Studying attention to IPCC climate change maps with mobile eye-tracking

easier to process. This difference could also be influenced by methodological factors: fixation detection algorithms might perform differently on screen-based stimuli versus in situ observa- tion, with the latter possibly affected by participant mobility during painting viewing. Lastly, in both settings, fixation counts were higher in paired viewing conditions, indicating a consis- tent trend across this metric. Interestingly, while dwell times were comparable between single and paired map viewing conditions, they differed significantly in painting viewing, suggesting a minor preferential difference between the two types of stimuli. 3.3 Climate change anxiety scale (CCAS) responses Irrespective of the viewing condition, whether single or paired, the analysis of the climate change anxiety scale involved four subscales of the 22-item questionnaire, with responses gath- ered using a 5-point Likert scale. For a visual representation of the responses, refer to S8 Fig for the frequency plot of individual items, S9 Fig for the aggregated frequency plot across the four subscales, and S9 Table for a statistical breakdown. Overall, levels of climate change anxiety were relatively low across the sample. For the first two, often highly correlated subscales (cognitive-emotional impairment and functional impairment), participants typically reported low scores. More than half of the responses were “never,” and approximately a quarter were “rarely.” However, a small subset of participants exhibited mid or high scores on these measures, suggesting notable levels of climate-related

anxiety for a minority of participants within the group. Furthermore, responses on the personal experience with climate change subscale, and more distinctly on the behavioural engagement subscale, were comparatively higher. While the dis- tribution of responses on the personal experience subscale was relatively even across all five points of the Likert scale, the behavioural engagement subscale showed a negatively skewed distribution. This suggests that on average, participants either exhibited or aspired to positive behaviours towards addressing the climate crisis. When investigating the relationships between the four subscales of the climate change anxi- ety scale (CCAS), the 5-point Likert scale responses were treated as ordinal data, assigning val- ues from 0 (never) to 4 (almost always). Despite the frequency distributions varying across the subscales, their relationships were examined through cross-correlation, using Spearman’s rho and Kendall’s tau-b. These analyses showed positive linear correlations between the subscales, as illustrated in S10 Fig, and detailed in S10 Table. This suggests that a combined CCAS score could be formed as a cohesive construct for further analysis. To explore the relationship between this aggregated CCAS score (treated as ordinal data) and the four primary gaze metrics (treated as continuous data), we conducted correlation anal- yses. The results, however, indicated no significant correlations; all relationships were effec- tively flatlined across the metrics. This lack of significant findings implies that there is no

immediate or obvious connection between main gaze metrics and self-reported climate anxi- ety, as detailed in S11 Table. Given these outcomes, we did not proceed further, such as divid- ing participants into low and high anxiety groups based on median, quartile, or range-based splits. 4. Discussion and conclusions The present study established the relevance of using mobile eye-tracking to examine how peo- ple look at climate change maps, and provided insights for scaling it up to more naturalistic settings, notably social ones. Despite individual differences, on average, the comparison between conditions during map viewing suggests that social interaction can subtly alter gaze patterns. Paired viewing was associated with higher fixation counts, shorter average fixation PLOS ONE Studying attention to IPCC climate change maps with mobile eye-tracking PLOS ONE | https://doi.org/10.1371/journal.pone.0316909 January 10, 2025 11 / 17