G. Dzemydaitė,  G. Melnik-Leroy, L. Aidokas, G. Dzemyda, V. Marcinkevičius, D. Melnikienė, A. Usovaitė

G. Dzemydaitė, G. Melnik-Leroy, L. Aidokas, G. Dzemyda, V. Marcinkevičius, D. Melnikienė, A. Usovaitė

Is my visualization better than yours? Analyzing factors modulating exponential growth bias in graphs

Is my visualization better than yours? Analyzing factors modulating exponential growth bias in graphs 

Gerda Ana Melnik-Leroy1, Giedrė Dzemydaitė2* , Linas Aidokas1, Gintautas Dzemyda1, Virginijus Marcinkevičius1, Danguolė Melnikienė1,3, Ana Usovaitė1,4 

*  

1Institute of Data Science and Digital Technologies, Vilnius University, Lithuania 

2Faculty of Economics and Business Administration, Vilnius University, Lithuania 

3Faculty of Philology, Vilnius University, Lithuania 

4Department of Graphical Systems, Vilnius Gediminas Technical University, Lithuania 

 

Humans tend to systematically underestimate exponential growth and perceive it in linear terms, which can have severe consequences in a variety of fields. Recent studies have attempted to examine the origins of this bias and to mitigate it by comparing the use of the logarithmic vs. linear scales in graphical representations. However, these studies have yielded conflicting results regarding which scale induces more perceptual errors.

In the current study, through an experiment with a short educational intervention, we further examined the factors modulating the exponential bias in graphs and suggested a theoretical explanation for our findings. Specifically, we tested the hypothesis that each scale can induce misperceptions in a particular context. In addition, we explored the effect of mathematical education by testing two groups of participants: those with a background in humanities versus in formal sciences.

The results of this study confirm that, when used in an inadequate context, both logarithmic and linear scales can dramatically affect the interpretation of visualizations representing exponential growth. In particular, while the log scale leads to more errors in graph description tasks, the linear scale misleads individuals when predicting the future trajectory of exponential growth. The second part of the study revealed that the difficulties with both scales can be reduced through a short educational intervention. Importantly, while no difference between participant groups was observed before the intervention, participants with a better mathematical education showed a stronger learning effect in the posttest. The findings of this study are discussed in light of a dual-process model. 

 

Keywords: cognitive bias; exponential growth; graph perception; logarithmic scaling; mathematical literacy 

 

Biography  

 

Giedrė Dzemydaitė, Assoc. Professor, Faculty of Economics and Business Administration, Vilnius University, Lithuania. Giedrė Dzemydaitė earned her PhD in economics in 2016. She currently holds the position of associate professor and senior researcher at Vilnius University in Lithuania. Her research interests include regional development, production efficiency, structural change and innovation. Giedrė collaborates on interdisciplinary studies, particularly with digital technologies, data scientists, mathematicians, and psychologists. She has experience working on projects related to economic analysis, policy formation, and interdisciplinary teamwork. Additionally, she is involved in international Erasmus+ capacity-building projects with partners from Europe and Asia. Giedrė is a member of the European Regional Science Association and serves on the board of the Baltic Section.