In recent years, social media channels like Facebook and Twitter have brought about a fundamental transformation of communication. Besides being utilized for personal communication, social media has also developed into a broadly and often exclusively used source of news and political information (Bode, 2015; Fletcher & Nielsen, 2018; Metzger & Flanagin, 2015). Given that anybody can produce and share information on social media, communication characteristics such as real time communication, unlimited distribution, high connectivity, and the lack of editorial supervision pave the way for floods of information which severely complicate social media recipients’ evaluation of the quality of online content. Therefore, it is crucial to gain a detailed understanding of how recipients assess the credibility of a message or source.

With regard to the question of how incoming information is processed, dual process models like the elaboration likelihood model (Petty & Cacioppo, 1986) and the heuristic systematic model (Chaiken, 1987) distinguish two ways of information processing which are activated depending on recipients’ ability and motivation. On the central route information is thoroughly processed, whereas on the peripheral route recipients are more likely to base judgments (e.g., concerning credibility) on simple cues. Furthermore, source, message, and meta-informational cues are assumed to trigger cognitive heuristics (Kruglanski & Gigerenzer, 2011; Metzger & Flanagin, 2015; Sundar, 2008) – mental shortcuts which do not include all available information to reduce the cognitive load (Shah & Oppenheimer, 2008) and are mostly unconsciously applied by individuals (Tversky & Kahneman, 1974).

To this end, three empirical studies were conducted to explore (1) the cues upon which recipients base their credibility judgments in social media communication, (2) whether cue patterns are universal among different platforms and communication contexts, (3) whether social media cues and features and their effect on credibility perceptions can be examined by applying automated methods, (4) whether the relation between cue and judgment underlies a cognitive heuristic and (5) whether the operation of this heuristic can be measured by means of task latencies indicating effort reduction as core function of cognitive heuristics.

As a first step towards a systematic investigation, Study 1 (N = 341; postings: 1366) aimed to investigate the role and interplay of cues available in social media communication and thus, tested the impact of source expertise, likes, shares, pictures, and topic involvement on evaluations of politicians’ Facebook postings. The results revealed that source cues led to higher credibility judgments, whereas higher numbers of likes and shares unexpectedly led to decreased credibility. Contrary to expectation, recipients’ involvement, need for cognition, and conformity with the message did not moderate the effects. The second study (N = 2626; ratings: 24823) sought to investigate not only a larger data set but also a different social media platform, namely Twitter. For large data sets, most researchers have suggested automated approaches to perform binary classification in order to determine information veracity, while studies have rarely considered recipients’ perspectives and multidimensional psychological credibility evaluations. To fill this gap and gain more insights into the impact of a tweet’s features on perceived credibility, a survey was conducted asking participants to rate the credibility of crisis-related tweets. The resulting 24823 ratings were used for an exploratory feature selection analysis, which revealed that credibility judgments are most affected by meta-informational features such as the number of followers of the author, the number of tweets produced, and the ratio between number of tweets and days since the creation of the author’s Twitter account. Even though these features are classically defined as meta-informational (as they represent numbers aggregated by the system), they are strongly connected to the source of information, in this case the Twitter account holder.

Since both studies thus demonstrated source cues to be the most important anchors for individuals’ credibility assessments, the third study (N = 185) was conducted to examine whether the relation between the expertise cue and resulting judgments and decisions is guided by a heuristic, namely the expertise heuristic. Therefore, a 2 (difference in expertise: yes vs. no) X 2 (number of conflicting cues: 1 vs. 2) x 2 (valence of additional cues: positive vs. negative) within-subject design was applied, asking participants to select one of two presented information sources described only by four attributes (source expertise, ratings of other users, picture, length) and related cue values. The findings indicate that the presence of the expertise cue reduced respondents’ task latencies significantly. Nevertheless, the use of the expertise cue was found not to be used independently from additional information such as valence of the additional cues. This contradicts the notion of attribute substitution (Kahneman & Frederick, 2002), according to which heuristics are based on only one single cue. In sum, the results contribute to a more detailed understanding of the expertise heuristic used in social media communication as triggered by the cue source expertise. Regarding the selection of information sources, recipients perceived source expertise to be the most important cue, and if this cue was present and positive for one of the given alternatives, the decision was easier and faster.

Overall, this dissertation provides empirical evidence regarding how social media recipients evaluate the credibility of content. This evaluation was found to be based mainly on source expertise cues, for Facebook as well as for Twitter, and for politicians’ postings as well as for crisis-related communication. Consequently, source cues accelerated individuals’ decision-making by means of effort reduction. In this vein, it can be argued that the relation between cue and judgment is guided by a cognitive heuristic, namely the expertise heuristic. In sum, the current work extended previous research on heuristics using self-reports or focus groups by providing empirical evidence through the measurement of effort reduction. The results can be further applied to develop support measures for users by highlighting relevant cues and features in the realm of interface design or media education.