WITHDRAWN - CONCRETENESS AND SIMPLICITY EXPLAIN THE EFFECT OF NUMERICAL AND GRAPHICAL RISK FORMATS ON PERCEIVED LIKELIHOOD AND CHOICE

Purpose: A good understanding of risk information is essential part of informed decision making about health and wellbeing. We investigated to what degree the perception and comprehension of quantitative risk information depends on the features of different numerical and graphical formats used in view of concreteness and their simplicity.  

Method: . One-hundred ninety two students from the Free University of Amsterdam participated in the study. Three different numerical formats were used, i.e. percentages (‘X%’), frequencies with round decimal denominators (‘X in 100/1000’), and frequencies that are standardised to the smallest numerator (‘1 in X’). These formats differ in the degree by which they refer to concrete frequencies and in their simplicity in conveying a numerical ratio.  For the graphical formats, we used bar charts and icon charts. To focus on the quantitative aspect of risk information, we used games of chance. We used an experimental mixed 3x2x2 design with two between factors and one within factor.  Measures for understanding information: perceived likelihood, choice preference, comparisons of different likelihoods. Process measures: response times, subjective evaluation of information. We also assessed participants’ numeracy

Result: The effects of different numerical formats are more pronounced than those of graphical representations. The numerical format ‘1 in X’ is perceived to convey larger likelihoods than the formats ‘X in 100’ and percentages. Likelihoods shown as ‘1 in X’ took less time to interpret and, together with percentages, they were deemed relatively easy to imagine. Adding a graphic and especially a bar chart yielded smaller perceived likelihoods, while the participants found it easier to imagine the likelihood information when an icon chart was added. Notably, the effect of adding a graphic on perceptions was stronger among people with low numeracy.

Conclusion: Perception and interpretation of likelihood information depends on the concreteness and simplicity of the risk formats used. The numerical format ‘1 in X’ may then be favourable as it is concrete and simple. While using percentages is often deemed unfavourable as it would be less concrete than equivalent frequencies, this argument is not supported by our findings. The effect of using of graphics on understanding is likely to be small, which may be larger for people with less numeracy skills.