The Shock of the Old? The Value of Looking Back When Studying the Mercurial World of Political Campaigning[1]
Kate Dommett, University of Sheffield
Sam Power, University of Sussex
http://dx.doi.org/10.17169/refubium-43527; PDF
‘Disinformation on steroids’ (Leingang, 2024), ‘An evolution in propaganda’ (Robins-Early, 2023) and ‘Deepfakes pose ‘perfect’ threat to next election’ (Dathan, 2024). These are a few of many examples of the often sensationalist coverage of the effect that AI will have on elections in 2024. With more voters headed to the ballot box in countries around the world than has happened in recent history (Ewe, 2023), scholars are increasingly in demand to offer our expertise on how and why this technology is being used, and what impact it will have on elections and wider society.
There is, unfortunately, no easy answer to these questions. Whilst journalists are often keen to reach out and learn from our expertise, our capacity to generate rapid, evidence-based answers is severely curtailed. Data is often not available, access for interviews and observation can be challenging, and researchers themselves are increasingly subject to threats of litigation for conducting this kind of research. By the time we have been able to marshal scientific evidence, public ideas around the impact of new technologies have often been shaped, and public and policy-making attention has frequently moved on to consider new technologies (Orben, 2020).
Recognizing this incessant demand for knowledge and the challenges of generating evidence-based responses, in this piece we consider how scholars can respond. Reflecting on our own recent experience of studying data-driven campaigning (Dommett et. al., 2024a), we assert the importance of establishing clear conceptual boundaries about the precise meaning of any new technological phenomena and suggest that efforts to draw lessons from the past can play an important role in shaping our understanding of the new. Advocating a methodologically pluralist approach, we seek to steer debate away from a recurring focus on transformation or threat, to consider the more nuanced and even mundane impacts that technologies can also exert.
The Study of Digital Elections
Innovation in election campaigning is a longstanding topic of public and academic interest. Looking back over recent history, the focus has been on new developments and their significance for campaign practice and success. From the impact of radio and television to more recent attention towards websites, social media, and online advertising, we recurrently think about how these new tools may affect elections. We’ve seen particular interest in the impact of online microtargeting, data-driven campaigning and, most recently, artificial intelligence. Each of these technologies has been seen to transform and threaten the practice of elections, inspiring extensive coverage of each development.
Take, for example, data-driven campaigning. In the wake of the Cambridge Analytica scandal in 2018 there was wide-ranging interest in the significance of data for electoral politics, and the potential threat posed by the ‘Digital influence machine’ (Nadler, Crain and Donnovan, 2018). Numerous stories emerged profiling the accounts of whistle-blowers (The Guardian, no date) and there was a desire to know more about the use of data to advance a campaign’s electoral fortunes. Looking back at this question now, we have a range of evidence about the use of data in election campaigns.
Scholars have studied the practice of data collection and analytics to look at the precise way in which data is being mobilized in campaigns (Dommett et al. 2024b; Kefford, 2021; Walker and Nowlin, 2022) and a body of work has interrogated the impact that seeing targeted political content online has on voter’s attitudes towards campaigners and their voting intentions (Aggarwal et al., 2023; Coppock et al., 2020; Haenschen, 2023; Decker, 2023; Lavigne, 2020; Tappin et al., 2024; Zarouali et al., 2022). At the time however, whilst there was some relevant research (e.g. Anstead, 2017; Kruikemeier et al., 2016), there was little on which journalists could draw to assess the prevalence, impact, and significance of these tools. It was therefore common to see stories such as Time Magazine’s article ‘Facebook’s New Controversy Shows How Easily Online Political Ads Can Manipulate You’ (Ghosh and Scott, 2018) which claimed that:
‘The real story is about how personal data from social media is being used by companies to manipulate voters and distort democratic discourse. In this regard, it appears the Trump campaign had a decisive and ill-gotten advantage in the quest to exploit personal data to influence voters. And they used it to the hilt.’
In many instances, academic evidence was absent from these stories. Early coverage of data-driven campaigning was therefore often characterized by a lack of empirical evidence. This meant there was little clarity about exactly what was meant by the term data-driven campaigning and which aspects of this activity were the cause of concern – something only addressed several years later (Dommett et al. 2024c).
