![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
Literature on agenda-building dynamics has neglected to assess the impact of contextual factors on the interplays between the issue attention of political actors and of the media. I fill the void by highlighting how media storms cause significant changes to the electoral communicational environment. Using a custom dataset compiled through an automated content analysis, I empirically examine patterns of issue salience during the 2015 Canadian federal election. The results support three main points. First, media storms do emerge during election campaigns. Second, media storms cause two main types of changes in the informational environment that characterize non-storm periods: (1) a reduction in the variety of issues included in the daily campaign coverage, and (2) a higher concentration of media attention on the storm-generating issues. Third, coverage of media storms compels political parties to engage with them, especially if they can exploit these storms with minimal risks. These findings suggest that some electoral contexts may be less conducive to political actors’ influence. They also offer evidence in support of the mediatization theory, according to which media market logic can take precedence over political normative logic in guiding the decisions of political actors.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Boydstun, Hardy, and Walgrave (Citation2014) also name “multi-media-ness” as a fourth criterion, indicating that the first three criteria should “register as such across multiple news outlets in a given media system” (p. 511). This criterion is implicitly incorporated in my research design, which includes two leading national newspapers and two leading nightly newscasts.
2 Press releases were collected daily from each party's campaign website. Facebook posts and tweets were collected from each party's official account (English language). The final dataset includes 1309 newspaper articles, 296 televised news reports, and 2751 political messages, unequally divided among the parties – 537 for the Conservative Party, 958 for the New Democrats, and 1256 for the Liberals.
3 Both the transcripts of the nightly news broadcasts and the newspaper articles were available through the Eureka database. All content was searched using the following keywords: “Harper” OR “CPC” OR “Conservatives” OR “Trudeau” OR “LPC” OR “Liberals” OR “Mulcair” OR “NDP” OR “New democrat*” OR “elect* AND Canad*”. All news reports were manually examined to remove non-relevant content.
4 After having activated the exclusion list – a list of words with little semantic values such as pronouns, conjunctions, etc. – and the lemmatization process, which reduces words to their lemma, i.e. the base form of all of its inflectional forms. The second process allows word like “studies” and “studying” to be considered as declination of the word “study”.
5 For example, sovereignty was associated with Quebec independence and with the military defense of Canada's national sovereignty. Two rules were created: (1) Sovereignty needed to be in the same sentence as Quebec, Clarity Act, or Referendum for the first case, and (2) in the same sentence as Arctic or North in the second case.
6 The policy domains are Culture, Economy, Environment, Ethics, Foreign Affairs, Immigration, Justice, Military, Politics, Science, and Welfare.
7 For example, the “Duffy” campaign issue is about a Conservative senator who was accused, among other things, of improperly claiming primary residency outside of Ottawa, Canada's capital, in order to claim living expenses for his time working there. Political operatives close to the prime minister were involved in the affair, which soon became a symbol of Harper's ethical lapses. The dictionary tracked this issue's presence through words such as “Duffy”, “Novak”, “Wallin”, “Wright”, and “Bayne”. However, it would be highly misleading to consider each appearance of these keywords to be a separate mention of the Duffy issue.
8 The corpus includes 38,532 paragraphs.
9 In this paper, storms are assessed through their presence in the daily electoral coverage. Therefore, to be classified as a storm, a given issue as to be mentioned in at least 20% of all daily paragraphs that include a reference to at least one electoral issue. In their seminal paper on the question, Boydstun, Hardy, and Walgrave (Citation2014) operationalized the concept as issue presence in the first three pages of their selected mass media outlets, where the whole article is the unit of analysis and where they (subjectively) coded what was the “primary” (p. 518) issue under discussion. I think that my operationalization is more stringent, as it based on a more precise unit that implicitly accounts for the amount of attention devoted to all issues. It does not, however, account for non-electoral coverage.
10 Another issue almost met those thresholds. The “veiled oath of citizenship” controversy obtained a mean media coverage of 18.9 percent between September 29 and October 12. It was not included in the media storm pool because I wanted to keep the same thresholds as the other studies on media storm.
11 This value is very similar to that of paragraphs mentioning both the Duffy issue and the New Democrats. This is probably explained by the fact that the NDP's communications regarding Duffy's trial were quite negative, with an average TS of –1.90. In comparison, the average TS of the Liberals’ paragraphs mentioning Duffy's affair was 1.02.
13 As a robustness check, I performed a similar procedure including only significant issues – those mentioned in at least two different paragraphs on the same day. Results were of similar direction and amplitude, but significant at only the p < 0.1 threshold.
14 As a robustness check, I also tested this proposition by computing the daily skewness of every day of storm and non-storm coverage. Results were similar, and the differences between both groups of values were significant, as revealed by an ANOVA test (p < 0.001).
15 For the sake of parsimony, I included only issues with an average daily presence higher than 1 paragraph.
16 This is an adjusted value; 5 days in which the New Democrats did not publish anything mentioning issues have been excluded.
Additional information
Notes on contributors
David Dumouchel
David Dumouchel recently received his Ph.D. from Université de Montréal. He is a member of the Groupe de recherche encommunication politique and of the Centre for the Study of Democratic Citizenship. His main research interests are agenda-building,electoral campaigning, and electoral tactics and strategies.