Method: N=232 German and Spanish students participated in two experiments, each with several tasks involving different treatments and sizes of risk reductions. In the first experiment, we manipulated two factors: 1) whether risk reductions were presented numerically in relative or in absolute terms; and 2) whether icon arrays were added to the numerical information. We measured 1) accuracy of understanding of risk reduction (c.g. Schwartz, Woloshin, Black, & Welch, 1997), 2) perceived helpfulness of treatment and intention to use it, 3) perceived helpfulness and credibility of information. In the second experiment, we manipulated 1) total number of icons used in arrays, and 2) position of icons symbolizing sick persons (at the beginning or at the end of an array). We measured 1) perceived seriousness of baseline risks, and 2) perceived helpfulness of treatment.

Results: Experiment 1. Accuracy of understanding of risk reduction was better when numerical information was in the form of absolute rather than relative risk reductions (average odds ratio of correct answers was 10.4 across three tasks). In addition, icon arrays improved accuracy of understanding of both numerical forms (average odds ratio was 4.5). Visual representation increased perceived helpfulness of treatment and willingness to use it (average Cohen's d across different tasks was .23 and .35, respectively), but this increase was not as large as that resulting from numerical presentation in relative instead of in absolute terms (d=.31 and .41). The information on risk reductions was perceived to be more helpful (average d=.32) and somewhat more credible (average d=.20) when icon arrays were added to numerical information. Experiment 2. Both perceived seriousness of baseline risk and perceived helpfulness of treatment were higher when larger total number of icons was used (average d=.27 for baseline risk and d=.43 for risk reduction). Position of icons symbolizing sick persons had inconsistent effects on those perceptions.

Conclusions: Icon arrays are helpful in communicating risk reductions because they increase accuracy of understanding and perceived helpfulness of information. However, it is important to understand that some of their properties, such as overall number of icons used, can affect risk perceptions independently of the information about risk reductions.