Social norms and climate-friendly behavior of adolescents

children, in particular, have an initial tendency to rely on adults for social information [61, 62]. Thus, we believe that special caution is warranted when examining the effect of peer infor- mation on children’s and adolescents’ beliefs and behavior. The fact that empirical expecta- tions shifted in response to the information about others’ low donations indicates that the adolescents in our sample did notice the treatment information. However, we can of course not rule out the possibility that a stronger intervention, with the drawbacks just mentioned, may also have triggered a behavioral change. While the subtle treatment manipulation may explain the overall results, it does not explain our age-specific results. After all, the eighth graders reacted as predicted: they conformed to the observed behavior and donated less to the CO2 offsetting scheme. The older students in Grades 9 and 10 also changed their empirical expectations in response to the information that peers donated low amounts, but, in contrast to the younger students, this information did not affect their behavior. We see two possible explanations for this age difference, which may also explain why, on the aggregate level, the peer information had only a limited effect. First, previous studies have shown that learning about the behavior of others is more effective when the reference group exhibits more similarities [43, 46, 48]. For example, Agerstro¨m et al. [63] provided university students with information about the donation behavior of either students from the same uni-

versity or of students in Sweden in general. Only the information about the donation behavior of students from the same university increased average donations. To account for this deter- mining factor, in our experiment, we chose a reference group that objectively shares many similarities with the surveyed adolescents; namely, they were the same age and from the same school type and city. However, it is possible that the chosen reference group was not specific enough for the adolescents, especially for the older students in Grades 9 and 10. After all, young people at this age are in the process of finding their own identity, and differentiation from the crowd is an essential part of this process. As a consequence, older adolescents may have a narrower reference group than younger ones. Our age-specific findings could align with this explanation: younger students in our sample considered students from another school (but the same age and city) as a reference group and thus updated their empirical expec- tations after receiving the information about these students’ donation decision and eventually adapted their own behavior. Older students, on the other hand, had a narrower reference group; while they also changed their empirical expectations, they did not change their behavior. Another, possibly complementary, explanation is based on the stages of moral develop- ment. The ability to make principled moral judgments develops during adolescence [64, 65]. As a result, older adolescents are less guided by intrapersonal conformity; they reason more in

terms of internalized norms and values. Applied to our study, this means that the older stu- dents likely already had stronger personal normative beliefs. This is indicated, for example, by the fact that they had a higher likelihood of having attended Fridays for Future protests, and that in the control group older students donated more than younger students. If the older stu- dents were more likely to form their donation decision on basis of their personal normative beliefs, then it is not surprising that the shift in empirical expectations did not suffice to influ- ence their decision-making. In sum, our findings suggest that the role of social norms for individual behavior is likely to change over the course of adolescence. Norm-based interventions can thus lead to mixed effects if not carefully designed for a specific age group. In particular, we observe that younger adolescents are more prone to reduce their climate-friendly behavior in response to the infor- mation that students from another school donated low amounts. Interventions aimed at mak- ing climate-related behavior more salient thus risk backfiring [32, 47], especially among PLOS ONE Social norms and climate-friendly behavior of adolescents PLOS ONE | https://doi.org/10.1371/journal.pone.0266847 April 27, 2022 12 / 16