An extended logarithmic visualization improves forecasting accuracy for exponentially growing numbers, but residual difficulties remain
Year: 2024
Type: article
Abstract: Humans find it notoriously difficult to predict the future development of numbers in scenarios where the data exhibits exponential growth. This study explored how employing logarithmically scaled graphs can improve forecasting accuracy in such scenarios. Experiment 1 shows that a modified visualization improves forecasting, mitigating the inaccuracies encountered with linear and ordinary logarithmic depictions. The modification consists of putting the y-axis on the right side of a logarithmically scaled graph and extending the x-axis to the range of the forecast period. This effect was independent of general graph literacy, and participants were more confident in their estimates. To uncover the role of tick marks in estimation accuracy, we conducted a second experiment manipulating the presence of minor tick marks and varying target values systematically with respect to their proximity to the next major tick mark. Participants performed worse for target values midway between two major tick marks and no accuracy benefits related to the presence of tick marks. Analysis of eye movements during the same task suggests that the poor utilization of minor tick marks is not simply due to a lack of attention but to difficulties in converting the location into the corresponding numerical value.
Authors: Ben H. Engler, Florian Hutzler, Stefan Hawelka
Institution: University of Salzburg
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $1,500
Funder: Austrian Science Fund
OpenAlex ID: W4403667930