Designing for Interaction: Determining the Most Influential Aesthetic Factors for Effective Visualisation of Uncertainty

Visualisations offer a variety of novel ways to depict data to a wider range of users. Moreover, they provide the ability to transform raw data into intuitive visual mechanisms for communication. Despite the developments in visualisation techniques, an area of depiction that has struggled to advance is uncertainty. The visualisation of uncertainty offers an additional dimension for the data by presenting confidence and error rates. Whilst nearly all predictive data sets visualised contain uncertainties, there is still little impulse to actively represent uncertainty. A growing area of interest in the visualisation world is the field of aesthetics; the authors ask the question ‘could aesthetics be applied to address the issues surrounding visualising uncertainty’. This paper reports on the design and delivery of a study to evaluate the effectiveness of aesthetics for the depiction of uncertainty. In particular, the evaluation of how to practically determine how we assess the influence of aesthetic dimensions for the presentation of uncertainty. The paper reports on the strategies employed in this study to assess user’s decision around aesthetic designs, whilst determining what aesthetic combinations elicit the most uncertain visual representation. The findings show that certain aesthetic combinations in a line graph visualisation portrayed a higher level of uncertainty than others and that particular combinations triggered affective responses based on how the visualisation influenced/impacted a participant. In detail, how the textured line characteristics can be displayed aesthetically (combined with either emphasis or scale) to encourage optimal user experiences of uncertainty in a diverse participant group. The paper highlights how a user’s decision on which texture was most uncertain can be overturned when presented with varying levels of emphasis and scale. In summary, this paper contributes to a more in-depth understanding of how to design for and evaluate aesthetic uncertainty visualisations that encourage interaction with the data.