Manipulating the Perception of Credibility in Refugee Related Social Media Posts

Abstract

This paper describes a controlled web-based study (n=126), investigating whether the perception of the credibility of refugee-related Tweets can be influenced by cues already reported in the literature for social media content generally. We provide empirical evidence that both a Tweet's popularity and the presence of links -even neutral links created by URL shortening services -may increase a user's belief that the Tweet contains credible information. This is important because the propagation of false information relating to refugees on social media sites has been well documented.

The Impact of Twitter Features on Credibility Ratings - An Explorative Examination Combining Psychological Measurements and Feature Based Selection Methods

“…Tweets of authors with numerous followers will benefit from large amounts of likes and retweets which might also persuade readers of the quality of the posting. This is in line with results of a user study by Aigner and colleagues [16] who found that credibility ratings mainly depend on the number of retweets and likes indicating a bandwagon effect. In this way, likes and retweets can be understood as recommendations of content by other users and might be taken as an anchor for rating something as credible.…”

“…An interesting finding is presented by Aigner and colleagues [16] who conducted a study focusing on how recipients evaluate the believability of news on twitter in the area of refugee related information. They demonstrated that tweets were rated as more credible if they received a higher number of retweets and likes, and that this is even true if the tweets were factually false.…”

“…predictions through the comparison of models which are designed to detect the veracity of facts [4], [13], [14], [15] or investigate isolated aspects potentially influencing perceived credibility in lab scenarios with self-reported measurements [16], [17], [18]. This is not to say that research on the veracity of facts is not timely or important, however, we decided to focus on perceived credibility here, as the corresponding psychological mechanisms are understudied.…”

Improved LDA Model for Credibility Evaluation of Online Product Reviews

“…M.J. Metzger et al [6] proposed a credibility assessment model based on Naive Bayes algorithm. J. Aigner et al [7] conducted a controlled web-based study to investigate whether the perception of the credibility of refugee-related tweets can be influenced by cues already reported in the literature for general social media contents. M. Alrubaian et al [8] proposed a model consisting of six integrated components operating in an algorithmic form to assess the credibility of tweets.…”

“…(5), p(d|t) represents the probability of the distribution of document d in the case of known topic t, which can be derived from the Bayesian formula shown as Eq. (7). In order to get document-topic distribution, we multiple π and θ to get a matrix denoted as σ wihich size is M × K. Based on the definition in 3.3, the distribution π can be considered as M matrices with dimension 1 × S, while θ can be regarded as M matrices with dimension S × K. Multiplying the matrix with dimension 1 × S by the matrix with dimension S × K, a matrix with dimension 1 × K can be obtained.…”