Data Visualization Design · Quiz 1 · May 2026

**1. Complete Solutions to Offline Paper Part A (Very Short Answer Questions, 6 Marks):** - **Q1 (4 pts) — Classify the four plots by their visualization form:** - **Figure A (US Birthdays by Month and Day grid):** **Heatmap** (or Matrix / Calendar Heatmap), encoding numerical birth frequency using color saturation across a 2D grid of months (rows) and days (columns). - **Figure B (United States Adults Ages 20 to 24 by County):** **Choropleth Map**, encoding a quantitative demographic variable across geographic county polygons using a sequential color scale. - **Figure C (Alluvial/flow connections between States and Causes of accidental deaths):** **Alluvial Diagram / Sankey Diagram** (or Parallel Sets), showing many-to-many flow magnitudes between categorical dimensions (`States` and `Cause`). - **Figure D (Treemap / Marimekko / Nightingale Rose / Radial chart shown in paper):** **Treemap** (nested rectangular areas proportional to hierarchical/categorical counts). - **Q2 (1 pt) — Gestalt principle illustrated by grouping dots/elements enclosed inside a shared boundary or shaded region:** - **Common Region** (or **Enclosure** / **Proximity** depending on the diagram): elements located within the same closed boundary are perceived as belonging to the same group. - **Q3 (1 pt) — Stevens' Power Law (\( S = I^p \)) sensation vs. physical intensity curves:** - **Length** has exponent \( p \approx 1.0 \) (linear relation, middle straight line), making length the most accurate visual channel for encoding quantitative magnitude; **Area** has \( p \approx 0.7 < 1 \) (compressive sub-linear curve); **Electric shock / Brightness / Saturation** has \( p > 1 \) (expansive super-linear curve) or \( p < 1 \). **Answer: A — I have answered this question in the offline paper.** A: **Correct:** Confirms that Part A (Questions 1–3) was answered in the offline exam booklet. B: **Incorrect:** Indicates the question was skipped. C: **Incorrect:** This question applies to all candidates taking the exam.

**1. Complete Solutions to Offline Paper Part B (Multiple Choice Questions 4 to 12, 9 Marks):** - **Q4 (1 pt) — Stevens' measurement scale (Nominal, Ordinal, Interval, Ratio):** - Categories with a natural ranking/order (e.g., disease stage I, II, III, IV or satisfaction ratings) where differences are not uniform belong to the **Ordinal** scale; temperatures in Celsius/Fahrenheit belong to the **Interval** scale; measurements with a true zero (weight, height, Kelvin, counts) belong to the **Ratio** scale. - **Q5 (1 pt) — Visual encoding effectiveness (Mackinlay's ranking for quantitative data):** - **Position along a common scale** is the most accurate channel for quantitative data, followed by Length, Angle/Slope, Area, Volume, and Color Hue/Luminance. - **Q6 (1 pt) — Preattentive visual attributes:** - A single target differing in **Color Hue** (e.g., one red circle among blue circles) or **Orientation/Size** pops out preattentively in \( < 200\text{–}250 \text{ ms} \), whereas conjunctions of two features require serial visual search. - **Q7 (1 pt) — Scott Berinato's 2x2 typology (Conceptual vs. Data-driven, Declarative vs. Exploratory):** - A conceptual + declarative diagram explaining a framework/process (like a cycle or org chart) is **Idea illustration (C)**; conceptual + exploratory is **Idea generation (D)**; data-driven + declarative is **Everyday dataviz**; data-driven + exploratory is **Visual discovery (A)** / **Visual confirmation (B)**. - **Q8 (1 pt) — Continuous Digital Elevation Model (DEM) giving terrain height at every location across Mumbai:** - **Correct choice: C. Raster, where each cell stores an elevation value.** Continuous spatial fields (like terrain elevation or satellite imagery) are naturally represented as a regular grid (raster) of elevation pixels. - **Q9 (1 pt) — Resizing country territories so each country's area is proportional to its population:** - **Correct choice: A. Cartogram.** A cartogram distorts/resizes geographic boundary areas in proportion to a quantitative variable such as population. - **Q10 (1 pt) — Two-way categorical plot where column width is proportional to group size and column height is split by treatment proportion:** - **Correct choice: B. Mosaic plot** (or Marimekko chart), which visualizes contingency tables by making tile widths proportional to marginal probabilities and heights proportional to conditional probabilities. - **Q11 (1 pt) — Color palette choice for diverging data (e.g., temperature anomaly or profit/loss around zero):** - **Diverging color palette** (two contrasting hues diverging from a neutral midpoint at zero). - **Q12 (1 pt) — Tufte's Data-Ink Ratio / Chartjunk principle:** - Maximize the **data-ink ratio** by removing non-data ink, 3D effects, and redundant gridlines/decorations (**chartjunk**). **Answer: A — I have answered this question in the offline paper.** A: **Correct:** Confirms that Part B (Questions 4–12) was answered in the offline exam booklet. B: **Incorrect:** Indicates the question was skipped. C: **Incorrect:** This question applies to all candidates taking the exam.

