Visual art inspired by climate change—An analysis of audience reactions to 37 artworks presented during 21st UN climate summit in Paris

3. In the third step, the research focus was on the characteristics of the artworks. The question is if the variables identified as relevant in the first step on the person level differed between artworks. To achieve this, the model tested in step 1 was reconfigured as a full two-level model. All dependent variables in the layers "climate change-related cognitions" and "art experience" were tested for significant variance on the artwork level and all variables except the support for climate policies had significant variation between artworks. The artwork- level variance was then regressed on the artwork features. Since a large selection of charac- teristics was recorded, this was done in a stepwise procedure, where each level 2 variable was first regressed on all artwork features and then, one-by-one, features with the highest p-level were removed until only significant influences remained. We are aware that this is a highly exploratory strategy, very vulnerable for capitalizing on chance. The results of this analysis should thus be treated as a first indication only. The results of this analysis are reported in Fig 4. Results In this section, we first report the final structural model of the relations between the variables and behavior, then the test of the interaction hypothesis, and finally, the test of a full two-level model with influences of artwork characteristics. How does an art experience relate to support for climate policy? Fig 2 displays the results of the model test. All relations are confirmed, and the model receives

a good model fit (see numbers in the figure). The displayed estimates on the arrows are stan- dardized regression weights. According to the tested model, the first step in the process is the emotional activation that is related to experiencing an artwork. Both positive and negative emotional reactions then relate significantly to the mediating cognitive variables perceived quality and reflections on the art- work. The relation of negative emotions with reflections on the artwork seems to be stronger than the relation of positive emotions, whereas the perceived quality of the artwork is slightly stronger related to positive emotions. The more negative and positive emotions are elicited, the higher is the perceived quality, and the stronger are the reflections on the artwork. The Fig 4. Artwork level part of the multilevel model with variance in emotional responses, identification with the artist, perceived quality of the artwork, and reflection on the artwork modeled on the artwork level (Nlevel1 = 714, Nlevel2 = 30). https://doi.org/10.1371/journal.pone.0247331.g004 PLOS ONE Visual art inspired by climate change — An analysis of audience reactions PLOS ONE | https://doi.org/10.1371/journal.pone.0247331 February 19, 2021 10 / 18