Understanding the online relationship between politicians and citizens. A study on the user engagement of politicians’ Facebook posts in election and routine periods

ABSTRACT

Social media have offered politicians a way to reach a broader audience and citizens a dynamic way to respond and interact with politicians’ communication. In this study we focus on how two important dimensions of the social media messages of politicians impact different types of user engagement: the distinction between political and private posts and the degree of emotionality of the post. Additionally, we compare the amount and types of interaction between routine periods and election periods. Supported by automatic data gathering and coding we analyze all Facebook posts of 124 Belgian politicians for a period of more than two years (N = 34,408). Our results indicate that different types of Facebook posts lead to different types of user engagement. Private posts generate more reactions, while political posts are more often shared and commented on. Additionally, Facebook posts with positive, and, negative emotional language garner more interaction than those with less emotionality. Finally, during election campaigns both politicians and citizens are more active. There is a proliferation of the amount of Facebook messages that politicians post, and these messages also score higher on engagement.

KEYWORDS:

• Social media
• Facebook metrics
• user engagement
• political communication
• automatic coding
• emotions
• election campaign
• machine learning

Disclosure statement

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

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Notes

1. Substitutes who (temporarily) filled in the parliamentary seat of a ministers were also included.

2. For more info see https://www.crowdtangle.com

3. A BERT-transformer using the GroNLP/bert-base-dutch-cased pretrained model (12-layers, GELU activation).

4. For Reactions, all sub types of reactions were aggregated (likes, wow, haha, sad, angry and love).

5. Skewness for Positive words = 3.47 and skewness for negative words is 3.12.

Additional information

Funding

This work was supported by the BOF GOA

Notes on contributors

Jeroen Peeters

Jeroen Peeters is a PhD candidate at the University of Antwerp (Belgium) and a member of research group Media, Movements and Politics (www.M2P.be). His research focusses on the issue strategies and social media communication of politicians.

Michaël Opgenhaffen

Michaël Opgenhaffen (Ph.D) is Associate Professor of digital news and journalism at the Institute for Media Studies, KU Leuven, Belgium. His research focuses on the production and consumption of digital and social media news.

Tim Kreutz

Tim Kreutz is a PhD candidate at the Computation Linguistics Research Group (CLiPS) at the University of Antwerp. His main field of interest is Automatic Content Analysis of texts, specifically analyzing sentiment and emotional language in political documents.

Peter Van Aelst

Peter Van Aelst is a professor of political communication at the University of Antwerp (Belgium) and a founding member of the research group Media, Movements and Politics (www.M2P.be). His current research focuses on the relationship between politicians, journalists and citizens in the digital age. He has published extensively on agenda-setting, election campaigns and the interactions between journalists and politicians in comparative perspective.