecniche sempre più sofisticate per una più precisa inferenza causale; dal crescente interesse per l’approccio Bayesiano alla conferma di ipotesi esplicative; dalla discussione su potenzialità e limiti del disegno sperimentale e sulla sua traduzione configurativa e Booleana (QCA); dalle sfide poste dai Big Data per la ricerca sociale; dalla discussione sul rapporto fra impianto teorico, disegno della ricerca e metodi, fra standard quantitativi e qualitativi, fra diverse nozioni di causalità e il loro retroterra ontologico. Dato questo quadro di sfondo, l’obiettivo di questa sezione di studio è favorire la discussione metodologica sulla ricerca empirica e sugli usi creativi degli approcci esistenti, sia nella Scienza Politica sia in altre discipline. Lo scopo è richiamare l’attenzione sulle scelte di metodo degli studiosi attraverso uno scambio di idee che parta dalle singole esperienze di ricerca, coniugando panel e paper che condividano prospettive ed approcci diversificati. Pertanto si invitano a presentare proposte di panel che affrontino i differenti problemi di metodo nelle diverse fasi del processo di ricerca (dalla dimensione ontologica ed epistemologica del disegno della ricerca; agli studi comparati; alle tecniche di analisi statistica dei dati). A tal fine, promuoviamo anche la partecipazione di candidati al dottorato e la discussione di ricerche in corso allo scopo di fornire elementi di riflessione sui lavori in itinere.
 

Panel 11.2 Tavola Rotonda: La ricerca empirica nelle Relazioni Internazionali. Metodi e prospettive a confronto


Negli ultimi due decenni la ricerca empirica nell'ambito delle Relazioni Internazionali ha acquisito notevole forza. I temi di ricerca spaziano, andando a intersecarsi con la politica comparata, l'economia, l'analisi delle politiche pubbliche. Allo stesso modo, i metodi utilizzati dai ricercatori nei loro studi si sono diversificati e raffinati. Riteniamo sia importante un momento di riflessione sul rapporto fra teoria e ricerca empirica, così come un confronto fra ricercatori che lavorano prevalentemente con metodi quantitativi e ricercatori che lavorano prevalentemente con metodi qualitativi, per evitare la formazione di compartimenti stagni e lavorare alla complementarietà degli approcci. Proponiamo dunque una tavola rotonda che metta a tema il percorso svolto fino a ora dalla ricerca empirica nelle Relazioni Internazionali, proponga riflessioni sulle direzioni future e favorisca un confronto costruttivo fra esperti di metodi diversi.
I partecipanti alla tavola rotonda che hanno già confermato la propria disponibilità sarebbero i seguenti:
Stefano Costalli (Università di Firenze) - Chair
Fabrizio Coticchia (Università di Genova)
Federica Genovese (University of Essex)
Francesco Moro (Università di Bologna)
Chiara Ruffa (Swedish Defense University)
Andrea Ruggeri (University of Oxford)

Chairs: Stefano Costalli

All the things we IR. Or few little things IR quals should know about IR quant and few little things IR quants should be aware about themselves.
Andrea Ruggeri (andrea.ruggeri@politics.ox.ac.uk)
AbstractA sterile and confrontational IR debate between quantitative and qualitative methods is fading away among institutions where the discipline has reached a certain maturity. The salient issue should not be the method scholars use, but the rigor in applying a clear and transparent research design and its related apt methods. Moreover, the “qual” and “quant” labels are misleading because both approaches cover very large and variegate toolkit for IR scholars. Coming from a quantitative angle, I suggest that IR scholars should think more carefully about “their” methods’ limitations and the lessons that they can learn from thinking qualitatively, and vice versa. In my paper, I explore two theses and elaborate a dilemma. The first thesis is that scholars who use qualitative methods cannot be quantitative-deaf. They should use quantitative methods to strengthen their research design. Most importantly this would be beneficial in terms of case selection, omitted variable–bias, and outliers. The second thesis asserts that quantitative scholars are not reflexive enough and tend to trade quantity over quality. The so-called “identification strategy revolution” presents itself as solution to correlational regression-studies based on marginal role of new variables. However, neither of the two is panacea. Reading and learning from colleagues using qualitative methods, quantitative scholar should think more careful about the difference of the causal effect of a variable against the causal mechanics of the effect. Moreover, non-linearity, sequencing/processes and data quality should be other important lessons to learn. The two theses combined, though, introduce a dilemma: how an IR scholar can do all this? Individual, intensive and specialist training, individual authorship (essential feature for some academic assessment), and also high article-demands seem to provide a full incentives’ structure against the implementation of the theses above. Coauthoring, long-term project and the return to the writing books could help us to mitigate this dilemma.

