SDS Seminar
The SDS Seminar series is a space for SDS Hub participants to share and listen to work by scholars from Edinburgh and further afield. These take place twice a semester. Unlike the SDS Workshop series, papers are not circulated in advance as this is a venue for more conventional academic seminars.
Seminars are hybrid and Zoom links for each event are circulated on the SDS Hub mailing list prior to the event.
Past
19 March 2025 Leveraging Social Media to Understand and Crowdsource Belief Change Interventions Zachary Horne
About the speaker: Zachary is a Lecturer in Psychology at the University of Edinburgh. He joined the School of Philosophy, Psychology & Language Sciences in 2020. Before starting at Edinburgh, he was an Assistant Professor of Psychology at Arizona State University from 2018 to 2020. He completed his PhD in psychology at the University of Illinois in 2017. Zachary's research investigates attitude change and the formation of attitudes, especially about science and morality. He is interested in how people explain the world around them and how they generate arguments. Zachary addresses these issues using a range of methods including behavioural studies with adults and children, data mining and machine learning techniques, surveys of experts, and Bayesian statistical modelling.
5 March 2025 Monarchs on the Move: Geographical Inequalities and the Making of Anglo-Britain (1870 - 1949) Marta Pagnini
My paper contributes to the sociological analysis of nationalism by focusing on the role of the monarchy in historically consolidating geographical inequalities in Britain. In particular, I examine the role that the British monarchy played at the turn of the twentieth century in consolidating, through uneven royal visits, an Anglo-centric nation and the North-South divide. The paper analyses 60,173 locations visited by British monarchs and senior members of the royal family between 1870 and 1949, as officially reported in the Court Circular, the official record of royal activities (N > 150,000) published since 1803 in major national newspapers. In terms of methods, I digitised the Court Circular data with web-scraping and OCR techniques. I then extracted spatial variables from the corpus using customised named entity recognition models and obtained information on the longitude, latitude and postcode of these locations using the Google API. The data reveals segregation patterns of visits at the national, regional and city levels, with a continuous dominance of English territories, the South of England, and London, alongside the persistent exclusion of Ireland and Wales. The role of Scotland, despite its prominence in the Victorian era, wanes in the twentieth century. Despite being increasingly easier to move from one location to the other with more modern means of transportation, visits did not disperse but instead gravitated more and more around London, symbolically reinforcing a deeply unequal geographical union.
About the speaker: Marta Pagnini is an ESRC-funded PhD candidate at the Department of Sociology of the London School of Economics and Political Science. Her research project, under the supervision of Professor Mike Savage and Dr Milena Tsvetkova, examines the historical processes of legitimation and consecration of status hierarchies by the state. To that end, she analyses with text mining, time series and social network analysis the British monarchy's public activities listed in the daily Court Circular from 1837 until present. Marta completed an MSc in Sociology at the LSE in 2020. She obtained a BA in Sociology from the University of Paris IV Sorbonne and a BA in Political Science at the University of Pisa. Before joining the LSE as a PhD fellow, she worked as Market Research Analyst at Euromonitor International and as Research Assistant in Economic History for the University of Groningen.
4 March 2024 Platform-Controlled Social Media APIs Threaten Open Science Emily Godwin and Darja Wischerath
Social media data hold tremendous value for studying behavioural patterns over time and across contexts at individual, group and population levels, and are relevant to a broad range of disciplines in the social and behavioural sciences. Researchers can mainly access these data via platform-provided application programming interfaces (APIs), but these come with restrictive usage terms that mean studies cannot be reproduced or replicated. In this seminar we discuss how platform-owned APIs hinder access, transparency and scientific knowledge, other ways of accessing data, as well as legislative endeavours to promote researcher access to data.
About the speaker: Darja Wischerath is a doctoral researcher at the University of Bath. Her research explores how belief in conspiracy theories can lead to violence and violent extremism through the mediating factors of psychological needs, social networks, and violence legitimating narratives. Emily Godwin (MSc, University of Bristol) is a Doctoral Researcher in Cyber Security at the University of Bath. She is interested in the spread of mis/disinformation online, with a particular focus on conspiracy theories, as well as the ways in which online spaces are moderated.
5 February 2024 LocalView: Scaling Up the Study of Local Politics and Policy-Making in the United States Soubhik Barari
Despite the fundamental importance of American local governments for services like education and public health, local policy-making remains difficult to study at scale due to a lack of centralized data. In this talk, I introduce LocalView, the largest existing dataset of real-time local government public meetings, the central policy-making process in American local government. In sum, the dataset covers more than 100,000 videos and their corresponding textual and audio transcripts of local government meetings publicly uploaded to YouTube from more than 1,000 places and 2,000 distinct governments across the United States between 2006-2023. I discuss how the data are processed, downloaded, cleaned, and publicly disseminated (at localview.net) for analysis across places and over time. Finally, I show how LocalView can be useful for social science and journalistic applications such as measuring political polarization in local politics and tracking shifts in public health attention across geography.
About the speaker: Soubhik Barari is a quantitative political scientist and data scientist. His academic research explores the dynamics and effects of online misinformation, public deliberation in local politics, media coverage in U.S. elections, and political cues from corporations. Currently, he is a Quantitative Social Scientist at NORC at the University of Chicago and an Adjunct Assistant Professor of Political Science at Columbia University. He holds a Ph.D. in Political Science and an A.M. in Statistics from Harvard. Previously, he has worked in research and data science roles at SurveyMonkey and Microsoft Research.
