The Weighted Average Illusion: Biases in Perceived Mean Position in Scatterplots

Matt-Heun Hong, Jessica Witt, Danielle Albers Szafir

External link (DOI)
View presentation:2021-10-28T17:30:00ZGMT-0600Change your timezone on the schedule page
2021-10-28T17:30:00Z
Exemplar figure, described by caption below
We know from prior work that estimation biases can arise in common bar charts and line charts. But what about scatterplots, a seemingly precise technique for presenting multidimensional data? In this presentation we discuss an illusion which arises when presenting more than two data dimensions using scatterplots, which may lead to measurement biases.
Abstract

Scatterplots can encode a third dimension by using additional channels like size or color (e.g. bubble charts). We explore a potential misinterpretation of trivariate scatterplots, which we call the weighted average illusion, where locations of larger and darker points are given more weight toward x- and y-mean estimates. This systematic bias is sensitive to a designer’s choice of size or lightness ranges mapped onto the data. In this paper, we quantify this bias against varying size/lightness ranges and data correlations. We discuss possible explanations for its cause by measuring attention given to individual data points using a vision science technique called the centroid method. Our work illustrates how ensemble processing mechanisms and mental shortcuts can significantly distort visual summaries of data, and can lead to misjudgments like the demonstrated weighted average illusion.