Text analysis e big data. Nuove vie per le relazioni internazionali?
Federica Genovese (fgenov@essex.ac.uk)
AbstractNel corso della tavola rotonda mi piacerebbe discutere opportunità e limiti di nuovi metodi di data collection e data exploration nel campo delle Relazioni Internazionali (IR), tipo social media analysis e quantitative text analysis. Credo sia utile investigare le complicazioni di questi metodi quando le risorse di ricerca sono scarse o quando le domande di ricerca sono molto grandi (come e’ spesso in IR). Inoltre mi piacerebbe sentire le opinioni di altri colleghi su quali siano i “subfields” di IR dove questi nuovi metodi siano più facilmente applicabili, e quali invece siano le domande di ricerca in IR che continueranno a richiedere più “classici” metodi di data collection e analisi tipo hand-coded data collection o secondary data analysis.

Content, discourse and frame analysis in IR: methodological challenges
Fabrizio Coticchia (fabrizio.coticchia@unige.it)
AbstractThe talk will illustrate the state of the art of the current methodological debate on content and discourse analysis in International Relations. A specific attention will be devoted at the following issues: the main challenges related to the technical process of content analysis (e.g., the role of software, the use of language for non-native speakers, etc.), the problematic distinctions among controversial concepts such as narratives, frames and master-fames, the “quali-quantitative” nature of content analysis, and the “practical” problems concerning qualitative discourse and publishing in IR journals.

Archeology as a source for IR research.
Francesco N. Moro (francesco.moro4@unibo.it)
AbstractArcheology as a source for IR research. The presentation will focus on the methodological, and to an extent substantive, contributions that more attention to archeological findings and methods can give to research International Relations. In particular, it will address issues related to evolving agency in IR and the increasing need of fine-grained data drawn from a distant past.

Between a rock and a hard place? Case study research and the quantitative shift in political science.
Chiara Ruffa (chiara.ruffa@fhs.se)
AbstractIn recent years, quantitative approaches in political science have gained traction, with their undeniable advantages in terms of generalization, transparency and comparability. Qualitative researchers still have a space, which is, however, getting eroded by both the appeal of quantitative approaches and the challenge posed by post-positivist turns. Within the positivist tradition, several scholars advocate for the necessity of a mixed-method approach, combining qualitative insights into broader quantitative studies. On the one hand, quantitative scholars often see case study research as useful but complementary, serving at best an ancillary role to shed light on causal mechanisms. On the other hand, qualitative scholars conceive of quantitative approaches often as reductionist and dismissive of sophistication and context-specific factors. This contribution explores the challenges and opportunities of mixing quantitative and qualitative methods and is keen to promote a discussion on how to combine quantitative and qualitative in creative and fruitful ways.

 

Panel 11.3 Large data across time and space: Addressing problems and solutions to the analysis of observational large-N data


While a new wave of experimental studies has recently populated most social science literatures, the large majority of social science data is still observational in nature. This panel collects papers that discuss the use of large-N methods and the statistical analysis of observational data in political science and in other disciplines. We invite studies in comparative research that use applied statistical approaches to analyze data where the heterogeneity may be on time, space units, or both. Papers on secondary panels and historical datasets are as welcome as papers on “big data”, “text-as-data”, and “quasi-experimental” and survey data. The panel will actively seek to address the challenges with doing causal inference with the data at hand as well as connecting problems and solutions to observational large-N data analysis in different subfields, from international relations to institutional studies to comparative behavior.

