Books
Journal and proceedings publications
Blogs
Software
Migration and direct democracy: the case of referendums in Switzerland
Matt Ryan, Paolo Spada, Masood Gheasi, Edson Utazi and Daniele Mantegazzi
Presented at the European Consortium for Political Research (ECPR) conference, 2023 in the Panel for Challenges to the Inclusion of Citizens in Digital and Direct Democracy
In the last decades, migration has been a hot topic at the national, cantonal, and municipal levels in Switzerland (among other countries) and several referendums obtained sufficient consensus in imposing restrictions on migration or banning certain cultural representations (such as ban on minarets of mosques). This study investigates the spatial-temporal dimension of voting patterns in the Swiss system on different types of referendums related to migration, and observes their relationship with linguistic, socioeconomic, and local characteristics in a multilevel temporal spatial model. Particular attention is dedicated to voting pattern’s variations related to the three Swiss instruments of direct democracy: mandatory referendums, optional referendums, and popular initiatives. The results highlight the existence of significant differences among Swiss municipalities in their voting patterns on referendums related to migration and indicate that these differences are associated with inequalities in local economic welfare, education, age, language, and political ideologies. Moreover, differences in voting behaviour are observed in relationship with the three different instruments of direct democracy, and these differences vary depending on the local socio-demographic and socio-economic characteristics. Overall, this study suggests that the availability and exploitation of different instruments of direct democracy allows giving more voice to more and different people.
Read the paper here.
Perceived constraints and missed opportunities? Exploring the tensions between adopting open science and innovating democracy
Lala Muradova, Matt Ryan, Rafael Mestre, and Masood Gheasi
Conference paper at the European Consortium for Political Research (ECPR) conference, 2021
Panel Advancing Open Science Practices Within the Democratic Innovations Field
In this article, we studied the application of four commonly-discussed open science practices (OSP) (pre-registration, data sharing, replication and open access publishing) within the subfield of democratic innovations, by analysing published and peer-reviewed papers in the field (N = 6384). A minority of articles have used one or more OSPs. Replication and pre-registration accounted for less than 1% of research articles. Full data availability of research materials (data, code for replication, etc.) was only found in approximately 3.5% of our sample; though this practice has increased over the last 10 years reaching close to 8% in the year 2020. At the same time, we find an increase in the application of open publishing over time, reaching almost 50% of the publications in recent years. We conclude by arguing that OSPs can enhance the validity and rigour of research and can consequently contribute to improving the practice of democratic innovations and their policy effects. We also recognize potential perceived constraints of OSP on creative discovery within the field and discuss whether open research conflicts with aspects of exploratory discovery, by shedding light on some misperceptions.
Read the paper here.
Gender Inequality and Deliberation: a meta-analysis
Masood Gheasi, Matt Ryan, Jessica Smith, Rafael Mestre
Working paper
Deliberation is an essential component of a healthy democracy. Through deliberation citizens listen to, learn from, and engage with different discourses. Furthermore, diversity (especially inclusion of different genders, races, ethnicities, and other groups who have been historically disadvantaged) is important for improving fairness through reducing prejudice and discrimination in a deliberative body. Among a larger volume of research on diversity and in particular gender differences in deliberation, researchers find significant gender gaps in participation and inclusion. There are a number of qualitative studies on gender inequality in deliberation, but this topic is relatively understudied using advanced large-N statistical methods. Recent experimental work has begun to test gender gaps in deliberative settings and possible interventions which may lead to more equitable deliberations. In the first meta-analysis in this area, we pool the results of these experimental studies to better understand the relationship between deliberation and gender, and how it may be moderated by the rules of deliberation. The analysis of 10 studies of a rich quality yielded 579-point estimates. In this paper we focus not only on the magnitude of the effect sizes, but also on their direction and statistical significance. Aggregated results reveal a more nuanced picture regarding gender gaps in deliberation. Perhaps surprisingly, the aggregated effect size of gender across experiments is almost zero, showing only very small inequality in gender deliberation at a superficial level. However, our results provide strong confirmation that the decision rules can shape deliberation in favor of women in a small group discussion.
Read the paper here.
Crisis and Attitudes to Democracy: A Comparative Study
Masood Gheasi, Matt Ryan, Annie Tubadji, Rafael Mestre
Working paper
Research on the effect of economic crises on support and desire for democracy has rarely considered important differences in regime types. In this study, we asked what the consequences of the 2008 economic crisis were on support and desire for democracy in different regimes. Significantly we include geographies beyond the USA and Europe. The paper explores democratic support in different regimes namely, democracies, hybrid-democracies, hybrid-autocracies and autocracies. Our analysis is based on two waves of the World Values Survey (wave 5 and 6). It shows that the major economic crisis of 2008 decreased peoples’ support for democracy in autocratic and hybrid autocratic countries, while support increased in hybrid democratic countries. We further explore the importance of the political context for these shifts in attitude to democracy under the same economic shock. The empirical analysis suggests that in autocratic regimes people’s pro-democratic attitudes (valuing freedom of speech, and general support for democracy) have declined after the economic crisis. The same results also hold for hybrid autocratic regimes, while in hybrid democratic regimes people’s support for democracy has increased after the economic shock of 2008. Our results show that the effect of major economic shock on support for democracy is conditioned by regime type.
Paper available upon request.
