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

total fixation duration (also referred to as dwell time), fixation count, average fixation dura- tion, and proxy saccadic scanpath length. Instances of transient engagement, such as brief glances at fiducial markers, were excluded from the analysis due to their minimal duration. As previously described in the Data Analysis section, the pre-processed, enriched data were obtained using the reference image mapper (RIM) technique. 3.1 Descriptive statistics for maps The main descriptive statistics for the maps reflect averages across all stimuli, without differen- tiation by map type. Supplementary statistics were broken down either by map projection timeframe or projection type. Sample sizes were 35 for the single viewing condition and 12 for the paired viewing condition, with each of the 10 stimuli viewed under both conditions. This resulted in a total of 350 observations for the single viewing and 120 for the paired viewing conditions. Initial visualisation using cumulative heatmaps revealed a strong upper-central tendency in viewing patterns, indicating that regions on and around continental Europe received signifi- cant attention across most cases. These heatmaps also illustrated comparable spatial patterns between single and paired viewing conditions. Notably, a considerable number of fixations were concentrated on the scales and on areas of the maps displaying the minima and maxima values of the corresponding scale. These areas often represent the most salient regions in terms of contrast and colour, suggesting that such bottom-up factors are major determinants of spa-

tial attention location among participants. For an illustrative overview of these patterns, refer to Fig 1, which displays the heatmaps of the maps. In comparing single and paired viewing conditions (NSampleSingle = 35, NStimulusSingle = 350 and (NSamplePaired = 12, NStimulusPaired = 120), the total fixation duration, or dwell time, aver- aged around 30 seconds per map, showing some variations within each condition but relatively minimal variation between conditions: MSingle = 29.70 s (±.25), MPaired = 32.92 s (±.94). This represented approximately 10% of the total dwell time across each map for both conditions. The number of fixations, or fixation count, was relatively lower in the single viewing condition compared to the paired viewing condition: MSingle = 55.12 (±.87), MPaired = 70.70 (±2.47). In line with this difference in the fixation count, and as a derivative metric to the previous two, the average fixation duration was longer in the single viewing condition compared to the paired viewing condition: MSingle = 586.43 ms (±11.29), MPaired = 504.45 ms (±15.29). The proxy scanpath length, measured as the cumulative sum of Euclidean distances between suc- cessive fixations based on reference images with 1920 × 1080 pixel dimensions, was shorter in the single viewing condition compared to the paired viewing condition: MSingle = 19534.89 px (±365.87), MPaired = 26288.61 px ±1036.95), refer to Fig 2 for an overview as box plots and S1 Table for aggregate metrics. Despite the relatively low sample size and exploratory nature of the research, a nonparamet-

ric ANOVA (Kruskal-Wallis test) was used to analyse these metrics. While fixation duration showed no significant differences, fixation count, average fixation duration, and scanpath length displayed significant differences between single and paired viewing conditions (see S2 Table for detailed breakdown of χ2, df, p, ε2 values). Additionally, descriptive statistics were reported based on two categorisations: map projec- tion timeframe (near and far future) and map projection type (main temperature, sea surface temperature, sea level rise, anthropogenic CO2 emissions, and fine particulate matter PM2.5). Metrics were similar between the two timeframes but showed some variation across the five types of projections. The overall variance was generally larger, especially for the maxima, in paired condition for all metrics except average fixation duration. These results are presented in 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 6 / 17