Chapter 2 Quiz: How the Eye Sees

Chapter 2 Quiz: How the Eye Sees

Instructions: Answer all 20 questions. For multiple-choice questions, select the single best answer. For true/false, state whether the claim is true or false and provide a one-sentence justification. For short answer, respond in 2-4 sentences. For applied scenario questions, justify your answer using specific principles from the chapter.

Multiple Choice (10 questions)

Q1. Pre-attentive processing is best described as:

(a) The conscious effort required to decode a legend (b) The automatic, parallel extraction of visual properties occurring in under 250 ms (c) The process of reading axis labels before interpreting data points (d) The viewer's deliberate comparison of two chart elements

Q2. According to the Cleveland-McGill hierarchy, which of the following encoding channels produces the MOST accurate quantitative judgments?

(a) Area (b) Color saturation (c) Angle (d) Position on a common scale

Q3. A pie chart encodes proportional values primarily through which visual channel?

(a) Position (b) Length (c) Angle (d) Color hue

Q4. Jacques Bertin classified the retinal variable "color hue" as selective but NOT ordered. In practical terms, this means:

(a) Color hue can distinguish categories but cannot convey a natural ranking (b) Color hue cannot be used in any chart (c) Color hue is more accurate than position for quantitative data (d) Color hue should always be used for continuous variables

Q5. In a scatter plot, you color-code points by category (blue for Group A, orange for Group B). Which Gestalt principle causes the viewer to perceive all blue points as belonging together?

(a) Proximity (b) Enclosure (c) Similarity (d) Closure

Q6. Stevens's Power Law, with an exponent of approximately 0.7 for area, predicts that:

(a) Viewers will overestimate differences in area (b) Viewers will accurately perceive area proportions (c) Viewers will systematically underestimate differences in area (d) Viewers will perceive area more accurately than length

Q7. A line chart of five time series with similar colors and frequent crossings is difficult to read. Which combination of Gestalt principles explains the difficulty?

(a) Similarity groups all lines together; connection within each line conflicts with the color-based grouping; crossings disrupt continuity (b) Proximity separates the lines too much (c) Closure causes the viewer to see closed shapes instead of lines (d) Enclosure prevents the viewer from distinguishing the lines

Q8. A designer uses a rainbow color scale (red-orange-yellow-green-blue-violet) to encode temperature values from 0 to 100 degrees. What is the primary perceptual problem?

(a) The colors are not pre-attentive (b) Color hue is a categorical channel applied to continuous quantitative data, creating false boundaries and no natural ordering (c) Rainbow scales always cause color vision deficiency issues (d) The scale has too many colors for the visual system to process

Q9. Change blindness is MOST likely to affect chart comprehension in which scenario?

(a) A viewer looks at a single static chart with clear color encoding (b) A viewer compares two versions of a chart displayed side by side (c) A viewer sees a chart on one presentation slide, then a modified version on the next slide (d) A viewer reads a chart title before examining the data

Q10. The working memory limit for visual information is approximately:

(a) 1-2 items (b) 3-5 items (c) 10-12 items (d) 20 or more items

True/False (4 questions)

Q11. TRUE or FALSE: Pre-attentive processing can detect a conjunction of two features (e.g., a red circle among red squares and blue circles) as efficiently as it detects a single feature (e.g., a red item among blue items).

Q12. TRUE or FALSE: In the Cleveland-McGill hierarchy, length encoding is ranked as more accurate than angle encoding for quantitative judgments.

Q13. TRUE or FALSE: The Gestalt principle of connection explains why drawing a line through sequential data points in a scatter plot causes the viewer to perceive them as a single series.

Q14. TRUE or FALSE: A bar chart that does not start at zero on the y-axis distorts the length encoding, because the visible bar lengths no longer accurately represent the data values.

Short Answer (3 questions)

Q15. Explain the relationship between Bertin's retinal variables and the Cleveland-McGill hierarchy. How are the two frameworks complementary?

Q16. A chart designer wants to highlight a single outlier data point in a scatter plot containing 200 points. Describe two different pre-attentive attributes the designer could use to make the outlier pop out, and explain why each would work.

Q17. Describe the Gestalt principle of closure and give one specific example of how it can be used to reduce visual clutter in a chart.

Applied Scenario (2 questions)

Q18. You are designing a chart to compare the average test scores of students across eight schools. The scores range from 72 to 89 (out of 100). A colleague proposes the following designs:

Design A: A pie chart where each school's slice represents its score as a proportion of the total across all schools.
Design B: A horizontal bar chart with schools on the y-axis and scores on the x-axis, starting at zero.
Design C: A horizontal dot plot with schools on the y-axis and scores on the x-axis, starting at 60 (to zoom in on the relevant range).

Rank these three designs from most to least effective for the comparison task. For each design, identify the primary encoding channel used for the score variable and reference the Cleveland-McGill hierarchy to justify your ranking. Address the zero-baseline issue where relevant.