Argument Mapping in Public Consultation
Paolo Spada, Matt Ryan, Adam Meylan-Stevenson, Masood Gheasi, Rafael Mestre, Richard Gomer, Tim Norman (University of Southampton)
Working paper
Democratic innovations have blossomed in recent years as part of attempts to avert a democratic malaise. Attention of scholars has focussed on devices that aim to augment public consultation by further democratising their processes of recruitment, deliberation, and decision-making. Many of these deliberative processes aim to elicit considered judgements through collective reasoning of a diverse group of citizens. A key challenge is ensuring inclusion and equality of voice for participants while integrating expertise. Can existing and emerging technologies help rather than hinder this process? In computer science and related fields, work on argument mining, mapping and reasoning has developed tools for identifying, mapping and interpreting argument structures and disagreements as they emerge. There are at least two potential uses of these tools in democratic innovation: First by classifying and mapping arguments we can develop visual presentations of the structure of a discussion and the key components of arguments. Such a tool can then be used to reduce redundancy, neatly structure discussion around support for claims, and expose unstated assumptions. Secondly, techniques can be used to identify signals or predictors of anti-democratic modes of speech allowing early intervention in discussions. The potential for computational linguistics to enhance political deliberation among citizens remains mostly untapped. In this paper, we assess the extent to which using tools of argumentation in public consultation affects democratic goods by leveraging a set of semi-structured interviews with stakeholders that have been working on hybrid and digital democratic innovations in the past ten years.
Paper available upon request.
Deliberation, Argumentation, and Democracy
Rafael Mestre, Matt Ryan, Masood Gheasi, et al.
Paper prepared for APSA Annual Meeting, Montreal, 2022
Advanced democracies face a plethora of wicked problems of governance linked to increased polarisation of politics, the spread of misinformation, and decreased trust in democratic institutions. One common point of contact in all these issues is deliberation and/or argumentation, both online and offline. Computer scientists and philosophers of linguistics have tried to understand how arguments can be identified, their quality assessed, mapped, and improved, with techniques ranging from artificial intelligence and machine learning to qualitative methods. Democratic Innovations, inspired by theories of deliberative and participatory democracy have focused on institutional engineering to increase inclusion and capacity of citizen voices. This paper maps the different foci of responses to these problems and identifies missed opportunities for clever collective response. We show that conceptual confusion as well as differences in focus on argumentation, dialogue or discourse has led to underuse of deliberative insights in argument mapping, in turn reducing the impact of advances in argument identification in democratic innovation. The paper discusses implications for some of the leading tools and platforms currently in widespread use. We find that approaches in the social sciences have advanced strong normative criteria, as well as detailed policy implication, but lack a mid-range theory to explain how affordances of design affect communication in fora. Elsewhere certain approaches linking computer science and philosophy have offered strong conceptual theory to guide efficient product design but suffer from a lack of attention to normative questions of how social outcomes are achieved. Drawing on these insights we produce a novel organising perspective to guide efficient discovery of solutions within this pressing research agenda.
Read the paper here.
M-Arg: MultiModal Argument Mining Dataset for Political Debates with Audio and Transcripts
Rafael Mestre, Razvan Milicin, Stuart E. Middleton, Matt Ryan, Jiatong Zhu, and Timothy J. Norman
Proceedings of the 8th Workshop on Argument Mining, 2021 https://aclanthology.org/2021.argmining-1.8/
Argumentation mining aims at extracting, analysing and modelling people’s arguments, but large, high-quality annotated datasets are limited, and no multimodal datasets exist for this task. In this paper, we present M-Arg, a multimodal argument mining dataset with a corpus of US 2020 presidential debates, annotated through crowd-sourced annotations. This dataset allows models to be trained to extract arguments from natural dialogue such as debates using information like the intonation and rhythm of the speaker. Our dataset contains 7 hours of annotated US presidential debates, 6527 utterances and 4104 relation labels, and we report results from different baseline models with highest accuracy of 0.86 with a multimodal model.
Potential and Pitfalls of Audio-as-Data: alignment, features and classification models
Rafael Mestre and Matt Ryan
Working Paper
Political science is a field rich in a great variety of multimodal sources of information, from televised debates to parliamentary briefings or TV interviews. The paper bridges a gap between computer science and political science in the field of multimodal data analysis using audio. The adoption of multimodal analyses in political science (for example including video or audio with text-as-data approaches) has been relatively slow due to the unequal distribution of computational power and skills needed to analyze this type of political data. This paper focuses on solving challenges encountered when trying to analyse audio sources and advancing potential of multimodal data analysis in political science. Using a dataset of all televised US presidential debates from 1960-2020, we focus on three types of features encountered when analyzing audio data: Mel-frequency cepstral coefficients (MFCCs), low level descriptors (LLDs) like pitch or energy, and audio embeddings or encodings like Wav2Vec. We showcase four applications using this dataset: a) forced alignment of audio and text using MFCCs, timestamping transcripts and speaker information to the audio source; b) speech characterization using LLDs; c) custom-made classification models—speaker recognition—with audio embeddings and MFCCs; and d) emotional recognition models using Wav2Vec for classification of discrete emotions and their location in the valence-arousal-dominance spectrum. We provide explanations to help understand how these features can be applied for different political research questions, for both experienced researchers and those who want to start working with audio.