Four Experiments on the Perception of Bar Charts

Area of Science:

• Data Visualization
• Human-Computer Interaction
• Cognitive Psychology

Background:

• Bar charts are a prevalent data visualization tool.
• Previous research by Cleveland & McGill highlighted performance differences between various bar chart designs.
• The underlying reasons for these perceptual differences remained unexplored.

Purpose of the Study:

• To experimentally investigate and explain the perceptual differences in bar chart reading accuracy.
• To build upon and validate the findings of Cleveland & McGill's foundational study.
• To identify specific sources of error in bar chart interpretation.

Main Methods:

• Conducted a series of controlled experiments to assess human perception of different bar chart configurations.
• Measured accuracy and response times for perceptual tasks involving aligned and stacked bar charts.
• Analyzed participant errors to understand cognitive processes during bar chart reading.

Main Results:

• Confirmed Cleveland & McGill's findings regarding the relative accuracy of different bar chart designs.
• Identified specific factors contributing to lower accuracy on stacked bar charts compared to aligned bar charts.
• Demonstrated that bar proximity significantly influences comparison accuracy.

Conclusions:

• The study provides empirical evidence explaining the perceptual advantages of aligned over stacked bar charts.
• New insights into the cognitive mechanisms underlying bar chart perception were gained.
• Proposed novel hypotheses for future research on bar chart design and interpretation.