Algorithm as Recommending Source and Persuasive Health Communication: Effects of Source Cues, Language Intensity, and Perceived Issue Involvement

Algorithms are now playing significant roles in online health information selection and recommendation. A question arises as to when and why people would be persuaded by the content they recommend. We conducted a 4 (recommending source: algorithm, other users, a friend, the CDC) x 2 (language intensity: high vs. low) experiment to find out. Participants (N = 299) were exposed to a health-related public service announcement embedded in a social media post. The results showed that overall, an algorithm induced a similar level of compliance intention compared with other recommending sources. We also found a significant three-way interaction when comparing the effects of the algorithm and the CDC: for individuals with low issue involvement, the algorithm was less persuasive when paired with a message with high language intensity. In contrast, for high-involvement individuals, the algorithm elicited more fear than the CDC when recommending an assertive message, partially leading to more compliance intention.