Social norms and climate-friendly behavior of adolescents

survey packages laid out on the chairs in the schools’ assembly halls. The questionnaire ver- sions differed only slightly on one page, and with students sitting far apart from each other (see Fig D.1–D.3 in S1 File) and being supervised by eight members of the research team, we can say with confidence that the students were not aware of the different versions during the study. The study was preregistered prior to data collection https://aspredicted.org/jz76g.pdf (Please note that the numbering of the hypotheses has changed). The dataset, analysis scripts, and original instructions (in German) can be accessed here: http://doi.org/10.17605/OSF.IO/ Z6G54. An English translation of the instructions can be found in the Section D.2 in S1 File. Hypotheses We expect that providing information about how much relevant others (i.e., students of the same age, school type, and city) chose to donate will alter students’ perceptions about the pre- vailing social norm and, consequently, behavior. We assessed these changes with the help of the two incentivized belief elicitations and the costly behavioral measure. In belief elicitation 1, we targeted the empirical expectation and asked the students to guess how much their schoolmates of the same grade would donate on average. We hypothe- sized that this empirical expectation would be higher in HIGH, whose students received infor- mation about relevant others donating large amounts, in comparison to CONTROL (H1.1), in which no information about others’ donations was given. We expect the empirical expectation

to be lower in the LOW condition compared to CONTROL (H1.2) and HIGH (H1.3). With belief elicitation 2, we assessed the normative expectations and asked the students, following the mechanism of Krupka and Weber [57], to guess how socially appropriate their schoolmates of the same grade would perceive each possible donation amount. For the analy- sis, we will cluster the evaluations into small (0–2 €), medium (3–5 €), and high (6–8 €) dona- tion amounts and will focus only on the evaluation of small and medium amounts, since we expect that high donation amounts will always be evaluated as socially desirable. (For the indi- vidual donations amounts in CONTROL, we observed that donations of 50% or more were judged on average as socially appropriate.) We hypothesize that students will rate small and medium donation amounts lower on the social appropriateness scale in the HIGH condition than in CONTROL (H2.1). On the other hand, we hypothesize that the perceived social appropriateness of the same amounts will be higher in the LOW condition than in the CONTROL (H2.2) and HIGH (H2.3) conditions because the information that relevant others donated only little makes donating small amounts more socially acceptable. Consequently, we assume that the information about the behavior of others also changes one’s own decision-making. Therefore, for observable behavior, we expect average donations Table 1. Treatment overview. Treatment Description Observations Total Grade 8 Grade 9 Grade 10 CONTROL No information 164 51 53 60

LOW Information on relevant others donating low amounts 159 49 50 60 HIGH Information on relevant others donating high amounts 168 53 52 63 HYPO-LOW Information on hypothetical low donation amounts 70 22 21 27 HYPO-HIGH Information on hypothetical high donation amounts 67 22 19 26 https://doi.org/10.1371/journal.pone.0266847.t001 PLOS ONE Social norms and climate-friendly behavior of adolescents PLOS ONE | https://doi.org/10.1371/journal.pone.0266847 April 27, 2022 6 / 16