Perceptual Biases in Font Size as a Data Encoding.

Area of Science:

• Information Visualization
• Human-Computer Interaction
• Perceptual Psychology

Background:

• Data is frequently encoded using font sizes in visualizations like word clouds and cartographic labels.
• Font size as a visual encoding can be intuitive but may introduce perceptual biases.
• Factors irrelevant to the encoded data, such as word length or character height, might influence judgments.

Purpose of the Study:

• To empirically investigate the extent to which font size, word length, and character height influence comparative judgments in data visualizations.
• To identify potential biases in font size perception that are unrelated to the actual data values.
• To propose mitigation strategies for perceived biases in font-size-encoded data.

Main Methods:

• A series of empirical studies were conducted to test participants' comparative judgments of font sizes.
• Participants were presented with various visualizations where font size encoded data, with controlled variations in word length and character height.
• Statistical analyses were used to quantify the influence of irrelevant factors on perceived font size.

Main Results:

• Irrelevant factors like word length and character height do influence comparative judgments of font size.
• The degree of influence from these factors was found to be less significant than commonly assumed.
• Specific character shapes and word lengths were identified as having the largest potential to bias perception.

Conclusions:

• While font size is a common encoding in visualizations, its use requires careful consideration of potential perceptual biases.
• The impact of confounding factors like word length is present but may be overestimated.
• Strategies exist to mitigate these biases, ensuring more accurate data interpretation from font-size-encoded visualizations.