Hope vs. Fear: How Public Opinion Shapes Political Communication in Referendum Campaigns

Over the last decades, research on the sentiment of political communication has primarily concentrated on its impact on political preferences (Basil, Schooler, and Reeves 1991; Abbe et al. 2000) and on voter turnout (Ansolabehere, Iyengar, and Simon 1999; Brooks 2000). Consequently, scholars working on negative electoral communication have also focused their researches on voting behavior. According to their work, negative communication is both more likely to be remembered by the electorate (Shapiro and Rieger, 1992) and more attention grabbing in general (Wentura, Rothermund and Bak, 2000) than positive communication. This is consistent with recent findings in psychology, according to which fear, as a primary emotion, dominates hope in the perception of individuals and collectivities (Jarymowicz and Bar-Tal, 2006). Research in political communication has shown that approximately one-third of campaign communication was negative (Sabato, 1981) and that negative ads frequently used fear appeals (Kaid, 1991). This framing of risk strategically presents a political choice as the decision between two avenues, the adverse option being pictured as the most risky by the transmitter (Stromback and Kiousis, 2011). That type of rhetoric is alleged to be especially effective for the status quo option during independence referendums (Nadeau, Martin and Blais, 1999). At this point in time, scholars have not studied the impact of fluctuation in voting intentions – as reported by polls – on a political organization's use of negative communication during referendum campaigns. In the literature, polls are rather used as an explicative variable for other trends: public opinion, turnout level, or mass media behavior. The literature doesn't provide an answer to the following question: what explains the fluctuation of the use of negative messages by political organizations during a given political campaign? This paper aims to answer this question by analyzing political communication produced by the YES and the NO organizations, Yes Scotland and Better Together, during the 2014 Scottish independence referendum campaign. Data consists of all the tweets distributed by the official Twitter account of both organizations (@YesScotland and @UK_Together) between June 15, 2014 and September 18, 2014. Results are obtained by a multinomial logistic regression. This analysis demonstrates that having an advance in the polls had a statistically significant influence on the tweet sentiment of each organization, but in a contrasted way: Yes Scotland's messages were more positive when it was ahead in the polls, while Better Together's messages were more negative when it was leading, and vice-versa. There also exists a sharp contrast of sentiment between primary and secondary types of political messages for both organizations. If one only looks at the primary messages written by the official accounts, the difference of sentiment is striking: the messages produced by the YES side are substantially more positive than those produced by the NO side. The secondary messages – written by third parties – shared by these same official accounts show the opposite pattern. This supports Macskassy's (2011) view that shared messages, at least on Twitter, tend to be different in nature from the ones produced by the organizations themselves.