People are naturally attracted towards beauty in many areas of life. What may be less apparent, however, is that beauty can produce overgeneralizations: when we think something is beautiful, we also tend to believe that it has better qualities.
Chujun Lin and Mark Thornton (2022) recently tested this aesthetic bias with data visualization to see whether we can be fooled by beautiful data. Using hundreds of graphs, they found consistent evidence of a bias. The more beautiful a graph, the stronger the tendency to trust the information in it, even when the graphs were intentionally misleading (like the graph on the right side below). Beauty explained around 15-20% of the variance in trust in the data.
The paper nicely demonstrates how susceptible we are towards the aesthetic bias. In a world that relies heavily on data storytelling and visualizations, it is important that we strengthen data literacy and become more sensitive towards such overgeneralizations.
Link to original study (preprint): https://doi.org/10.31234/osf.io/dnr9s