‘Look at How Corrupt They Are!’: How Anti-Political Discourse from Politicians Affects Their Own Image and the Image of Politics

ABSTRACTIn this article, we use a survey experiment (N = 1,626) to test whether anti-political Facebook messages from politicians improve the character evaluation of the messenger while damaging the image of politics. We take account of people’s preexisting anti-political attitudes and their support for the political messenger as important moderators. Results show that anti-political discourse from politicians has minimal effects on the attitudes of audiences toward the messenger and toward politics. We find limited support for three underlying mechanisms: anti-political messages give (some) audiences new impressions that improve the personal image of the messenger (impression formation), anti-political discourse from politicians alienates some of their supporters (backlash effect), and anti-political discourse has more effect on people who do not hold anti-political attitudes (attitudinal incongruence). Overall, our results show that politicians reap little benefit from using anti-political discourse, as such discourse may worsen their own image and the image of politics among certain audiences.DisclaimerAs a service to authors and researchers we are providing this version of an accepted manuscript (AM). Copyediting, typesetting, and review of the resulting proofs will be undertaken on this manuscript before final publication of the Version of Record (VoR). During production and pre-press, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal relate to these versions also. Disclosure StatementWe have no known conflict of interest to disclose.Notes1 Participants exposed to a favored political party read a Facebook message from: Vlaams Belang (15.8% of participants; the Flemish populist radical-right party), N-VA (12.7%; Flemish nationalist party), Open Vld (12.7%; the liberal party), CD&V (13.4%; Christian democratic party), sp.a/Vooruit (11.4%; the socialist party), Groen (14.1%; the green party), or PVDA (7.3%; the workers’ party). Participants exposed to an unfavored political party read a Facebook message from: Vlaams Belang (39.1%%), N-VA (10.6%), Open Vld (6.3%), CD&V (4.2%), sp.a/Vooruit (11.2%), Groen (10.4%), or PVDA (18.1%)..2 We use the package “lavaan” in R (Rosseel, Citation2012). All models were estimated using WLSMV estimation.3 Our interpretation of the mediation analysis assumes that all relevant mechanisms are considered. This assumption is tentative given that (a) some important variables might not have been considered in the mediation analysis and (b) the causal (cognitive) mechanisms linking the mediator to the outcome variable(s) were not directly observed. For a full discussion on the problem of unobserved variable bias (or confounders) in causal analysis see Imai et al. (Citation2011).Additional informationNotes on contributorsDieter DekeyserDieter Dekeyser holds master’s degrees in criminology (2012), sociology (2014), and statistics (2015), and received his PhD in sociology at Ghent University (Belgium) in 2022. His research interests include populism, political attitudes, media effects, public opinion research, cognitive sociology, and research methodology.Henk RooseHenk Roose received his PhD in sociology at Ghent University (Belgium). He is currently working there as an associate professor teaching Methodology and Cultural Sociology. His research focuses on the link between cultural participation and social stratification using Geometric Data Analysis, the empirical analysis of aesthetic dispositions and survey methods.