Chairs: Margherita Belgioioso, Federica Genovese, Fedra Negri

Discussants: Michele Fenzl, Francesco N. Moro

Exploring the relationship between ideological party similarity and negative campaign
Fedra Negri (fedra.negri@unimi.it), Luigi Curini (luigi.curini@unimi.it), Alessandro Nai (alessandro.nai@sydney.edu.au)
AbstractThe ongoing economic crisis has exacerbated party competition: several democracies are experiencing a decrease in turnout, populist parties' electoral success and mainstream parties' position-shift towards more populist stances. Given this scenario, the contribution narrows on parties' strategic behaviours during electoral campaigns by investigating the relationship between parties’ ideological positioning on the left-right spectrum and negative campaign. In detail, we maintain that a party's incentive to highlight non-policy based valence issues in the electoral confrontation increases as the spatial distance separating this party from its ideologically adjacent competitors decreases. Intuitively, this is likely to happen because, when parties are near to one another on policy issues, they have to find different means to distinguish themselves: investing in a non-policy based valence campaigning issue provides parties with that opportunity. This expectation is tested on 19 democracies worldwide.

Income Inequality and Union Membership in Advanced Democracies
Michele Fenzl (mfenzl@essex.ac.uk)
AbstractIn an age of increasing income inequality, we know very little about its effects on our political systems. This is the first study to systematically analyze the effect of income inequality on labor organizations. In particular, this study argues that inequality reduces their effectiveness, because – counterintuitively – it decreases membership in unions. To address endogeneity and reverse causality between inequality and unionization, the empirical testing adopts an instrumental variable and finds that inequality depresses union membership, based on data for 14 advanced economies from 1963 to 2007. To explain this phenomenon, this study proposes that inequality undermines the perception of unions’ effectiveness. As a consequence, workers have decreasing incentives to join and remain members of a union. Survey data from the Eurobarometer provides confirmatory evidence on the negative relationship between inequality and satisfaction with unions. The findings of this study have important implications for the future functioning of labor institutions, and more generally for our understanding of the politics of inequality. The results further contribute to an exploration of the why established democracies have failed in curbing rising inequality.

Biting the hand that feeds? External support, population dependence and rebel groups’ portfolio of killings
Margherita Belgioioso (mbelgia@essex.ac.uk)
AbstractWhat motivates some rebel group to seek cooperation with local populations rather than using them as a target of terrorist violence? What explain the variation in the portfolio of killings across rebel groups? This study implements and actor oriented approach to explain how different types of non-state actors transnational support affect rebel groups’ relative allocation between terrorist and conventional violence. Rebels receiving fungible financial support are less likely to target civilians than combatants. Rebels have incentive to invest financial support domestically rather than internationally. This is more economically efficient and it maximizes the possibility to secure less volatile resources from the population in the future. In turn, increased rebel dependency on local population generates incentives to restrain the use of terrorism. Rebels receiving hardly fungible military support are more likely to target more civilians than combatants. Military resources are efficiently invested in warfare activities without the need to increase reliance on the population and it is hard to convert military resources in assets to be invested in future, less volatile returns from the population. I model rebel groups’ portfolio of killings as a proportion of terrorist-related deaths and battle-related deaths. The empirics support all the hypotheses and are consistent with the argument that the counterproductive effects of terrorism offset its tactical advantages when rebels depend on local population.