Observing such trends, academics can often be frustrated by the claims made in journalistic coverage and policy debates. The prevalence of speculative and sensationalized depictions of the power new technologies have to fundamentally transform the electoral landscape can be out of kilter with our own expectations. Our capacity to generate timely evidence to test these claims is, however, limited. Whilst serendipity can sometimes lead our research agendas to align neatly with the latest technological development, more often new technologies gain coverage before scholars have had the chance to gather empirical evidence. This means we’re often reliant on gathering it after the fact, and often in ways shaped by the prevailing emphasis on transformation and threat. Indeed, we are incentivized by funding competitions to design projects assessing the threats and democratic challenges that have gained popular attention. Whilst such interventions are important, academics can reinforce prevailing narratives, which can often be overstated. Or, as political scientist Phillip Cowley put it, “[t]here is an inverse relationship between the importance of any election campaign technique and the amount of attention devoted to it” (Cowley, 2010).
Looking to the past
Observing these tendencies, we identify two potential ways in which academics can shape early public discourse. First, we argue there is a vital role for scholars to play in establishing clear conceptual boundaries about the precise meaning of any new technological phenomena. If we can better conceptualize and situate certain developments, we can better understand exactly what is new and where commonalities with old practices lie. In the realm of data-driven campaigning, for example, conceptualizing this activity as “A mode of campaigning that seeks to use data to develop and deliver campaign interventions with the goal of producing behavioural or attitudinal change in democratic citizens” (Dommett et al., 2024b), helped to reveal the longstanding antecedents of data collection by campaigns. By showing this history, and that techniques were often facilitated by the free provision of state data, we raised more acute questions about the precise forms of data collection and analytics that were occurring and why these more novel practices were substantively different to, for example, the segmentation and targeting of voters using direct mail. Clear conceptualization of new technology and its relationship to pre-existing phenomena can therefore help to contextualize the novelty of any new approach and often leads to a more nuanced and frankly mundane assessment of new technologies.
Second, we contend that scholars can highlight lessons from past technologies in efforts to shape understanding and generate expectations about the latest innovations. This approach helps to build upon existing theories and empirical research and foregrounds material considerations that affect how new technologies are likely to be adopted and employed, helping to counter technological myths (Aagaard & Marthedal, 2023). Previous research on data-driven campaigning, for example, shows that political campaigns are often reticent about employing new technologies, are cautious of risks, and are also materially constrained in their capacity to exploit new tools (Dommett et al., 2024b). Elsewhere, work on ‘fake news’ shows that exposure is more likely to reinforce partisan priors than have a radical effect on voting behavior (Sindermann et al., 2020). Thinking about these lessons for current debates around AI, these findings suggest that AI tools are unlikely to be rapidly adopted or used by these organizations to their full capacities for some time to come. Likewise, they indicate that AI-generated ‘deepfakes’ are likely to reinforce pre-existing viewpoints as opposed to fundamentally reforming people’s political viewpoints. By drawing on existing theories and findings generated in other, cognate areas, scholars can help to shape expectations about effects prior to new empirical research emerging.
In recommending efforts to establish contextual boundaries and to look to the past, we also assert the value of studying campaign activity and technologies in a longitudinal perspective, seeing new developments within the context of existing practices. We acknowledge, however, that the task of studying previous campaign activity is far from simple. Over recent years, analyzing electoral practice has become increasingly challenging and scholars now find many of the methodological tools we have historically deployed to be insufficient. For those working in the qualitative tradition, access for interviews and ethnographic observation has often become harder as campaigners become more cautious about sharing their campaign approach. Additionally, campaigners can be wary about sharing documentation about party practices or are subject to non-disclosure agreements. There are also risks for researchers studying political campaigns, especially when looking at those employing controversial tactics or advancing radical ideological agendas (Brown and Searles, 2023).
For those conducting observation of campaign output, other challenges have emerged. The capacity to systematically gather data has been frustrated by successive changes in data availability, with well-documented restrictions to APIs (de Vreese and Tromble, 2023), withdrawal of other monitoring infrastructure (such as Crowdtangle) and a tendency for digital platforms to litigate against academics reportedly breaking terms of service to gather data via scraping or addons (Adler and Maréchal, 2023). Even where data is available, this infrastructure is also limited and unreliable, placing constraints on our ability to research what is happening online (Edelsen et al., 2018; Leerssen et al., 2018). Cumulatively, these developments have made the process of generating data on campaign practices and impacts harder, something that, in turn, makes it challenging to predict future developments based on past insights.
