How consumers process sustainability messages in a randomized two-wave experiment on household food waste reduction in Japan

Abstract Information campaigns are widely used to reduce household food waste, yet little is known about how consumers process the messages they receive. This study investigates whether differently framed messages trigger qualitatively different cognitive-moral responses and whether randomized message assignment changes self-reported outcomes. In a randomized two-wave experiment with Japanese consumers (Wave 1 n = 2069; Wave 2 n = 959, five-week follow-up), participants received information framed around tax burdens (Tax frame), global hunger (Hunger frame), household monetary costs (Cost frame), or a general-information control. Open-ended responses were classified into five Norm Activation Theory (NAT; [31]) stages using two independent large language model coders with researcher adjudication, supplemented by a blinded author re-coding exercise. Framing causally shifted message interpretation: the Tax and Hunger frames increased the probability of ascribing responsibility (+ 5.1 and + 5.8 percentage points), while the Cost frame uniquely elevated personal norm activation (+ 4.5 points). However, intention-to-treat estimates show no effects on self-reported preventive practices, normative orientation, or policy support. The one persistent effect is negative: five weeks later, the Tax frame reduced food waste reduction motivation (− 0.15 SD), a pattern consistent with motivational crowding-out, although the mechanism was not directly tested. Information framing changes cognitive-moral interpretation but not self-reported practices—a process-level distinction that helps explain the intention–behavior gap and informs the design of household sustainability communication.