Dual paths of information behavioral choice influenced by information and emotion: The moderating effect of user credibility
Article excerpt
by Na Wang, Yaming Zhang, Yaya Hamadou Koura In public opinion events, social media is often used as an important channel for the public to collect, process, and share information, and the topics and information on social media stimulates people…
by Na Wang, Yaming Zhang, Yaya Hamadou Koura
In public opinion events, social media is often used as an important channel for the public to collect, process, and share information, and the topics and information on social media stimulates people to behave in creating related contents. This study integrates the elaboration likelihood model (ELM) to examine how information topics (central cue) and emotional responses (peripheral cue) shape user information behavior (likes, comments, forwards), and how user credibility moderates these effects. 56,663 Weibo posts (August 22 to September 5, 2023) about “Japan’s announcement of nuclear contaminated water discharge” were used for empirical analysis. Topics were identified via Latent Dirichlet Allocation (LDA), yielding four categories: Japanese aquatic products trade (Topic 1, 1.850%), seawater blackening (Topic 2, 8.822%), radioactivity exceedance (Topic 3, 13.286%), and global impact (Topic 4, 76.043%). Seven emotions were scored using the Dalian University of Technology affective lexicon. User credibility was coded from verification type. Baseline regression controlling for richness, mention, hashtags, URL, and text length showed that information topics significantly increased likes (β range: 0.164 to 0.333, p < 0.001), comments (β range: 0.063 to 0.138, p < 0.001), and forwards (β range: 0.083 to 0.156, p < 0.001), with the largest effects on likes. For emotions, anger, disgust, good and surprise positively influenced information behaviors (β range: 0.004 to 0.060, p < 0.01 to p < 0.001), while sadness, happy and fear showed negative or non‑significant effects. User credibility positively moderated all these relationships, with interaction β values ranging from 0.010 to 0.066 for topics and from 0.006 to 0.040 for emotions. Likes were the most influenced behavior, followed by forwards, then comments. The study demonstrates that information topics and emotions, moderated by user credibility, drive distinct information behaviors. Authorities should leverage high‑credibility users, highlight impactful topics, and monitor emotional tones, especially negative ones, to guide public opinion effectively.