Introducing the Peacekeeping Operations Corpus (PKOC)
Jessica Di Salvatore (jdisalvatore1@gmail.com), Elio Amicarelli (elioamicarelli@rocketmail.com)
AbstractScholars have long used United Nations official documents on peacekeeping missions to extract information for their analyses. However, traditional approach to this task has two main shortcomings. First, it is inefficient: reading, identifying and manually extracting information from texts is a time-consuming and costly process. Second, relying on non-machine readable sources precludes the community from benefiting from advancements in quantitative methodology, especially with regard to Natural Language Processing and Machine Learning techniques. In order to overcome these limits, the Peacekeeping Operation Corpus (PKOC) will provide a highly structured and multiformat machine-readable collection of the United Nations Secretary General's Reports on all peacekeeping missions from 1995 to 2016. PKOC will allow researchers to interrogate documents in a quicker and more systematic way, hence making traditional tasks such as creation of variables easier and reproducible. By reducing cost of interrogation, PKOC will also enhance the agility of researchers in exploring aspects of peacekeeping operations involving their dynamics and reporting of activities on the ground. In addition, the corpus will open the doors to the use of new methodologies in peacekeeping research as well as to automated approaches to event extraction. In this paper we discuss PKOC’s core characteristics and how they can be leveraged for quantitative analyses. We present three applications showing how PKOC (i) improves efficiency in retrieving textual information from the whole body of UN reports, (ii) allows the automated coding of new variables suitable for regression analysis and (iii) enables researcher to apply quantitative content analysis methodologies.

Big data and causal discovery: what we can learn from the EXPOsOMICS project
Virginia Ghiara (vg201@kent.ac.uk)
AbstractIn this paper, I use the biomedical project EXPOsOMICS, that studies the causal links between exposure and disease, as a case study to investigate how big data studies can gather mechanistic evidence. Two points are highlighted. First, I suggest that the EXPOsOMICS project can be considered a new form of process tracing (PT) based on the search for mechanistic evidence. Recently, Beach and Pedersen (2013) identified three variants of PT: theory-testing, theory-building, and explaining-outcome PT. The EXPOsOMICS project seems to pursue all these goals, furthermore, it appears to use a novel strategy to trace causal mechanisms: rather than selecting case studies or using experimental studies, the huge volume of data enables scientists to perform what I call a “large-scale PT”. I claim that this form of PT is not limited to the medical field and that, methodologically, it has a great advantage: researchers can be confident that their data contains the average features of the population under study. Second, the emergence of big data has enabled scientists to collect and use data whose heterogeneity is not only on time and space. Data is now collected at different levels and can be used all together to investigate the same phenomenon. The EXPOsMICS project illustrates this point as it examines biological, individual, and environmental data. This variety, however, causes some problems: it is questionable how data should be integrated, and what role different data should play in the study of causal mechanisms. Indeed, even causal mechanisms can be located at different levels: in the literature, a distinction is often proposed between constitutive mechanisms, breaking down a phenomenon into its constitutive parts, and horizontal mechanisms, tracing the causal history of a phenomenon. In the EXPOsOMICS project, scientists tend to reduce all the macro-level factors (such as the socio-economic factors) to the micro-level, establishing causal links between micro-environmental factors and micro-disease. This approach prioritizes constitutive mechanisms and appears in line with the discussion about the micro/macro relationship in the social domain: as illustrated by the so-called Coleman Boat, the main idea is that causal influence for macro factors can only work through disaggregated effects at the micro (individual) level. I suggest that the understanding of this micro/macro relationshiop is indispensable for big data to really enhance our study of causal mechanisms

 

Panel 11.4 Methodological Approaches to the Study of Public Engagement


Democratic systems are facing several challenges in the last decades. On the one hand, the increasing lack of interest in politics and distrust in political institutions and practices, such as elections (Albrecht 2006), is broadening the gap between representatives and represented, making visible the so-called ‘crisis of representation’. On the other hand, while facing complex and controversial issues, democratic policy-making processes often turn into conflicts, protests and citizens' mobilisations, which end up in stalemates or tentative solutions.
Participatory and deliberative approaches to decision making are presented as tools to answer both political dissatisfaction and disinterest and policy ineffectiveness. In particular, direct involvement of ordinary citizens in policy making is seen as a way of improving the three dimensions of democratic systems: politics, polity and policy.
This panel aims at exploring the potentials and pitfalls of participatory and deliberative methods, by looking at different designs (face-to-face, on-line, mixed), jurisdictions (local, national and international level) and issues (constitutional reforms, environmental issues, social policies, housing policies, culture, fiscal issues etc.).
The panel welcomes empirical papers from different methodological perspectives (case studies, comparative analyses, explorative and descriptive statistics), but also more theoretically-oriented contributions.

