Social norms and climate-friendly behavior of adolescents

to be higher in the HIGH condition compared to the CONTROL condition (H3.1) and lower in the LOW condition compared to the CONTROL (H3.2) and HIGH (H3.3) conditions. Results In the following section, we examine the impact of the two information conditions, HIGH and LOW, first on the basis of aggregated data and one-sided Mann–Whitney U tests in the direc- tions following our predictions. Panel A in Fig 1 depicts the distribution of the empirical expectations. In the HIGH condi- tion, students guessed that schoolmates of their grade would donate, on average, 60% of their endowment to CO2 offsetting. This donation is similar to what students in the CONTROL condition expected: 56% (z = -1.439, p = 0.075). Thus, we do not find support for H1.1. In the LOW group, in contrast, students believed that students of their grade would contribute, on average, 46% of their endowment; hence, they expected contribution of a significantly lower share than the CONTROL (z = 4.532, p < 0.001) and HIGH (z = -5.752, p < 0.001) groups did. We thus find support for H1.2 and H1.3: students altered their perception about the common behavior in response to the information provided, particularly when they received information that other students of the same age acted in a more self-regarding way. A second element of social norms is the beliefs about what others think one should do: the normative expectations. To assess the social appropriateness of each donation amount, we clustered the evaluations into small (0–2 €), medium (3–5 €), and high (6–8 €) donation

amounts. Panel B shows that high donation amounts were evaluated as very socially appropri- ate in all conditions. For medium donation amounts, the information that peers donated large amounts (HIGH) increased the variance, but did not alter the assessment in a significant man- ner (z = 0.542, p = 0.29). The information that peers donated low amounts (LOW), on the other hand, tended to increase the rated social appropriateness of medium amounts, but this change was also not statistically significant (comparison to CONTROL: z = -1.023, p = 0.153; comparison to HIGH: z = -1.472, p = 0.070). The assessment of small donation amounts, in contrast, was similar between students who received the HIGH information and the CON- TROL group (z = -0.243, p = 0.596). When students instead obtained the information that peers donated low amounts (LOW), small donation amounts were judged as more socially appropriate. However, this observation was not statistically supported by the non-parametric tests (comparison with CONTROL: z = -1.307, p = 0.096; comparison with HIGH: z = -1.126, p = 0.130). Hence, we do not find support for H2.1, H2.2, and H2.3. Lastly, we examined how the treatments influenced the donation decisions (Panel C). When students received the HIGH information, their donation decisions were, on average, similar to those by students in the CONTROL group (HIGH: 63% and CONTROL: 64% of the endowment; z = 0.273, p = 0.608). In the LOW condition, on the other hand, students showed a tendency to decrease their donations in response to the information about peers donating

small amounts (LOW: 59% vs. CONTROL: 64%); however, no statistically significant differ- ence in donation behavior can be detected compared to the CONTROL group (z = 1.365, p = 0.086) or the HIGH condition (z = 1.079, p = 0.140). Thus, we did not find support for H3.1, H3.2, and H3.3. Thus, in our setting, information about peers’ behavior has only a lim- ited impact on the normative expectation and actual behavior on average. This finding is in line with the results of experimental research conducted with adults [44]; learning about the behavior of others does not automatically lead to a change in one’s own behavior. To probe the robustness of our results, we performed multivariate regression analyses controlling for other factors that may have shaped students’ expectations and donation deci- sions, namely, individual characteristics, such as gender, age, migration background, students’ PLOS ONE Social norms and climate-friendly behavior of adolescents PLOS ONE | https://doi.org/10.1371/journal.pone.0266847 April 27, 2022 7 / 16