Trust and Recall of Information across Varying Degrees of Title-Visualization Misalignment
Trust and Recall of Information across Varying Degrees of Title-Visualization Misalignment
复制标题
不同程度的标题-可视化错位对信息的信任和回忆
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
Karrie Karahalios
中科院分区：
文献类型：
--
作者：
H. Kong;Zhicheng Liu;Karrie Karahalios
可视化正在成为传播数字错误信息的一种手段。先前的研究表明，可视化的解释可以通过倾斜的标题来操纵，这些标题只支持视觉故事的一面，但人们仍然认为可视化是公正的。在这项工作中，我们研究是否继续存在这样的影响时，标题和可视化表现出更大程度的不一致：标题的信息不同于视觉提示的信息在可视化，和标题的信息与可视化相矛盾。我们发现，虽然标题与矛盾的倾斜触发更多的人识别偏见相比，标题与一个错误的倾斜，可视化持续认为是公正的大多数。此外，人们对可视化信息的回忆更频繁地与标题对齐，而不是可视化。基于这些结果，我们讨论了利用文本组件来检测和打击基于视觉的错误信息与基于文本的倾斜的潜力。
Visualizations are emerging as a means of spreading digital misinformation. Prior work has shown that visualization interpretation can be manipulated through slanted titles that favor only one side of the visual story, yet people still think the visualization is impartial. In this work, we study whether such effects continue to exist when titles and visualizations exhibit greater degrees of misalignment: titles whose message differs from the visually cued message in the visualization, and titles whose message contradicts the visualization. We found that although titles with a contradictory slant triggered more people to identify bias compared to titles with a miscued slant, visualizations were persistently perceived as impartial by the majority. Further, people's recall of the visualization's message more frequently aligned with the titles than the visualization. Based on these results, we discuss the potential of leveraging textual components to detect and combat visual-based misinformation with text-based slants.