Chairs: Linda Basile, Stefania Ravazzi

Discussants: Antonio Floridia

Income inequality and voter turnout in American elections.
Michele Fenzl (mfenzl@essex.ac.uk)
AbstractResearch reports conflicting conclusions on the relationship between inequality and turnout. Some studies conclude that inequality increases turnout, and others report the opposite effect. This study moves to resolve these inconsistent findings by proposing a theory that suggests the relationship is curvilinear. It analyzes elections in American states from 1980 to 2012, and the empirical evidence shows that low levels of inequality stimulate turnout, by enhancing group identification, and mobilizing poor voters. High levels of inequality, however, reduce turnout because they harm the economic homogeneity of society, they create uninvolved social groups, and they stimulate democratic disaffection in impoverished voters. Empirical evidence shows this curvilinear relationship holds both cross-sectionally and longitudinally. Moreover, this study adopts econometric techniques that resolve estimation biases due to cross-sectional dependence among units. The results are robust to a number of model specifications, and to the measurement of turnout both on the eligible population and on the voting-age population.

Bridging the Representation Gap? Experimenting Participation with the Online Deliberation
Francesco Olmastroni (olmastroni3@unisi.it), Linda Basile (basile7@unisi.it), Pierangelo Isernia (isernia@unisi.it), Olivier Parnet (olivier.parnet@kantarpublic.com)
AbstractThese are difficult times for the European Union. The refugee and the financial crises, the security threats coming from neighbour countries, as well as the Brexit, are putting both European values and the political system under strain. How such challenges should be addressed by both national and European institutions, and whether they would eventually hinder the European project is a relevant question as ever before. People’s claims for effective responses are turning to hostility, if not anger, towards the political elites and the European project; across Europe, citizens are increasingly seduced by those parties promoting a fierce opposition to the EU. In this scenario, citizens feel that their voices are not heard by politicians in their national capitals, or in Brussels. On the other hand, policy makers often fail to understand citizens’ concerns, but also to effectively get their message across to people. The advent of social media seems to have narrowed the distance between people and political leaders, and create more room for interaction among people; nevertheless, the “unmediated” nature of the new media has also some drawbacks. The E-Voice project sought to bridge the alleged gap between citizens and elites in Europe. To achieve this goal, it matched the need for an improved interaction – both among citizens and between people and politicians – with the need for an informed environment, where discussing on the current challenges that Europe is currently facing on migration, economy and security. From October 17 to the 27, 2016, about 300 citizens from 10 European countries joined the e-VOICE online platform. For each country, people were then randomly assigned to one out of three national groups composed of about 10 members each. e-VOICE allowed participants to interact in a virtual agora, featuring different activities and roles: debating issues; asking questions to experts; interacting with politicians; playing games, designed to make people decide on principles of redistributive justice and mechanisms of cooperation in democratic societies; and formulating policy proposals. This paper proposes a description of the event, as well as an examination of the socio-demographic background of the participants, as compared to other participatory (face-to-face) events. Finally, it will present some preliminary results and findings and shed lights on the possible use of this instrument for further applications.

Designing Anticipatory Policies through the Effective Use of ICTs: A Bottom-up Approach
Giliberto Capano (Giliberto.capano@sns.it), Elena Pavan (elena.pavan@sns.it)
AbstractThe paper seeks to achieve a better understanding of how and under what conditions current digital communication technologies can become an asset to the design of effective policies. In order to do so, we bridge two strands of reflection that have hitherto developed quite independently – i.e., policy design studies and researches on policy crowdsourcing. Indeed, on the one hand, policy design studies have so far considered communication processes only tangentially and most often in relation to the “public delivery” of already formulated policies. On the other hand, studies on policy crowdsourcing have focused more on the aspects of political participation and much less on the potential and actual roles that ICTs play or might play in policy design. In this context, no systematic reflection has been yet developed on how ICTs, their features and their uses by both policy-makers and the public may affect and, potentially, enhance policy finding and design processes. In response to this situation, our paper proposes a bottom up approach to define what constitutes an “effective use of communication” in policy design. We present the reconstruction realized through the employment of digital archeology techniques of three different cases in which ICTs have been employed by governments within policy making efforts: Iceland’s recent experiment to redraft collectively its constitution; La Buona Scuola, the latest Italian public education law; and the Finnish Avoin Misteriö, a platform for crowdsourced legislation. By exploring the different modes in which ICTs have been integrated in the formulation of these three policies, we begin to disentangle different and more or less effective ways of exploiting ICTs’ networking and communicative potential for designing successful public policies.

