Engaging Populism? The Popularity of European Populist Political Parties on Facebook and Twitter, 2010–2020

ABSTRACT

Scholars have argued that populists disproportionately benefit from social media, and there is evidence that they attract more engagement than other politicians and parties in several countries. We systematically evaluate the relationship between populism and online popularity using a novel dataset covering over four hundred parties active on Facebook and Twitter in thirty European countries between 2010 and 2020. We use Bayesian hierarchical models to assess the relationship between populism and three types of engagement on each platform. We find that populists received more engagement than other parties on Facebook, but there was no consistent relationship on Twitter. Among these parties, only right-wing populist parties had any widespread advantages. Our results highlight considerable heterogeneity across countries and over time, providing new insight into the scope and scale of populists’ online popularity. This includes evidence that left-wing and centrist parties attracted more engagement in a handful of countries. We evaluate several explanations for these results, finding that negative macroeconomic indicators and increased migration are positively correlated with populist engagement, while internet and social media adoption have mixed impacts that benefit right-wing populists.

KEYWORDS:

• Populism
• social media
• political parties
• Facebook
• Twitter

Acknowledgment

We thank the participants of the 2021 Politics and Computational Social Science Conference, the 2023 American Sociological Association Annual Conference, George Berry, Felix Hagemeister, Ranjit Lall, Sang Lee, Byungkyu Lee, Barum Park, Emily Parker, and Ben Rosche for their valuable comments and suggestions on earlier versions of this work. We are grateful to Daniele Loprieno and Debasmita Bhattacharya for their research assistance and to the Office of Advanced Research Computing (OARC) at Rutgers, The State University of New Jersey for providing access to the Amarel cluster that contributed to the results reported here.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Data Availability Statement

The data described in this article and the code to replicate the analyses are openly available at https://github.com/t-davidson/engaging-populism-replication

Open scholarship

This article has earned the Center for Open Science badges for Open Data and Open Materials through Open Practices Disclosure. The data and materials are openly accessible at https://github.com/t-davidson/engaging-populism-replication.

Supplementary Material

Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/10584609.2024.2369118

Notes

1. We could not obtain comprehensive, longitudinal data on leaders’ social media activity, particularly for smaller parties. We consider the implications of our focus on parties in the discussion.

2. Appendix A provides further information on data collection and processing.

3. We distinguish between Northern Ireland and Great Britain rather than grouping both as the United Kingdom. See Appendix A for discussion of how we addressed overlap with Sinn Fein (SF), which is also active in the Republic of Ireland but uses the same social media accounts across both countries.

4. Two engagement types were introduced during the period: replies to Twitter in 2010 and shares to Facebook in November 2012. Due to infrequent usage in the year of introduction, we consider these engagements from 2011 and 2013 respectively. We exclude the emotional “reactions” that were introduced on Facebook in 2016 and “quote retweets” introduced in 2015 because they have no equivalent on the other platform and cover a shorter period.

5. We exclude any tweets considered retweets when calculating the engagement metrics on Twitter since engagements with retweets are not always associated with the retweeting account. For example, the tweet in our sample with the highest overall engagement was from Barack Obama and was retweeted by multiple parties. Any engagements are associated with Obama’s original tweet rather than the retweeted version, so including such tweets would significantly inflate these parties’ engagement metrics. Additionally, due to the way engagements were recorded by Twitter, most retweets do not include data on likes.

6. We also expect that parties with more followers will, ceteris paribus, get more engagement, but we cannot control for follower count because historical audience data was not available on either platform.

7. We also tested Poisson models, but they failed to converge even with many iterations, careful prior specification, and alternative parameterization. Appendix D discusses differences between Gaussian, Poisson, and Negative Binomial models, demonstrating how the latter performs best on posterior predictive checks.

8. We use 90% percentile intervals here because there is less information to estimate the parameters – there are 3,133 party-years in the model but only 30 countries, 11 years, and 316 distinct country-years – and the posterior distributions are positively skewed, which can make 95% HDI posterior intervals overly conservative. See Kruschke (2015) for further discussion of interval choices.

9. Spain (2017–20), France (2020), Italy (2020), and Sweden (2020).

Additional information

Notes on contributors

Thomas R. Davidson

Thomas R. Davidson is an Assistant Professor in the Department of Sociology at Rutgers University–New Brunswick. His research interests include far-right activism, populism, online hate speech, and computational social science.

Jenny Enos

Jenny Enos is a PhD Candidate in the Department of Sociology at Rutgers University–New Brunswick. Her research interests include anti-immigrant sentiment, ethnoracial bias, and political extremism.