Q19. A public health team wants to show vaccination coverage rates across 30 countries, grouped into three income categories (low, middle, high income). They have two candidate visualizations:

Candidate 1: A choropleth world map where each country is shaded from light (low coverage) to dark (high coverage), with income groups indicated by border style (solid, dashed, dotted).
Candidate 2: A grouped dot plot with vaccination rate on the x-axis (position encoding), countries listed on the y-axis, and income groups distinguished by color and vertical proximity (countries within the same income group are clustered together).

(a) For each candidate, list the encoding channels used and the variable each encodes.

(b) Using the Cleveland-McGill hierarchy and Gestalt principles, argue which candidate will produce more accurate comparisons of vaccination rates between specific countries.

(c) Is there any advantage to the choropleth that the dot plot lacks? When might that advantage matter more than encoding accuracy?

Analysis (1 question)

Q20. A business analyst produces the following chart for a quarterly review meeting: a 3D stacked bar chart showing revenue across four product lines for each of the four quarters. The chart uses perspective rendering (bars further from the viewer appear smaller), four distinct colors for the product lines, and a y-axis that starts at $10M instead of $0.

Conduct a systematic perceptual analysis of this chart. In your analysis:

(a) Identify every encoding channel in use and the variable it represents.

(b) Identify at least three specific perceptual problems, referencing the Cleveland-McGill hierarchy, Gestalt principles, and/or perceptual limitations from Section 2.6.

(c) For each problem, explain what the viewer will experience (what will be misperceived, what will be difficult to compare, what information will be lost).

(d) Propose a redesigned chart that addresses all identified problems. Specify the chart type, encoding channels, and which Gestalt principles your redesign leverages.

Answer Key

Q1: (b). Pre-attentive processing is automatic, parallel, and occurs in under 250 ms.

Q2: (d). Position on a common scale is ranked highest in the Cleveland-McGill hierarchy.

Q3: (c). Pie chart slices are decoded primarily by their central angle.

Q4: (a). Hue distinguishes categories but has no inherent order (red is not "more" than blue).

Q5: (c). Similarity causes elements sharing a visual property (color) to be grouped.

Q6: (c). The sub-linear exponent (0.7 < 1.0) means perceived area grows more slowly than actual area, so differences are underestimated.

Q7: (a). Similar colors group all lines together, conflicting with the connection within each individual line, while crossings break the continuity the eye expects.

Q8: (b). Hue is inherently categorical; mapping it to a continuous variable creates artificial categorical boundaries and lacks perceptual ordering.

Q9: (c). Change blindness occurs when a visual disruption (slide transition) prevents direct comparison, making changes between the two versions undetectable.

Q10: (b). Modern estimates place the visual working memory limit at approximately 3-5 items.

Q11: FALSE. Feature integration theory (Treisman) shows that conjunction searches require slow, serial processing because no single pre-attentive channel distinguishes the target.

Q12: TRUE. Cleveland and McGill's experiments placed length above angle in the accuracy ranking.

Q13: TRUE. Connection is the Gestalt principle that causes linked elements to be perceived as belonging to the same group, which is the mechanism behind line charts.

Q14: TRUE. Bar charts rely on the length of the bar from a baseline; a non-zero start truncates the visible length, making differences appear proportionally larger than they are.

Q15: Bertin's retinal variables provide a theoretical taxonomy of visual encoding channels and classify them by their suitability for different data relationships (selective, ordered, quantitative). The Cleveland-McGill hierarchy provides the experimental evidence for how accurately each channel is decoded. Bertin asked "what can each channel express?" while Cleveland and McGill asked "how accurately does each channel express it?" Together, they provide both a vocabulary for design choices and an empirical ranking to guide those choices.

Q16: (1) Color hue: Make the outlier red while all other points are gray or blue. Color hue is pre-attentive; the unique color will pop out across the entire visual field in under 250 ms. (2) Size: Make the outlier point substantially larger than the other points. Size differences are pre-attentive; the large mark will be detected without serial search. Either approach ensures the viewer's visual system flags the outlier before conscious analysis begins.

Q17: Closure is the Gestalt principle by which the visual system completes incomplete shapes, perceiving closed forms even when gaps exist. In chart design, this means gridlines can be drawn lightly or partially — the viewer's eye will complete the grid pattern mentally. For example, removing the top and right borders of a chart frame (leaving only the x- and y-axes) reduces visual clutter while the viewer's perception completes the implied rectangular frame through closure.

Q18: Ranking from most to least effective: C > B > A. Design C (dot plot, position encoding on a zoomed scale) uses the most accurate channel (position on a common scale) and focuses on the relevant range. Design B (bar chart, length encoding) uses the second most accurate channel, but starting at zero wastes visual space since the scores cluster in 72-89 — the bars will all look nearly the same length. A dot plot avoids the zero-baseline problem because dots encode position, not length. Design A (pie chart, angle encoding) is the least effective: the scores are close together, the chart does not naturally support the comparison task (it shows part-of-whole, not individual magnitudes), and angle is ranked well below position and length in the hierarchy. Note: the zero-baseline issue is a concern for bar charts (which use length encoding) but not for dot plots (which use position encoding), which is why C is preferred over B despite the non-zero start.

Q19 and Q20: Model answers provided in Appendix A. These require extended analysis and multiple valid approaches exist.