 

Panel 11.5 Words that matter: exploring the pitfalls and opportunities of studying political texts


Quantitative Text Analysis is one of the fastest growing areas of political methodology. Analytical tasks involving thousands of political documents were hardly conceivable only a few years ago, but today represent standard applications in an increasing number of areas of political research. On the one hand, this growth in text analysis was made possible by the diffusion of software and technological innovations, such as high-quality OCR scanners, Application Programming Interfaces (APIs) to access social media contents, and aggregators of news in electronic format such as Lexis-Nexis and Factiva. The combination of all these novel tools has thus generated a massive availability of political texts. On the other hand, this growth in text availability was accompanied by statistical innovations that made it possible to identify some of political scientists’ quantities of interest previously hidden in the text. For example, bag-of-words approaches such as statistical topic models enable researchers to pinpoint and to track the salience of political issues, identifying proportions of topics in the text. Scaling models further allow to identify the latent positions of political actors leveraging on the words contained in their documents.
All of this implies terrific scientific opportunities for the years to come. In fact, applications of text analysis have been recently implemented in the most various fields of Political Science, from public opinion to public policy, and from foreign policy to elite communication studies. The challenge in front of us is therefore the identification of new potential applications: what old theories might be refined with such promising quantitative text analysis approaches? And what new theories can be tested? However, we should also become aware of the potential pitfalls of these new techniques.
This panel invites papers exploring the most promising directions of text analysis. Both normative papers exploring the applicability of text analysis and its potential pitfalls, and applied papers presenting new empirical applications, are welcomed. A non-exhaustive list of potentially interesting papers involves: standard content analysis and dictionary-based applications; applications of statistical topic modelling, sentiment analysis, and descriptive analyses of texts (word frequencies and text visualisations); and normative papers identifying which theories can and which theories should not be tested with text analysis.

Chairs: Andrea De Angelis

Discussants: Paolo Bellucci

Applicazione delle tecniche di Content Analysis ai magazine di propaganda dello Stato Islamico. La chiamata alle armi di Rumiyah.
Suania Acampa (suaniacampa@gmail.com)
AbstractIl 2016 è stato l’anno in cui il terrorismo di matrice islamica ha sferrato un attacco frontale al cuore dell'Europa. La mia analisi parte dall’interesse di comprendere le tecniche e le tematiche utilizzate - dal gruppo estremista – a supporto dell’attività di propaganda, veicolata attraverso diversi strumenti comunicativi. Il mio lavoro è stato quello di analizzare, in prospettiva diacronica, i magazine di propaganda del sedicente Stato Islamico: una prima analisi ha interessato i magazine online Dabiq e Dar al-Islam editi da al-Ḥayāt Media Center e pubblicati – in lingua inglese e francese - da giugno 2014 (anno di proclamazione dello Stato Islamico) ad agosto 2016. Successivamente, lo Stato Islamico ha sostituito la pubblicazione di Dabiq e Dar al-Islam con il nuovo magazine online Rumiyha, pubblicato da Settembre 2016. Il magazine suscita curiosità soprattutto per il significato simbolico (e non) del nome attribuitogli (trad. di Roma). Questi tre magazine sono stati analizzati - in tempi diversi - attraverso le tecniche di Content Analysis, grazie alle quali è stata possibile una lettura multidimensionale del corpus testuale che ha permesso di evidenziare nettamente i differenti obiettivi comunicativi delle testate – strettamente legati alle diverse lingue utilizzate - e le tematiche di cui si sono avvalsi i jihadisti al fine di ottenere il feedback desiderato. L’elemento più significativo, sorto dall’analisi di Dabiq e Dar al-Islam, è quello relativo alla propaganda veicolata nei confronti delle donne, basata su un modo tutto nuovo di concepire la figura femminile da parte degli estremisti islamici, mentre il magazine Rumiyah si identifica come una vera e propria “metodologia dell’attentato”. La Content Analysis si è dimostrata così un ottimo strumento di analisi e comprensione delle strategie comunicative adottate dal nuovo network jihadista, il quale cerca di porsi, non più come un gruppo terroristico ma come uno Stato vero e proprio con confini definiti, promuovendo welfare e assistenza e, soprattutto, con obiettivi comunicativi chiari che mirano al proselitismo e all’induzione all’azione.