Tackling these challenges is by no means simple. Within our own research, we have countered them by taking an adaptive and methodologically pluralist approach. As primarily qualitative scholars, we were trained in the art of interviews, documentary analysis, and coding frameworks, with specific expertise in interviewing elite and grassroots campaigners, and scrutinizing publicly available documentation. Over the past few years, however, we have experienced the challenges described above, and so have broadened our methodological toolkit, using interdisciplinary collaboration with computer scientists to access available datasets and monitor the practice and shape of campaigning. We have also reconsidered existing data sources and have repurposed them to gain new insights. Taking this approach, we have been able to generate different forms of insight, helping us develop a longitudinal and contextualized conception of election campaigning in the UK (Dommett et al., 2024a). This approach can help to overcome the challenges encountered with any one method and can help to build an understanding of technologies deployed at past elections that can inform our expectations of newly available tools.
Conclusion
Studying digital elections is hard and countering the prevailing tendency to emphasize the transformational or threatening potential of new technologies is by no means easy. Yet we argue that it is essential to resist this tendency and to make more nuanced, if mundane, interventions in attempts to shape public discourse. We suggest that by conceptualizing and situating new technologies, and by drawing lessons from previous research and theory, scholars can produce important work. In advancing this approach, we note the many methodological challenges faced in attempts to generate such knowledge and welcome an increasingly (and genuine) pluralistic approach to inquiry. As technology continues to evolve and transform, looking to the past can help as much as imagining the future. Especially when studying the mercurial relationship between elections and technology.
References
Aagaard, P., & Marthedal, S. (2023). Political microtargeting: Towards a pragmatic approach. Internet Policy Review, 12(1), 1-22.
Aggarwal, M., Allen, J., Coppock, A. et al. A 2 million-person, campaign-wide field experiment shows how digital advertising affects voter turnout. Nat Hum Behav 7, 332–341
Adler, W. and Maréchal, N. (2023). To Protect Elections, Protect Researchers, Centre for Democracy and Technology. https://cdt.org/insights/to-protect-elections-protect-researchers/.
Anstead, N. (2017). Data-driven campaigning in the 2015 United Kingdom general election. The International Journal of Press/Politics, 22(3), 294-313.
Brown, D. K., & Searles, K. (2023). “New” Methods,“New” Challenges “New” Methods,“New” Challenges. Political Communication Report, 27, http://dx.doi.org/10.17169/refubium-39043.
Coppock, A., Hill, S. J., & Vavreck, L. (2020). The small effects of political advertising are small regardless of context, message, sender, or receiver: Evidence from 59 real-time randomized experiments. Science advances, 6(36), eabc4046.
Cowley, P. (2020). ‘Out with the new, in with the old’. Electionblog2010. https://electionblog2010.blogspot.com/2010/04/out-with-new-in-with-old.html.
Dathan, M. (2024). ‘James Cleverly: Deepfakes pose ‘perfect’ threat to next election. The Times. https://www.thetimes.co.uk/article/james-cleverly-deepfakes-threat-next-general-election-bwmjcdfpm.
Decker, H., & Krämer, N. (2023). Is personality key? Persuasive effects of prior attitudes and personality in political microtargeting. Media and Communication, 11(3), 250-261.
de Vreese, C., & Tromble, R. (2023). The data abyss: How lack of data access leaves research and society in the dark. Political Communication, 40(3), 356-360.
Dommett K, Power S, Barclay A, Macintyre A. (2024a) ‘Understanding the Modern Election Campaign: Analysing Campaign Eras through Financial Transparency Disclosures at the 2019 UK General Election’, Government and Opposition. 1-27, DOI: https://doi.org/10.1017/gov.2024.3
Dommett, K., Kefford, G., & Kruschinski, S. (2024b). Data-driven Campaigning and Political Parties: Five Advanced Democracies Compared. Oxford University Press.
Dommett, K., Barclay, A., & Gibson, R. (2024c). Just what is data-driven campaigning? A systematic review. Information, Communication & Society, 27(1), 1-22.
Edelson, L., Sakhuja, S., & McCoy, D. (2018). An analysis of Facebook’s archive of ads with political content. In Unpublished manuscript. https://c3media. vsos. ethz. ch/congress/2018/slides‐pdf/35c3‐9419explaining_ online_us_political_advertising. pdf.
