Data Visualization

Data visualization is the use of charts, graphs, and other visual tools to represent data. While visualizations can clarify and illuminate, they can also mislead — and the design choices that make them misleading are often not obvious. Critical thinkers must apply the same scrutiny to visual claims that they apply to verbal ones.

How It Appears Per Course

PHIL 252

Covered in Unit 6 as an extension of the broader discussion of representation and ambiguity. Visual bullshit exploits the same mechanisms as verbal bullshit — creating false impressions while technically remaining accurate.

Taxonomy of Misleading Visualizations

Problem	Description	Example
Duck	Visualization prioritizes aesthetics over accurate communication of data — “cute” but obscures the data	3D pie charts, pictograms with variable symbol sizes
Glass Slipper	Wrong chart type forced onto the data — “shoehorning.” Creates false sense of rigor by trading on authority of legitimate visualizations	”Periodic table of marketing” — looks scientific but categories are arbitrary
Axis Manipulation	Axes truncated, inverted, or use variable scale intervals to exaggerate or conceal differences	Y-axis starting at 97% instead of 0% makes small changes look huge
Bin Width Manipulation	Variable bin widths in histograms shift what appears large or small	Widening the “middle class” income bin makes it look disproportionately large
Proportional Ink Violation	Shaded area, length, or volume doesn’t correspond to actual values	Pie slices not proportional to their percentages
Right-Censoring	Omitting cases that haven’t yet reached the study endpoint, creating a false impression of outcomes	Removing still-alive patients from a survival study
Selection Bias	The sample visualized isn’t representative of the population the graph claims to describe	Surveying only high earners for an income distribution chart

Principle of Proportional Ink

“When a shaded region is used to represent a numerical value, the area of that region should be directly proportional to the corresponding value.”

Violations are everywhere: bar charts that don’t start at zero, 3D charts where perspective distorts relative sizes, pie charts with hand-drawn wedges.

Critical Questioning Protocol for Any Chart

1. Check the axes — Does the Y-axis start at zero? Are scales consistent?
2. Examine bins — In a histogram, are bin widths equal?
3. Assess form fit — Is the chart type appropriate for this data?
4. Check proportional ink — Does area correspond to value?
5. Investigate the sample — Who was counted? Could selection bias be present?
6. Check context and denominators — Are rates per capita? Are sample sizes reported?

Cross-Course Connections

Bullshit — new-school bullshit specifically uses data visuals to mislead
Bias — selection bias is a common source of misleading visualizations
FallaciesOfAmbiguity — visual ambiguity parallels verbal ambiguity
CriticalThinking — the same skeptical questions apply to charts as to verbal claims

Key Points for Exam/Study

• “Duck” = aesthetics over function; “Glass Slipper” = wrong format for the data
• Proportional ink principle: area = value (always check this)
• Right-censoring: removing ongoing cases artificially improves outcomes
• The same three questions work for visuals: Who made it? How do they know? What’s in it for them?
• Always ask what the chart would look like redrawn with the axis starting at zero, or with equal bins

Open Questions

• At what point does a design choice become dishonest vs. simply a legitimate presentational choice? Is there always a “most accurate” visual for any dataset?

Cross-course: DataVisualization-FinancialRatios — misleading financial ratio charts and annual report graphics as ADMN 201 application of PHIL 252 visualization critique