Biased Average Position Estimates in Line and Bar Graphs: Underestimation, Overestimation, and Perceptual Pull

Area of Science:

• Data Visualization
• Human-Computer Interaction
• Cognitive Psychology

Background:

• Position is generally considered the most accurate visual encoding for data representation.
• Other encodings like hue can be less precise and prone to systematic biases.

Purpose of the Study:

• To investigate potential systematic biases in data encoding using position.
• To determine if position, despite its precision, is susceptible to perceptual distortions.

Main Methods:

• Three empirical studies were conducted using visual data displays.
• Participants reported average positions of lines and bars after a short delay.
• Experiments included single and multiple data series (lines and bars).

Main Results:

• Position encoding, while precise, demonstrated systematic biases.
• Line positions were consistently underestimated.
• Bar positions were consistently overestimated, even in multi-series displays.
• A 'perceptual pull' effect was observed where estimates shifted towards other series.

Conclusions:

• Position encoding, though precise, is not immune to systematic perceptual biases.
• Underestimation of line positions and overestimation of bar positions are significant findings.
• The 'perceptual pull' effect highlights complex interactions in multi-series visualizations.