What’s the T? Personality Politics in Media Coverage Over Time: The UK Case 1992-2013

• In Event: Computational Methods for Studying Political Communication

Abstract

Over the last decades there has been a substantial amount of research about the phenomenon of the personalization of politics, especially in parliamentary democracies. The key tenet of the literature is that there has been an increase over time in the centrality of individual politicians at the expense of collective institutions (parties, cabinets and parliaments), which is manifested in, and in turn is reinforced by, personalized media coverage (e.g. Balmas et al. 2014; Boumans, Boomgaarden and Vliegenthart 2013; Karvonen 2010; Kriesi 2012; Mughan 2000; Rahat and Sheafer 2007). Within media personalization, the literature suggests that there has been a rise not only on the emphasis on individuals but also a greater focus on their personalities, including ‘hard’ and especially ‘soft’ character traits, which is understood to come in detriment to the coverage of policies and issues (Langer 2011; Van Aelst, Sheafer and Stanyer 2012; van Zoonen and Holtz-Bacha 2000). Most research, however, has focused on the first dimension (i.e. individualization) with much less systematic empirical work carried out on the personality politics dimension, especially outside the US. This is particularly problematic as it is the dimension that raises stronger normative concerns due to its potential impact on the quality of information citizens receive, electoral behavior and the quality of elected representatives.

To a large extent this strand of research has been limited by methodological difficulties. The complexity of analyzing this dimension of personalization has meant that, up to now, it generally has been carried out with manual content analysis, and hence only in relatively small samples of data. As a result, there is not enough evidence to confirm (or reject) the assumption that personality—and especially ‘soft’ traits—has become a major focus of media coverage, and even less to help us understand under which circumstances this is more the case. Thus, this study seeks to better understand the degree to which media focus on politicians’ personalities, and the characteristics that they choose to portray. Specifically, it aims to answer these three research questions: How much of the coverage focuses on politicians’ personalities and how does it compare to that with policy-focus? Which types of personality traits predominate? Has the characteristics of the coverage changed over time and which factors explain these changes? In order to answer these question, we have created a novel dataset constructed from over one million newspaper articles in the UK covering a continuous 22-year period (1992-2013). This dataset will enable us to carry out the analysis via computational quantitative text analysis techniques. Our approach first breaks newspaper articles into sentences, which then become the basis of analysis. These sentences are subsequently analyzed on two fronts. First, we identify policy and personality-focused content. In a second step, we categorize the personality-focused-text referring to party leaders in particular according to the types of traits they emphasize. Specifically, these sentences (over 26 million in total) are analyzed via Keyword-in-Context to keep the meaningful words before and after each politician’s name (our search terms). The selected words are subsequently analyzed for their content to classify coverage into categories via a Latent-Dirichlet Allocation algorithm. We furthermore use dictionaries to categorize words and clusters into appropriate policy dimensions and appropriate brand personality categories.

Our preliminary results indicate that there are four main types of coverage: policy, process, party, and personality. It should be noted that these types are co-occur in the same article and hence the importance of carrying out the analysis at the sentence level. The proportion of coverage focused on personality is relatively low and does not appear to have systematically increased over the period under study. Instead, it fluctuates both as a result of variables related to the individual and due to institutional factors such as campaign periods and the presence of a coalition government. It also varies depending on newspaper type, although not as much as might be expected. We do not yet have findings to report on the types of personality traits but it would appear that competence (or hard character traits) are more prevalent than those associated with ‘likability’ (or soft character traits). Overall these findings posit an interesting contribution to the literature. First, they come to show the exact extent to which, both during electoral and government periods, the press focuses on the personality of politicians and how this compare with other—and often argued more relevant—type of content. It also reveals some of the mechanisms through which coverage is personalized. We manage to do so with—in our knowledge—the largest coverage of any previous research project on the topic. Second, it offers an innovative method to identify personalized coverage, as well as other types of coverage. This will enable further research not only on personality coverage itself but also linking it to changes in public opinion, voting behavior, and even politicians reactions to media coverage. Finally, they help to provide some sound empirical bass for the much discussed normative questions about to the role of personality on the media reporting of politics.

Authors

• Inaki Sagarzazu, Texas Tech U

• Ana Langer, U of Glasgow

• Johannes Gruber, U of Glasgow