22 January 2024 The Microtargeting Manipulation Machine Almog Simchon
In recent years, there has been a growing concern over "psychological microtargeting", in which psychological features that cannot be directly observed, such as personality characteristics, are inferred from online behaviour and personal data, and are used to customize manipulative messages, for example to provoke political action/inaction or to spread misinformation. Such microtargeting is opaque to online users and its effects, to the extent that they are understood, should give rise to concern. Hence, there is an urgent need to "reverse engineer" microtargeting strategies by uncovering the targeting algorithms in action. This talk presents a proof of concept for such algorithmic reverse engineering.
About the speaker: Almog Simchon is an Assistant Professor in the Department of Psychology at Ben-Gurion University of the Negev. Formerly a Senior Research Associate at the University of Bristol and a member of its Technology Democracy and Cognition (TeDCog) group led by Stephan Lewandowsky. Almog completed his PhD at Ben-Gurion University of the Negev in 2021. As a computational social psychologist, Almog's research examines how social media, social psychology, and language interact.
20 November 2023 Ideological Polarization on Social Media Marilena Hohmann
Despite a multitude of studies of polarization on social media, it remains disputed whether polarization in digital public spaces is on the rise. To analyze ideological polarization in social media networks, a suitable measure is needed. We argue that this measure should account for three defining factors: how extreme people's opinions are, how much they organize into echo chambers, and how these echo chambers organize in the network. We propose a new measure of ideological polarization in social networks that can capture all these factors, demonstrate its sensitivity in synthetic experiments, and apply it to real-world data, examining political debates on Twitter and voting patterns in the US House of Representatives. With Karel Devriendt and Michele Coscia.
About the speaker: Marilena Hohmann is a second-year PhD student in Social Data Science at the University of Copenhagen. Her interdisciplinary PhD project explores how people are connected through their opinions, beliefs, and ideas, for instance, in politically polarized discussions.
6 November 2023 A Human-centric Approach to Researching Social Bias in NLP Technologies Eddie Ungless
Much current research into social bias in Natural Language Processing -- that is, the tendency for NLP technologies to reflect human biases such as sexism -- suffers from a superficial understanding of the problem. The issue of bias is treated as a mathematical kink that needs to be ironed out, after which the harm the model does will be irrefutably reduced. I argue this superficial understanding of bias will lead us to superficial solutions which ignore the role of human behaviour in determining the harm done by technology. In this talk, I advocate for a human-centric approach to studying bias in NLP, presenting two papers as case studies, each addressing queerphobia in publicly available NLP tools.
About the speaker: Eddie L. Ungless is a final year PhD student in the Centre for Doctoral Training (CDT) in NLP, funded by the UKRI. He has an interdisciplinary background spanning linguistics, psychology, digital media strategy and computer science. His work addresses social bias in NLP technologies.
23 October 2023 Computing Hearts and Minds: The Implications of Using Modern Quantitative Social Science for Public Communication Campaigns Ben Tappin
Public communication campaigns are increasingly using the methods of quantitative social science to achieve their goals. In this talk I consider some implications of this trend for the potential impact of campaigning on public attitudes and behaviour, presenting research analysing the implications of online survey experimentation and message-targeting using machine learning for the impact of political and public health campaigns, and whether results can be generalised to public communication "in the wild".
About the speaker: Ben is a quantitative social scientist interested in political behaviour, public opinion and attitude and behaviour change. He is completing an early career research fellowship, funded by the Leverhulme Trust, investigating the impact of personalised/targeted communication on public opinion and behaviour. He works out of the Centre for the Politics of Feelings and Royal Holloway University in London, and is a research affiliate of the Human Cooperation Lab at MIT.
9 October 2023 Can Neighbourhood Indicators Derived from Natural Language Processing of Local Newspaper Text Help Us Understand Population Health Resilience and Vulnerability? Eleojo Abubakar
It is of great interest to understand why some neighbourhoods experience markedly better, or worse, health outcomes than might be expected given the local deprivation. In this study, we use natural language processing (NLP) techniques comprising geo-referenced, topic modelling of local newspaper text data, to derive novel neighbourhood indicators that quantify the intensity of specific "themes" in the local press for the city of Edinburgh. We explore whether spatial variation in these themes is associated with spatial variation in population resilience to poor health. We operationalise population health resilience using the residuals from a regression model that takes health outcome as the dependent variable and the Scottish Index of Multiple Deprivation (SIMD) as the independent variable. Of the 55 themes that emerged from the NLP analysis, sixteen were significant determinants of neighbourhood health resilience across Edinburgh, with the strongest results observed for mental health. The themes that emerge capture aspects of community, civic/social participation and the nature of local services, that are not well represented in the SIMD (or other official neighbourhood data) and are theoretically plausible in terms of mechanisms underpinning their association with neighbourhood health resilience.
About the speaker: Eleojo is a health geographer concerned with the role of location, place, and context in shaping health outcomes. He is adept at applying a wide variety of data science and spatial analysis methods. He was presently a Post-Doc Research Fellow in Social Data Science at the University of Edinburgh. Prior to this, he was a Post-Doc Research Associate in Urban Studies at the University of Glasgow and a GIS specialist at Cathie Group, UK.