The effect of exposure to corruption information on specific and diffuse political trust: evidence from a survey experiment

How does the disclosure of factual information about corruption affect different dimensions of political trust? Existing literature suggests that exposure to corruption will have a strong negative effect on people’s reported levels of political trust. However, little empirical work has been produced to identify whether that effect is driven by a decrease in the individual’s trust in particular political agents and institutions or by a drop in the confidence they express with regard to the principles of the political regime as a whole. This thesis is the first to address this research question from an experimental angle. I employ a vignette-based survey experiment, based on original data collected from an online survey conducted in March 2020, through which I seek to measure the extent to which revealing information about corruption deteriorates various dimensions of political trust. I further analyze how political trust judgments change when individuals are confronted to different types of corruption, namely personalized or institutionalized. I find that when corruption revelations do not provide sufficient information to effectively attribute blame to particular political objects, specific political trust and trust in the principles of democracy deteriorate the most strongly, while trust in the overarching institutions of the state remains intact. This effect is amplified when the vignettes identify bribery as the most common form of corruption, which I theorize is the result of blame being primarily attributed to political agents. When the vignettes identify lobbying as the most common form of corruption, however, trust in the overarching institutions of the state suffers a meaningful decrease as well, which I argue occurs because blame attribution is no longer exclusively personal but indicates the existence of a systemic or procedural flaw. Given the lack of inter-temporal outcome measurement, I suggest that these results should be interpreted with caution and perhaps complemented with further longitudinal analysis.