Sentiment analysis of the agenda of the Spanish political parties on Twitter during the 2018 motion of no confidence. A compositional data approach

• Main Authors: Marc Blasco-Duatis, Germà Coenders
• Format: Article
• Language: English
• Published: Universidad de Alicante 2020-07-01
• Series: Revista Mediterránea de Comunicación
• Subjects: sentiment analysis; agenda-setter; attribute agenda-setting; compositional data (coda); twitter; biplot
• Online Access: https://www.mediterranea-comunicacion.org/article/view/2020-v11-n2-analisis-de-sentimiento-de-la-agenda-de-los-partidos-politicos-espanoles-en-twitter

This paper provides a sentiment analysis of the political parties’ discussion on Twitter about the motion of no confidence in the Spanish government in 2018. In particular, 2824 tweets from the official accounts of the 13 political parties represented in the Congress of Deputies were extracted and analysed. The methodology applied compositional data analysis and its visualisation employing Biplot (a visualisation tool that enables contrasting the relative importance of the elements under study). Unlike traditional approaches, this study emphasises the relative importance of the issues within the agenda while incorporating a third category, the sentiment analysis. The main findings concern the reliability of the method to represent compositionally the agenda and the agenda-setters, as well as the sentiment analysis. This confirms that the issues most notably associated with certain parties also reflect their projection on sentiment. In sum, this analysis sheds light on sentiment representation in the agenda-setters (attribute agenda), especially in the field of political communication.