Visual cues in estimation of part-to-whole comparison

TL;DR

This study compares the effectiveness of pie charts and horizontal bar charts with different visual cues in estimating segments in part-to-whole comparisons, providing practical recommendations for dashboard design.

Contribution

It presents experimental evidence on how visual cues influence estimation accuracy in pie and bar charts, informing better visualization practices.

Findings

Pie charts with certain cues perform comparably to bar charts in estimation tasks.

Visual cues significantly affect the accuracy of segment estimation.

Recommendations for visualization design in dashboards are provided.

Abstract

Pie charts were first published in 1801 by William Playfair and have caused some controversy since. Despite the suggestions of many experts against their use, several empirical studies have shown that pie charts are at least as good as alternatives. From Brinton to Few on one side and Eells to Kosara on the other, there appears to have been a hundred-year war waged on the humble pie. In this paper a set of experiments are reported that compare the performance of pie charts and horizontal bar charts with various visual cues. Amazon's Mechanical Turk service was employed to perform the tasks of estimating segments in various part-to-whole charts. The results lead to recommendations for data visualization professionals in developing dashboards.