Quiz: Data-Ink Ratio

20 questions. Aim for mastery (18+). If you score below 14, revisit the relevant sections before moving to Chapter 7.

Multiple Choice (10 questions)

1. Tufte's data-ink ratio is defined as:

(a) Data ink divided by the number of data points (b) Data ink divided by the total ink used to print the chart (c) Total ink divided by data ink (d) The ratio of colored pixels to white pixels

Answer **(b)** Data ink divided by the total ink used to print the chart. The formula is **Data-Ink Ratio = Data ink / Total ink**. Data ink is everything that directly encodes data (bars, lines, dots, colored cells). Total ink is data ink plus all non-data ink (spines, gridlines, borders, labels, decoration). Tufte proposed maximizing this ratio as a directional heuristic for clean chart design — not a literal numerical target.

2. Which of the following counts as "data ink" on a scatter plot?

(a) The x-axis line (b) The axis tick marks (c) The individual dots representing observations (d) The background color of the plotting area

Answer **(c)** The individual dots representing observations. The dots directly encode data — each dot is a specific observation from the dataset. The x-axis line, tick marks, and background color are all non-data ink. They serve comprehension or structural purposes, but none of them encodes the values in the dataset.

3. Which of the following is an example of "structural chart-junk"?

(a) A drop shadow under each bar in a bar chart (b) The top and right spines of a default matplotlib bar chart (c) A coffee cup icon embedded in a bar representing coffee sales (d) A legend that duplicates information already shown by direct labels

Answer **(b)** The top and right spines of a default matplotlib bar chart. Structural chart-junk consists of elements representing the chart's own structure — frames, spines, borders — that do not serve a reading function. Top and right spines enclose the plotting area without helping the viewer read values. Drop shadows are decorative chart-junk. The coffee cup icon is decorative (or thematic). The redundant legend is redundant chart-junk. Each is a different category.

4. The "declutter procedure" described in the chapter has three ordered steps. They are:

(a) Simplify, lighten, remove (b) Add, adjust, polish (c) Remove, lighten, simplify (d) Audit, categorize, rebuild

Answer **(c)** Remove, lighten, simplify. Remove first — delete everything that does not earn its place. Lighten second — reduce the visual weight of what remains. Simplify third — make the remaining elements simpler. The order matters: spending time lightening an element you will later delete is wasted work, so deletion comes first.

5. Which of the following is the best example of "redundant chart-junk"?

(a) A bar chart with a title above it (b) A line chart with both a legend and direct labels on the lines (c) A scatter plot with x-axis and y-axis labels (d) A heatmap with a color bar legend

Answer **(b)** A line chart with both a legend and direct labels on the lines. This is the definition of redundant encoding: the same information (which line is which) is shown twice. The viewer only needs one identification method. Direct labels are usually more efficient because they eliminate the eye movement between the line and the legend. A chart title, axis labels, and heatmap color bar are all non-redundant — each conveys information not available elsewhere.

6. The threshold concept for Chapter 6 is:

(a) More ink is always worse than less ink (b) Defaults are chart-junk (c) Color should never be used for decoration (d) Every chart needs a title

Answer **(b)** Defaults are chart-junk. The threshold concept states that the default output of any plotting library is a starting point, not a finished product. Default settings include structural, decorative, and redundant elements that add visual weight without adding information. The mental shift — treating defaults as something to override rather than accept — is the main transformation this chapter is trying to produce.

7. Which of the following is a "data desert" failure mode of over-aggressive decluttering?

(a) Removing the top and right spines of a bar chart (b) Deleting the y-axis and expecting the viewer to read magnitudes from bar lengths alone (c) Lightening gridlines from black to pale gray (d) Removing a figure border

Answer **(b)** Deleting the y-axis and expecting the viewer to read magnitudes from bar lengths alone. The data desert is what happens when decluttering goes too far and essential comprehension aids are removed. The y-axis of a bar chart is non-data ink but is essential for quantitative reading — without it, the viewer can see relative magnitudes but cannot determine actual values. The other options are appropriate decluttering operations that do not cross into data desert territory.

8. The Bateman et al. 2010 paper ("Useful Junk?") found that:

(a) All visual embellishment reduces comprehension (b) Thematic embellishment can improve memorability of charts without sacrificing comprehension accuracy (c) Tufte's data-ink ratio is mathematically wrong (d) Readers cannot tell the difference between decorated and plain charts

Answer **(b)** Thematic embellishment can improve memorability of charts without sacrificing comprehension accuracy. The Bateman study found that charts with thematic visual embellishments (e.g., a coffee cup in a chart about coffee sales) were more memorable weeks later than plain versions, without measurable loss in reading accuracy. The result complicates strict Tufte minimalism by showing that, for certain audiences and tasks, embellishment has value. It does not invalidate the data-ink ratio as a general heuristic — it identifies specific conditions under which the heuristic is not the whole story.

