Artistic representations of data can help bridge the US political divide over climate change

The study addresses how artistic representations of climate data compare to conventional scientific data graphs in shaping public emotions, credibility perceptions, learning, and perceived personal relevance of climate change. Motivated by persistent U.S. political polarization over climate policy and evidence that data-centric messaging can trigger motivated reasoning, the authors investigate whether integrating art with data can engage emotions constructively and reduce polarization. They hypothesize that artistic visualizations may evoke stronger emotions without sacrificing credibility or memorability, and they explore whether art, especially when paired with reflective prompts, can lessen partisan differences in perceived relevance of climate change.

Prior work shows growing political divides in the U.S. on climate action. Data graphs can prompt biased processing among skeptics, while calls have been made to infuse communication with emotion and shared experiences. Artistic transformations of climate data (e.g., warming stripes, watercolor-infused graphs) are widely disseminated, assuming they address the imaginative deficit of scientific data. Yet, abstractions risk obscuring information, trivialization, or misinterpretation, and some audiences may be alienated by art. Scientists often view art as beneficial for emotional connection and learning, and climate art can elicit strong emotions, but evidence that it improves knowledge retention or maintains credibility is limited. Political motivations drive selective attention to graphs; avoidance emotions like anxiety can promote openness to opposing views. Reflective, accuracy-oriented primes can reduce biased reasoning and increase meaningful engagement with art. These lines of research motivate testing whether artistic visualizations can reduce polarization and under what conditions (e.g., with reflective prompts).

Design: Two online controlled experiments with U.S. adults recruited from Forthright/Bovitz panels. Total N=671 across two waves (April 2022: N=319; August 2022: N=352). Inclusion: prior Instagram use. IRB: University of Wisconsin-Madison (ID: 2022-0232). Random assignment to a 2 (visual format: artistic representation vs. data graph) × 2 (graph specificity: detailed vs. simplified) between-subjects design. Stimuli were based on Diane Burko’s painting “Summer Heat, 2020” integrating a simplified Keeling curve (Mauna Loa CO2) and corresponding data graphs. In Study 1 (N=319), participants first viewed a stand-alone image (to prompt reflection and measure emotions), then a mock Instagram post embedding the same image with a caption; they answered reflective questions about meaning and emotions. In Study 2 (N=352), participants only saw the mock Instagram posts (no stand-alone image and no reflective prompts). Ecological validity was targeted by using Instagram post mockups. Measures: Emotions (5-point scale, 1=none at all to 5=a great deal) using 12 items previously applied to climate art. Positive emotions index: happiness, awe, inspiration, enthusiasm, hope (α=0.89; mean=1.90, SD=0.97). Negative emotions index: guilt, sadness, anger, anxiety, disappointment, uneasiness, fear (α=0.93; mean=1.97, SD=1.02). Art interest: five 7-point items (α=0.89; mean=4.35, SD=1.36). Information recall: five multiple-choice items about the image/caption, summed correct (KR-20=0.63; mean=2.89, SD=1.56). Perceived credibility of the post: six 7-point items (trustworthy, reputable source, accurate, lack of incorrect info [reverse], unbiased, objective; α=0.85; mean=4.70, SD=1.08). Perceived relevance of climate change: three 7-point items (relevance to daily life, personal effects, thinking about one’s role; α=0.87; mean=4.77, SD=1.47). Political leaning: standardized composite of party identification (1=Strong Democrat to 7=Strong Republican) and ideology (economic and social; 1=very liberal to 7=very conservative). Covariates and group equivalence checks included demographics (age, gender, education, race, income), preexisting climate concern (α=0.94), science/art education, social media use frequency, and Instagram art activity (α=0.90). Analyses: ANOVA and ANCOVA tested main and interaction effects (art vs. graph; detailed vs. simplified; interactions with art interest and political leaning). Study 1 tested reflective prompting; Study 2 replicated without reflective primes to assess moderation by reflection. Group homogeneity across conditions was verified by ANOVAs on demographics and relevant predispositions.

