“…Theoretically, it has been argued that icon arrays are beneficial because they communicate in terms of relative frequencies rather than single-event probabilities (Tubau et al, 2019), they explicitly present information about how often the negative outcome is expected to occur and not to occur , they are more easily and automatically processed (Ancker et al, 2006;Trevena et al, 2013), and they result in deeper and more meaningful gist represenetations as opposed to surface-level verbatim representations (Brust-Renck et al, 2013). Empirically, icon arrays have been shown to help people solve Bayesian inference problems (Böcherer-Linder & Eichler, 2019;Brase, 2014;Tubau et al, 2019), improve their comprehension of relative and absolute risk-reduction statistics (Galesic et al, 2009;, reduce their susceptibility to gain-loss framing effects , and change their beliefs, emotions, and behavioral intentions in potentially adaptive ways (e.g., Nguyen et al, 2019;Walker et al, 2020). It is important to note, however, that not all studies have shown advantages of icon arrays over numerical representations (Etnel et al, 2020;Ruiz et al, 2013;Wright et al, 2009).…”