E-Campaigning on Twitter and Facebook: The Effectiveness of Negative and Positive Campaigning during the German Federal Election 2017.
Ceron Andrea (Andrea.Ceron@unimi.it), Luigi Curini (luigi.curini@unimi.it), Wiebke Drews (wiebke.drews@eui.eu)
AbstractDue to decreasing levels of conventional political participation and growing political apathy, democratic political systems face an erosion of their core functions, including participation, representation and inclusion. In light of these developments, high hopes accompanied the communication revolution of the past decade. New technologies have substantially reduced participation costs and, simultaneously, increased the action repertoires of those willing to get engaged. These new opportunity structures have affected political participation both on the supply and demand side: social networking sites (SNS) do not only hold potentials for mass mobilization of the population, but candidates and parties themselves have signed up and created their own social media accounts for e-campaigning purposes. With most research on the role of the internet and online tools in political campaigns focusing on US presidential elections, there is a need to enrich the academic debate through systematic studies of e-campaigns in other political, legal and cultural contexts. The project aims to add to the discussion by investigating the effect of positive and negative campaign strategies on unsolicited voting intentions as expressed on both Facebook and Twitter in the anticipation of the German federal election of September 24, 2017. This paper presents the preliminary results as of the beginning of September 2017 with the goal of nowcasting electoral performance. To this end, two different approaches of text analysis are focused on: first, by means of qualitative content analyses of Facebook posts and Tweets by the seven main German parties, e-campaigning strategies will be classified as either positive or negative. Second, supervised sentiment analysis (SA) of all German Tweets and Facebook comments to party posts is used to measure the political preferences of users and, thus, the share of unsolicited voting intentions toward each of the seven main German parties. These two different types of data (parties’ posts and voters’ comments; parties’ and voters’ Tweets) allow for linking the daily variation in voting intentions to the political messages delivered by each party during the campaign as well as contrasting the usage and effects on different social media. The results will give insights into the extent to which technological advancements on the internet have indeed impacted upon political participation in the German context.

The Words that People Want: a Quantitative Text Analysis of the Five Stars Movement’s Political Discourse
Andrea De Angelis (andrea.deangelis@eui.eu)
AbstractIn this paper, I apply a recent advance in statistical modelling of text (Statistical Topic Models, STM) to study the evolution of the political discourse of a populist political movement, the Italian Movimento Cinque Stelle, from the initial mobilization stage to the current stage where it governs some Italian big cities and therefore it aims at influencing the policy-making process. I exploit the main communication channel of the M5S, namely the blog of its founder, the comedian Beppe Grillo. The hypothesis is that the recent electoral success of the movement, that gave it access the main Italian institutions, has triggered a process of institutionalization: to keep expanding its electoral base, the Movement had to change its discourse from the original anti-establishment rhetoric to include a more diversified and concrete policy platform. To test this hypothesis, I scrape the entire universe of the M5S blog posts (since 2005) to systematically analyse the topics at the centre of the political movement’s discourse. In particular, after having validated the specific most discussed topics that were identified, I track their saliency over time to explore how the political discourse has changed in the recent years. Preliminary findings show the tendency of the Movement in the last two years to focus more on the problems of immigration and to voice opposition to the European Union and related austerity policies. This, in turn, seems to suggest that the M5S institutionalization is producing a transition from a post-ideological stand of opposition to traditional left-right conflict, into a more traditional populist far-right discourse. STM modelling proves to be a very useful instrument to track dynamic turns in party discourse.