The interactive effect of recommendation subjects and message types on consumers' suboptimal food purchase intentions

Abstract

Purchasing suboptimal foods is an effective measure to reduce food waste, but there is still significant consumer resistance in purchasing decisions. Message interventions are widely adopted by retailers as a convenient, flexible, and cost-effective strategy. The interactive effect of recommendation subjects and message types on suboptimal food purchase intentions is investigated in this study employing a 2 (recommendation subjects: AI vs. human) × 2 (message types: fact-based vs. affect-based) between-subjects experimental framework. It has been discovered that AI recommenders enhance purchase intentions when delivering fact-based messages, whereas human recommenders are more effective when offering affect-based messages. Moreover, the mediating role of green identity is confirmed. The interactive effect between recommendation subjects and message types is moderated by scarcity cues, with fact-based messages being more effective when combined with long-term scarcity cues, while affect-based messages being more effective when combined with short-term scarcity cues. In this work, the application of "algorithm appreciation effect" and "algorithm aversion effect" is expanded to the study of suboptimal food purchase intentions. It proposes that providing consumers with specific types of messages from AI or human recommenders can increase their purchase intentions, thereby offering theoretical support and practical insights for retailers' message strategies in suboptimal food marketing.