• Emotions: Artistic representations elicited stronger positive emotions than data graphs (main effect of art: F(1,316)=8.16, p=0.005, η²=0.025). Detailed graphs (vs. simplified) elicited stronger negative emotions (main effect of detail: F(1,316)=13.21, p=0.000, η²=0.04). The original art piece generated the highest positive emotions (mean=2.17, SD=1.06). The edited data graph (no labels/title/scale) evoked the lowest positive (mean=1.54, SD=0.82) and negative emotions (mean=1.59, SD=0.77). - Moderation by art interest: Interaction between visual format and art interest was significant for negative emotions (F(1,314)=7.14, p=0.008, η²=0.022): among low art-interest individuals, data graphs evoked stronger negative emotions; among high art-interest individuals, artistic visualizations evoked stronger negative emotions. - Information recall: Main effect of art on recall was marginally significant across combined samples (F(1,668)=4.26, p=0.039); effects were small. - Perceived credibility: No significant difference between posts with artistic visualizations vs. data graphs (F(1,668)=0.39, p=0.535). - Political polarization in perceived relevance: In Study 1 with reflective primes, interaction between political leaning and visual format was significant (F(1,314)=8.14, p=0.005, η²=0.025): the association between conservatism and lower perceived relevance was stronger after viewing data graphs than after viewing artistic visualizations, indicating reduced polarization with art. In Study 2 without reflective primes, this interaction was not significant (F(1,347)=0.10, p=0.76), suggesting the depolarizing effect of art depends on reflection prompting.

The findings indicate that artistic representations can enrich data communication by evoking emotions without compromising credibility or recall. Art elicited stronger positive emotions than data graphs, potentially supporting engagement and meaning-making. Importantly, when individuals were prompted to reflect on the meaning and feelings evoked by the visuals, artistic representations attenuated the typical partisan divergence in perceived relevance of climate change observed with data graphs, addressing the central goal of reducing polarization. Audience characteristics matter: those with higher interest in visual art experienced stronger negative emotions from art, which may facilitate deeper interpretation. Despite concerns that abstract or artistic formats could reduce trust or learning, participants viewed posts as similarly credible and recalled information comparably across formats. These results support the strategic use of artistic visualizations, particularly alongside reflective engagement, to broaden the perceived personal relevance of climate change across political lines and to expand interdisciplinary collaboration among science, communication, and the arts.

This research provides experimental evidence that artistic representations of climate data can enhance emotional engagement and, when paired with reflective prompts, reduce political polarization in perceived relevance of climate change compared to traditional data graphs, without diminishing credibility or information recall. The work underscores the promise of integrating art into science communication to connect with diverse audiences. Future research should: (1) test a broader array of artworks and artists, isolating visual format from other content features; (2) identify specific artistic characteristics that influence emotions, credibility, and learning; (3) examine effects across cultural contexts, including Global South settings; (4) explore hopeful visual discourses and participatory, community co-creation approaches; and (5) assess impacts in real-world digital environments (e.g., Instagram, TikTok) and with unconcerned audiences.

• Stimulus specificity: Results are based on a single artwork by one American artist; findings may not generalize to other artworks, artists, or presentation contexts (e.g., in-person installations vs. online viewing). - Confounding visual elements: The art contained additional elements (e.g., map of Europe, glaciers) absent from data graphs, so effects may not be due solely to artistic format. - Sample and context: Non-probability U.S. samples of Instagram users limit generalizability; cultural and platform-specific factors may influence outcomes. - Reflection dependence: The depolarizing effect appeared only when participants engaged in prompted reflection; effects may be weaker without such engagement. - Measurement scope: Recall reliability was moderate (KR-20=0.63); emotional and relevance measures, while reliable, capture short-term reactions in an experimental setting.