Behavioral interventions motivate action to address climate change

Abstract

Mitigating climate change requires urgent action at individual, collective, and institutional levels. However, individuals may fail to act because they perceive climate change as a threat that is distant or not personally relevant, or believe their actions are not impactful. To address these psychological barriers, we conducted a large-scale "intervention tournament." In a sample of 7,624 participants, we systematically tested 17 interventions that targeted psychological mechanisms described by three key themes: Relevance, Future Thinking, and Response Efficacy. Interventions that emphasized social relevance were the most effective for motivating people to share news articles and petitions about climate change. Interventions that targeted future thinking were the most effective for broadly motivating individual actions (e.g., driving less, eating vegetarian meals) and collective actions (e.g., donating, volunteering) to address climate change. Interventions that emphasized the environmental impact of these actions reliably increased the perceived impact of pro-environmental actions, but did not consistently motivate action. Notably, interventions that targeted two or more mechanisms-such as imagining a future scenario that involved oneself or close others-were most effective. Importantly, our leading interventions were substantially more effective than prevalent existing strategies (e.g., carbon footprint information). Our findings are relevant to theories of behavior change, motivation, and information sharing, with potential applications across domains. Insights from our tournament could be applied to develop scalable online interventions and mass communication campaigns to address climate change.

Keywords: behavior change; climate change; information sharing; pro-environmental behavior; psychology.

Fig. 1.

Overview of interventions tested in the tournament, organized into three key themes: Relevance (Top), Future Thinking (Left), and Response Efficacy (Right). Some interventions, indicated in overlapping portions of the theme circles, leveraged multiple psychological mechanisms.

Fig. 2.

Overview of the intervention tournament. The Left panel lists all interventions tested; for some intervention strategies, we tested multiple variations. Where applicable, these subgroups are labeled in smaller boxes to the right of each parent intervention label. Interventions targeted different psychological mechanisms, indicated here with three color-coded themes: Relevance, Future Thinking, and Response Efficacy. Some interventions targeted multiple mechanisms (Fig. 1), marked with multiple colored bars to the right of each intervention box. For multitheme interventions, the leftmost box indicates the primary theme. The Control group did not complete any intervention task, and proceeded directly to completing the outcome measures after providing consent. Participants were randomly assigned to a group in a between-subjects design. The Right panel illustrates the primary outcome measures: ratings of future intentions and perceived impact regarding pro-environmental individual and collective actions related to climate change, and intentions to share news headlines and petitions about climate change. All participants completed the same set of outcome measures. In addition to the primary outcomes illustrated here, participants also completed a battery of secondary outcome measures, described in detail in SI Appendix. * denotes intervention conditions that were tested in a second wave of data collection; all interventions were compared with the same control group for consistency.

Fig. 3.

Results from the Climate Action Task, including action intentions (Left) and perceived impact of pro-environmental behaviors (Right). Results shown are estimates derived from Bayesian mixed-effects regression models. Point estimates indicate the treatment effect for each intervention condition (Intervention–Control, comparing the median values from each posterior distribution). Error bars mark 95% credible intervals surrounding the point estimates. Dependent variables were z-scored to provide standardized effect sizes. Dotted lines marks zero (no effect; no difference from Control group). Points are color-coded to reflect the three intervention themes: Relevance, Future Thinking, and Response Efficacy. Note that some interventions can be described by more than one theme (Figs. 1 and 2); colors here indicate a primary theme for each intervention.

Fig. 4.

Results for intentions to share news articles (Top) and petitions (Bottom) about climate change broadly on social media (“broadcast” sharing). Results for narrowcast sharing are visualized in SI Appendix, Fig. S2. Results shown are estimates derived from Bayesian mixed-effects regression models. Point estimates indicate the treatment effect for each intervention condition (Intervention–Control, comparing the median values from each posterior distribution). Error bars mark 95% credible intervals surrounding the point estimates. Dependent variables were z-scored to provide standardized effect sizes. Dotted lines marks zero (no effect; no difference from Control group). Points are color-coded to reflect the three intervention themes: Relevance, Future Thinking, and Response Efficacy. Note that some interventions can be described by more than one theme (Figs. 1 and 2); colors here indicate the primary theme for each intervention.