Abstract
Graphical representations have to be true to the data they display. Computational tools ensure this on a technical level. But we also need to take “flaws” of the human perceptual system into account. The sine illusion provides an example where human perception leads to systematic bias in the assessment of the optical stimulus, with a particularly notable impact on perception of time-series data with a seasonal component. In this article, we discuss the reasons for the illusion and various strategies useful to break the illusion or reduce its strength. We demonstrate the presence of the illusion in real-world and theoretical situations. We also present data from a user study, which demonstrate the dramatic effect the sine illusion can have on conclusions drawn from displayed data.
Citation
[1] S. Vanderplas and H. Hofmann. “Signs of the Sine Illusion - why we need to care”. In: Journal of Computational and Graphical Statistics 24.4 (Dec. 10, 2015), pp. 1170-1190. DOI: https://doi.org/10.1080/10618600.2014.951547.
@article{sineillusionjcgs,
title = {Signs of the Sine Illusion - why we need to care},
author = {Susan Vanderplas and Heike Hofmann},
journal = {Journal of Computational and Graphical Statistics},
publisher = {Taylor & Francis},
volume = {24},
number = {4},
pages = {1170-1190},
year = {2015},
doi = {https://doi.org/10.1080/10618600.2014.951547},
month = {dec},
}