Ewe, K. (2023). The Ultimate Election Year: All the Elections Around the World in 2024. Time Magazine. https://time.com/6550920/world-elections-2024/.
Ghosh, D. and Scott, B. (2018) ‘Facebook’s New Controversy Shows How Easily Online Political Ads Can Manipulate You’, Time Magazine, https://time.com/5197255/facebook-cambridge-analytica-donald-trump-ads-data/.
Haenschen, K. (2023) The Conditional Effects of Microtargeted Facebook Advertisements on Voter Turnout. Polit Behav 45, 1661–1681
Kefford, G (2021) Political Parties and Campaigning in Australia: Data, Digital and Field, Basingstoke: Palgrave MacMillan.
Kruikemeier, S., Sezgin, M., & Boerman, S. C. (2016). Political microtargeting: relationship between personalized advertising on Facebook and voters’ responses. Cyberpsychology, Behavior, and Social Networking, 19(6), 367-372.
Lavigne, M. (2020) ‘Strengthening Ties: The Influence of Microtargeting on partisan attitudes and the vote’, Party Politics, 1-12, Online first.
Leerssen, P., Ausloos, J., Zarouali, B., Helberger, N., & de Vreese, C. H. (2018). Platform ad archives: Promises and pitfalls. Internet Policy Review, 8(4).
Leingang, R. (2024). ‘Disinformation on steroids’: is the US prepared for AI’s influence on the election?’. The Guardian. https://www.theguardian.com/us-news/2024/feb/26/ai-deepfakes-disinformation-election.
Nadler, A., Crain, M., & Donovan, J. (2018) ‘Weaponizing the digital influence machine’, Data and Society, https://datasociety.net/wp-content/uploads/2018/10/DS_Digital_Influence_Machine.pdf.
Orben, A. (2020). The Sisyphean cycle of technology panics. Perspectives on Psychological Science, 15(5), 1143-1157.
Papakyriakopoulos, O., Hegelich, S., Shahrezaye, M., & Serrano, J. C. M. (2018). Social media and microtargeting: Political data processing and the consequences for Germany. Big Data & Society, 5(2). https://doi.org/10.1177/2053951718811844.
Robins-Early, N. (2023). An evolution in propaganda: A digital expert on AI influence in elections. The Guardian. https://www.theguardian.com/us-news/2023/jul/20/artificial-intelligence-us-elections.
Sindermann, C., Cooper, A. and Montag, C. (2020). A short review on susceotibility to falling for fake political news. Current Opinion in Psychology, 36(1): 44-48.
Tappin, B. M., Wittenberg, C., Hewitt, L. B., Berinsky, A. J., & Rand, D. G. (2023). Quantifying the potential persuasive returns to political microtargeting. Proceedings of the National Academy of Sciences, 120(25), e2216261120.
The Guardian. (no date) ‘The Cambridge Analytica Files;, The Duaridan, https://www.theguardian.com/news/series/cambridge-analytica-files.
Walker, D., & Nowlin, E. L. (2021). Data-driven precision and selectiveness in political campaign fundraising. Journal of Political Marketing, 20(2), 73-92.
Zarouali, B., Dobber, T., De Pauw, G., & de Vreese, C. (2022). Using a personality-profiling algorithm to investigate political microtargeting: assessing the persuasion effects of personality-tailored ads on social media. Communication Research, 49(8), 1066-1091.
Katharine Dommett is Professor of Digital Politics at the University of Sheffield. Her research focuses on digital campaigning, data and democracy. She has published extensively on these topics, including her latest co-authored book Data-Driven Campaigning in Political Parties: Five Advanced Democracies Compared, published by Oxford University Press. Professor Dommett works closely with electoral regulators, and was recently appointed to the Irish Electoral Commission’s Research Advisory Board. More details can be found at http://www.katedommett.com
Sam Power is Senior Lecturer in Politics at the University of Sussex. His research focuses on political financing, electoral regulation, and political campaigns. He has published widely on these themes and regularly provides expert commentary to the media and policymakers.
[1] Copyright © 2024 (Kate Dommett & Sam Power). Licensed under the Creative Commons Attribution Non-commercial No Derivatives (by-nc-nd). Available at https://politicalcommunication.org.