The present study investigates the role of source characteristics, evidence quality, crowdsourcing platform, and prior beliefs of the topic in adult readers' credibility evaluations of short healthrelated social media posts. Researchers designed content for the posts concerning five health topics by manipulating source characteristics (source's expertise, gender, and ethnicity), accuracy of the claims, and evidence quality (research evidence, testimony, consensus, and personal experience) in the posts. Then, accurate and inaccurate posts varying in these other manipulated aspects were computer-generated. Crowdworkers (N = 844) recruited from two platforms were asked to evaluate the credibility of ten social media posts, resulting in 8380 evaluations. Before credibility evaluation, participants' prior beliefs on the topics of the posts were assessed. Results showed that prior belief consistency and source expertise most affected the perceived credibility of accurate and inaccurate social media posts after controlling for the topic of the post. In contrast, the quality of evidence supporting the health claim mattered relatively little. In addition, the data collection platform had a notable impact, such that posts containing inaccurate claims were much more likely to be rated higher on one platform compared to the other. Implications for credibility evaluation theory and research are discussed.

翻译：本研究探讨了信息来源特征、证据质量、众包平台以及读者对主题的先前信念在成年读者对短篇健康相关社交媒体帖子可信度评估中的作用。研究者针对五个健康主题设计帖子内容，通过操控信息来源特征（信息源的专业性、性别和种族）、主张的准确性以及帖子中的证据质量（研究证据、证词、共识和个人经验）。随后，计算机生成了在这些操控方面各异的准确与不准确帖子。从两个平台招募的众包工作者（N=844）被要求评估十篇社交媒体帖子的可信度，共获得8380条评估结果。在可信度评估前，参与者对帖子主题的先前信念被评估。结果显示，在控制帖子主题后，先前信念一致性和信息源专业性对准确与不准确社交媒体帖子的感知可信度影响最大。相比之下，支持健康主张的证据质量影响相对较小。此外，数据收集平台产生了显著影响，例如包含不准确主张的帖子在一个平台上被评高分可能性远高于另一平台。文中还讨论了这些发现对可信度评估理论与研究的启示。