**1. Complete Solutions to Offline Paper Part D (Case Study on Global Water Stress Visualizations, 25 Marks):** - **Subquestion (i) — Critique and Deconstruction of Charts 1 to 4 (Data Types, Marks, Channels, and Encoding Effectiveness):** - **Chart 1 (Baseline Water Stress by Country — Choropleth / Bar Chart):** - *Data:* Geographic/Nominal (Country) + Ordinal/Quantitative (Water stress category: `<10%` Low to `>80%` Extremely High). - *Marks & Channels:* Area marks (country polygons) encoded with an ordered sequential/diverging color-luminance/hue ramp (yellow to dark red). - *Critique:* Immediately highlights regional clusters (Middle East & North Africa, South Asia), but suffers from geographic area bias (small high-stress nations like Bahrain, Kuwait, Lebanon, or Cyprus are nearly invisible on a global map). - **Chart 2 (Global Freshwater Use Over the Long-Run, 1901–2014 — Line/Area Time Series):** - *Data:* Temporal (`Year`, 1901–2014) + Quantitative (`Withdrawals` in \( \text{m}^3/\text{year} \), Ratio scale). - *Marks & Channels:* Line/Area mark with horizontal position for time and vertical position for total withdrawal volume. - *Critique:* Uses the highest-accuracy channel (aligned vertical position) to show the sharp post-1950 acceleration from ~\( 600 \text{ billion m}^3 \) to nearly \( 4 \text{ trillion m}^3 \). - **Chart 3 (Global Water Withdrawal by Sector, 1900–2010 — Stacked Area / Multi-line Chart):** - *Data:* Temporal (`Year`) + Categorical (`Sector`: Agriculture, Industries, Municipalities, Reservoirs) + Quantitative (`km}^3/\text{year}`). - *Critique:* Clearly shows that **Agriculture** accounts for the largest share (~70%) of global water withdrawals. However, for intermediate stacked bands (Industries, Municipalities), readers must judge vertical distances from a moving baseline, which is perceptually harder than a common baseline. - **Chart 4 (Agricultural Water Withdrawals, 2015 — Choropleth Map):** - *Data:* Country (Nominal/Spatial) + Annual agricultural water withdrawal in \( \text{m}^3 \) (Quantitative Ratio). - *Critique:* Uses binned sequential blue shades (`0` to `>50 billion m³`). Because it plots **total unnormalized volume** on a choropleth map rather than per-capita or per-hectare intensity, large countries (India, China, USA) dominate purely by size. - **Subquestion (ii)–(v) — Synthesis, Redesign Proposals, and Narrative Dashboard Design:** - **Addressing Area Bias in Chart 1 & Chart 4:** Pair the global choropleth map with a **ranked horizontal bar chart** (or Cleveland dot plot / population cartogram) of the top 25 water-stressed countries so small nations are clearly legible and accurately compared via common-scale position. - **Normalizing Chart 4:** Plot **agricultural withdrawal as a share (%) of renewable freshwater resources** or **per hectare of irrigated land**, and use a **bivariate plot / scatter plot** comparing total withdrawal (x-axis) vs. baseline water stress (y-axis) with bubble size = population and color = region. - **Improving Sector Comparison in Chart 3:** Use **small multiples** or a **100% stacked bar / line chart alongside absolute lines** on a common baseline so individual sector growth trajectories (Agriculture vs. Industry vs. Municipal) can be compared without baseline distortion. **Answer: A — I have answered this question in the offline paper.** A: **Correct:** Confirms that Part D was answered in the offline exam booklet. B: **Incorrect:** Indicates the question was skipped. C: **Incorrect:** This question applies to all candidates taking the exam.