![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
ABSTRACT
Given the importance of issue competition in a West European context and the growing use of Facebook in elections, this paper studies how politicians use Facebook to shape the campaign agenda. We analyze the issues addressed in 6,388 Facebook posts by candidates in the Danish 2011 parliamentary election. A limited share of Facebook updates is dedicated to issues. The Facebook agenda did not respond to standings in the polls, nor to the media agenda or public agenda. Comparing issue engagement of new candidates and rerunning candidates we find that the Facebook campaign agenda is not simply politics as usual.
Notes
1. In this paper, sitting MPs who rerun for office are referred to as “rerunners,” and challenging candidates who are not in parliament are referred to as “new candidates.” We avoid using the term “incumbents,” since in the Danish electoral context this may also refer to candidates who belong to the governing party.
2. Information on personal pages (even with the most relaxed privacy settings) is not accessible unless one has a friendship status on Facebook. Although there are personal pages for which the Facebook wall is visible even without friendship, the activity on these walls cannot be gathered and used for analysis.
3. There are six types of status categories: status, photo, link, video, swf (flash), and music. Although they reflect different types of status updates, all have some associated text (such as the poster’s words on the link, photos). We use all these text data as well. The distribution on the different types of updates is: 1,872 link, 5 music, 1,143 photos, 2,805 simple status, 13 swf, and 545 videos (plus 5 uncategorized types).
4. For a description of the techniques used to develop the dictionary see Hobolt and Klemmensen (Citation2008).
5. More precisely, this index calculates the ratio between the words that are found in both the dictionary and the texts and the total number of words in both the texts and the dictionaries. Its main advantage is that it makes the index highly comparable across texts and dictionaries with different lengths. However, it is also a calculation method that biases the identified issue content downward, because the intersection between two sets does not take into consideration whether a word from a dictionary appeared once or ten times in a document. Alternative operationalizations are of course possible (such as taking into account how many times a dictionary word appears in a text, or how many of the total words in a text are also found in the dictionary), but these are more sensitive to differences in text length. Furthermore, these operationalizations indeed influence the absolute value of the “issue talk,” but they yield the same results in the subsequent comparative analysis.
6. On August 27 and on September 10 no Megafon poll was carried out. For these days, the calculation of average scores is only based on the three other available national polls.
7. This possibility would be available in a vector autoregressive model (VAR) specification. Although VAR models are employed in similar analyses of online communication data (Bode et al., Citation2011) or in the analysis of communication reciprocity by different actors (Adams et al., Citation2005), they are also very costly in terms of the number of parameters to be estimated. For these exact reasons of possible overfitting with a sample size of 21, we decide to specify the simpler Granger models.
8. We chose the 21-post cut-point because the official campaign ran for 21 days.
9. The fact that party leader communication is a large proportion of the rerunner communication and we see such coherence patterns confers face validity to our measurement, because party leaders are crucial in deciding the election message and plan.
10. It is impossible to analyze both campaign stage and individual-related differences, because we cannot subset our text data to meaningfully capture a daily individual Facebook activity. Simply put, we do not have enough candidate Facebook communication gathered on a daily basis.
11. To reiterate, they might not voice the same opinions on these issues. What we analyze here is the rank of attention given to an issue, or in other words, relative salience.
12. Besides the amount of posts, the number of issues each candidate actually mentions (from the 12 coded issues) has an even more important role in this analysis. Many not-mentioned issues from one candidate suppress the coherence score. Only 21 candidates actually mention all twelve issues to some extent (13 new candidates and 8 rerunners), and for these candidates, the coherence scores are between 0.69 and 0.90. We carried out an additional analysis controlling for the number of issues mentioned, and for those cases there are no significant differences between rerunners and new candidates.
Additional information
Notes on contributors
Arjen Van Dalen
Arjen van Dalen is an associate professor at the Centre for Journalism, Department of Political Science and Public Management, University of Southern Denmark. His research interests are in the role of media in politics, in particular the relation between politicians and journalists.
Zoltán Fazekas
Zoltán Fazekas is a postdoctoral research fellow in the Department of Political Science and Public Management, University of Southern Denmark. His research focuses on political behavior on the voter and the elite level, and how our characteristics as human beings influence political attitude formation and political decision making.
Robert Klemmensen
Robert Klemmensen is a professor of political science in the Department of Political Science and Public Management, University of Southern Denmark. His research focuses on comparative political behavior, and it has appeared in journals such as Political Psychology, Comparative Political Studies, and Comparative Politics.
Kasper M. Hansen
Kasper M. Hansen is a professor in the Department of Political Science, University of Copenhagen. He has published extensively on voting behavior, turnout, and political campaigns. His work has been published in Political Behavior, Political Communication, and European Journal of Political Research, among other journals. See http://www.kaspermhansen.eu.