How Second-Order Is the Regional Level? An Analysis of Tweets in Simultaneous Campaigns

Abstract

This article puts the second-order theory for regional elections to the test. Not by analysing voting behaviour but with the use of campaign data. The assumption that regional campaigns are overshadowed by national issues was verified by analysing the campaign tweets of Flemish politicians who ran for the regional or national parliament in the simultaneous elections of 2014. No proof was found for a hierarchy of electoral levels but politicians clearly mix up both levels in their tweets when elections coincide. The extent to which candidates mix up governmental levels can be explained by the incumbency past of the candidates, their regionalist ideology, and the political experience of the candidates.

Sampling Twitter users for social science research: evidence from a systematic review of the literature

“…The researcher has no control over or intervention in the selection process. In studies with customized sampling frames, the sample of Twitter users is obtained by random selection (e.g., Bouteca et al 2017) or by setting filters such as number (minimum or maximum) of followers or number of statuses posted (e.g., Kearney 2019). Finally, when the sampling frame comes from randomly generated numbers, the sample automatically includes those Twitter users whose ID numbers match the randomly generated numbers (e.g., Liang and Fu 2015;Liang et al 2016).…”

“…In 4 studies, researchers resorted to these services to request a sample of individuals interested in collaborating with the investigation (being a Twitter user was mandatory) (e.g.,Tominaga et al 2015;Fiesler 2018). • Customized list of Twitter users in 8 studies the researchers built or acquired a list representing the target population of the study, such as a list of candidates to elections(Bouteca et al 2017), a list of followers of 12 Twitter accounts (Kearney 2019) and a list of scientific tweeters(Yu et al 2019).…”