Information credibility perception on Twitter

Information on Twitter is vast and varied. Readers must make their own judgements to determine the credibility of the great wealth of information presented on Twitter. This research aims to identify the factors that influence readers' judgements of the credibility of information on Twitter, especially news-related information. Both internal (within the Twitter platform) and external factors are studied in this research. User studies are conducted to collect readers' perceptions of the credibility of news-related tweets, Twitter features, and the impact of reader characteristics, such as a reader's demographic attributes, their personality and behaviour. Twitter readers are found to depend solely on surface tweet features in making these judgements such as the author's Twitter ID, pictures, or the number of retweets and likes, rather than the tweet's metadata as recommended in previous studies. In this study, surface features are related to cognitive heuristics. Cognitive heuristics are features that the mind uses as shortcuts for making quick evaluations such as deciding the credibility of tweets. There are three main types of cognitive heuristic features found on Twitter that readers use to determine credibility: endorsement, reputation and confirmation. This study finds that readers do not use only one single feature to make credibility judgements but rather a combination of features. External factors such as a reader's educational background and geolocation also have a significant positive correlation with their perceptions of a tweet's credibility. Readers with tertiary level education, or living in a certain location or environment, such as in a crisis or conflict area, are observed to be more careful in making credibility judgements. Readers who possess conscientiousness and openness to experience personality traits are also seen to be very cautious in their credibility judgements. Another insight provided by this research is the categorisation of readers' behaviours according to credibility perceptions on Twitter. The behavioural categorisations are defined by readers' behavioural reliance on Twitter's surface features when judging the credibility of tweets. The findings can assist social media authors in designing the surface features of their social media content in order to enhance the content's credibility. Furthermore, findings from this research can help in developing effective credibility evaluation systems by considering readers' personal characteristics