The Significant Effect of Visual Representations on Dichotomous Thinking

L'effet significatif des Visualisations sur le Raisonnement Dichotomique

Jouni Helske (1) , Matthew Cooper (1) , Anders Ynnerman (1) , Lonni Besançon (1)

1 LIU - Linköping University

Common reporting styles of statistical results for controlled experiments (confidence intervals and p-values) have been shown to lead to dichotomous inferences on the study outcome. We investigated in two controlled experiments whether several different visual representations could help limit such dichotomous interpretations. The results of our experiments suggest that some visual representations can help decrease binary interpretations of statistical results compared to textual representations of p-values and confidence intervals. All data analysis and scripts are available online.

Domaines: Interface homme-machine [cs.HC]

Jouni Helske, Matthew Cooper, Anders Ynnerman, Lonni Besançon. L'effet significatif des Visualisations sur le Raisonnement Dichotomique. Journée Visu 2019, May 2019, Paris, France. ⟨hal-02130437⟩