DAVI Speed Critiquing of Visualizations TA

DAVI Lecture Notes: Speed Critiquing of Visualizations

Introduction

• The session focuses on speed critiquing in visualization.
• Goal: Develop the ability to quickly identify major issues in visualizations.
• Importance: Saves time during exams and enhances critical analysis skills.

Key Objectives

• Quick Recognition: Spot errors instantly, similar to noticing a simple math mistake (e.g., 1 + 3 ≠ 5) or a misspelled word (e.g., “nember” instead of “number”).
• Efficient Critiquing: Avoid overanalyzing; identify the most significant issues immediately.
• Exam Preparation: Improve performance by swiftly pinpointing problems under time constraints.

---

Example 1: Misuse of Version Numbers in a Bar Chart

Visualization Description

• A bar chart displays software version numbers as quantitative data.
• Versions like 4.5 and 5.0 are plotted as if they represent measurable quantities.

Key Issue

• Version Numbers Are Not Quantities:
  • Version numbers are identifiers, not measurable data.
  • They do not represent quantities that can be meaningfully compared using bar lengths.

Explanation

• Inappropriate Chart Type:
  • Using a bar chart suggests a quantitative comparison.
  • Version numbers (e.g., Windows 95, Windows 98, Windows 2000) do not follow a numerical progression that reflects magnitude.
• Misleading Interpretation:
  • Plotting version numbers as quantities can imply false relationships (e.g., that a higher version number is proportionally better).

Additional Point

• 3D Effects as a Distraction:
  • The use of 3D distorts perception and adds unnecessary complexity.
  • However, the primary issue remains the misuse of data types.

---

Example 2: Glass Slipper Visualization

Visualization Description

• A complex 3D pyramid titled “Pyramid of Sentiment-Oriented Hierarchical Production Planning.”
• Data is forced into a pyramid shape without clear justification.

Key Issue

• Glass Slipper Effect:
  • Data is molded to fit a familiar but inappropriate visual metaphor.
  • The shape (pyramid) does not enhance understanding and may confuse the viewer.

Indicators of a Glass Slipper

1. Title Explicitly Describes the Shape:

   • Titles like “Pyramid of…”, “Metro Map of…”, or “Periodic Table of…” signal forced metaphors.

2. Data Misalignment:

   • The chosen shape does not naturally represent the data structure or relationships.

Explanation

• Complexity Without Clarity:
  • The visualization prioritizes aesthetic appeal over functional design.
• Ineffective Communication:
  • The forced shape hinders comprehension and does not facilitate meaningful insights.

---

Example 3: Call Center Satisfaction Plot

Visualization Description

• A scatter plot with bubbles representing incoming calls to a call center.
• Both axes represent time intervals, but with inconsistent durations (e.g., 1-hour vs. 2-hour intervals).
• Bubble size indicates call length; color indicates customer satisfaction.

Key Issue

• Inconsistent Time Intervals on Axes:
  • Time intervals on the axes are irregular, leading to misleading slopes.
  • Diagonals remain straight despite varying time intervals, which is mathematically incorrect.

Explanation

• Misrepresentation of Data:
  • Slopes should change with varying time intervals; the visualization fails to reflect this.
• Confusion in Interpretation:
  • Inconsistent labeling makes it difficult to accurately read and analyze the data.

Additional Points

• Overplotting:
  • While overplotting is present due to data density, it’s a secondary concern.
• Lack of Clarity on Bubble Size:
  • The visualization does not clearly explain what bubble sizes represent, adding to confusion.

---

Developing an Eye for Visualization Critiquing

Key Indicators and Signs

1. Bar Charts:
   • Cut-off Axes: Check for axes that do not start at zero, which can exaggerate differences.
   • Quantitative Data: Ensure bars represent true quantities.
2. Histograms:
   • Equal Bin Sizes: Bins should be consistent to accurately reflect data distribution.
   • No Spaces Between Bars: Unlike bar charts, histograms should have adjacent bars.
3. Data Mapping:
   • Identity Channels for Categorical Data: Use color or shape for categories.
   • Magnitude Channels for Quantitative Data: Use size or position for quantities.
4. Pie Charts and Similar Visualizations:
   • Percentages Must Sum to 100%: Ensure all segments collectively represent the whole.
   • Appropriate Use: Only use when displaying parts of a whole.
5. Stacked Charts:
   • Parts-to-Whole Relationship: Stacked elements should contribute to a total.
6. Normalization:
   • Adjust for Population or Size: Normalize data to allow fair comparisons (e.g., per capita metrics).
7. Line Charts:
   • Avoid Categorical Data: Line charts imply continuity; use them for time series or ordered data.
8. Lie Factors:
   • 3D Distortions: 3D effects can mislead by altering perception of size and distance.
   • Area vs. Radius: In bubble charts, map data to area, not radius, to maintain proportionality.
9. Color Scales:
   • Equidistant Color Steps: Ensure perceptual uniformity across the scale.
   • Increasing Luminance: Helps in grayscale conversion and avoids visual artifacts.
10. Temporal Data:
    • Chronological Order: Data should be plotted in time sequence.
    • Consistent Time Intervals: Irregular intervals can distort trends and patterns.
11. Numbers vs. Quantities:
    • Numeric Labels Aren’t Quantities: Version numbers, dates, and categorical codes should not be treated as measurable data.
12. 3D Visualizations:
    • Use with Caution: Can introduce occlusion and make data interpretation difficult.
    • Depth Cues: Lack of depth cues can mislead viewers.
13. Consistency:
    • Colors and Labels: Maintain consistent use across all visual elements.
    • Axis Labeling: Ensure scales and units are clearly and consistently presented.
14. Cumulative Data:
    • Be Wary of Trends: Cumulative graphs always trend upwards; analyze the rate of change instead.

---

Practice Exercises

Exercise 1: Identifying Mistakes

• Task: In groups, identify as many mistakes as possible in a given visualization within 3 minutes.

• Common Mistakes Found:

  • Missing Scales and Labels: Lack of axis labels or units.
  • Inappropriate Chart Types: Using bar charts for non-quantitative data.
  • Inconsistent Ordering: Data not sorted logically (e.g., smallest to largest).

Exercise 2: Critiquing a Multi-Page Document

• Approach:
  • Page-by-Page Analysis: Stop at each page to discuss potential issues.
• Issues Identified:
  • Color Misuse: Colors not used consistently or appropriately for data types.
  • Percentage Errors: Totals not adding up to 100% where expected.
  • Redundant Visual Elements: Use of unnecessary graphics that don’t enhance understanding.

Exercise 3: Quick Critique of a Complex Visualization

• Task: Identify critical mistakes in a complex chart within 2 minutes.

• Key Findings:

  • Cut Axes and Scaling Issues: Axes not starting at zero or inconsistent scales between charts.
  • Misleading Representations: Visual elements not accurately reflecting the data.
  • Inconsistency: Varied use of colors, labels, and scales across similar charts.

---

Conclusion

• Importance of Speed Critiquing:
  • Enhances ability to quickly assess and improve visualizations.
  • Valuable skill for exams and professional work.
• Practice Makes Perfect:
  • Regular exercises help in developing intuition for spotting errors.
• Final Tip:
  • Always question the choice of visualization techniques and whether they serve the data effectively.