9. Which of the following non-data ink elements should NOT typically be deleted during decluttering, even for an aggressively minimalist chart?

(a) A decorative drop shadow under each bar (b) A source attribution citing the data provenance (c) The top spine of a bar chart (d) A thick rectangular border around the figure

Answer **(b)** A source attribution citing the data provenance. Source attribution is non-data ink, but it serves essential comprehension and ethical purposes (Chapter 4). Deleting it leaves the viewer without information they need to evaluate the chart's credibility and without the context needed to interpret the data. The other three options — drop shadows, top spines, and figure borders — are typical targets of the declutter procedure.

10. The chapter argues that decluttering has an ethical dimension because:

(a) Cluttered charts are more expensive to produce (b) Cluttered charts are harder to read accurately, which makes viewers more likely to form impressions from decorative features rather than data (c) Tufte is an authority on ethics (d) Decluttering is required by most corporate style guides

Answer **(b)** Cluttered charts are harder to read accurately, which makes viewers more likely to form impressions from decorative features rather than data. The ethical connection links back to Chapter 4. A cluttered chart competes for the viewer's attention, and the most salient features often win that competition. If the salient features are decorative rather than data, the viewer's impression is formed on the wrong basis. A clean chart makes the data the most salient feature, which helps ensure that the impression the viewer forms is the impression the data supports.

True / False (5 questions)

11. "The data-ink ratio is a mathematical law: any chart with a higher ratio is provably better than any chart with a lower ratio."

Answer **False.** The data-ink ratio is a directional heuristic, not a mathematical law. Tufte himself did not claim it was a law. The principle pushes you to examine non-data elements and ask whether they earn their place, but the "right" ratio depends on context, audience, and the information the chart is trying to convey. Two charts with identical ratios can differ in quality for many other reasons, and a chart with a lower ratio can sometimes be better if it includes well-chosen comprehension aids.

12. "In Part II of this book, the chapters do not contain code. The principles in Chapter 6 are implemented through matplotlib code in Chapter 12."

Answer **True.** Parts I and II of this book are library-agnostic. Chapter 6 teaches you the declutter mindset and the taxonomy of chart-junk. Chapter 12 (Customization Mastery) shows you the specific matplotlib function calls — `ax.spines[...].set_visible(False)`, grid styling, tick customization — that implement the declutter operations. Separating principles from implementation is a deliberate pedagogical choice: the principles outlast the library, and you should internalize them before learning the functions.

13. "A chart title always counts as chart-junk because it is non-data ink."

Answer **False.** A chart title is non-data ink, but it is not chart-junk. A good title — especially an action title that states the finding (Chapter 7) — helps the viewer understand the data and serves essential comprehension. Non-data ink and chart-junk are not the same thing. Chart-junk is a subcategory of non-data ink: the part that does not earn its place. Titles, axis labels, source attributions, and essential gridlines are non-data ink that does earn its place.

14. "Every default setting in every plotting library should be overridden for every published chart."

Answer **True (with a small asterisk).** The chapter's position is that defaults are a starting point, not a finished product — and that every publication-quality chart requires deliberate design decisions rather than accepted defaults. The small asterisk: some defaults happen to be the right choice for your chart, and in those cases you do not need to change them. But you should always *check* each default and make a deliberate decision, not simply accept whatever the library produces. In practice, after enough charts, you will build a set of personal defaults (via matplotlib rcParams, for example) that override the worst library defaults globally.

15. "Decluttering a chart means removing information from it."

Answer **False.** Decluttering removes visual elements that do not encode information — structural chart-junk, decorative chart-junk, redundant encodings. The actual data — the bars, lines, dots, colored cells — is preserved. A successful decluttering produces a chart with the same information as the original but far less visual noise. If you find yourself removing data in the name of decluttering, you have crossed into a different activity: editing the analysis, not the chart.

Short Answer (3 questions)

16. In three to four sentences, explain the difference between "useful non-data ink" and "wasted non-data ink," and give one example of each.

Answer **Useful non-data ink** is ink that does not encode data but helps the viewer interpret the data — axis labels, source attributions, scale legends, reference lines for thresholds, and essential gridlines. Removing these would make the chart harder to understand even though they are not strictly data. **Wasted non-data ink** is ink that neither encodes data nor helps the viewer interpret it — decorative borders, drop shadows, redundant legends, 3D effects. Removing these makes the chart cleaner without sacrificing comprehension. Example of useful: the y-axis numerical labels on a bar chart (necessary for quantitative reading). Example of wasted: the top and right spines of a default matplotlib bar chart (enclose the plotting area but serve no reading function).

17. Explain in three to four sentences why the declutter procedure orders its steps as "remove, lighten, simplify" rather than some other order.

Answer The order minimizes wasted work. If you lighten or simplify an element that you are going to delete later, the lightening and simplification are wasted. By deleting first, you ensure that all the effort spent on Step 2 (Lighten) and Step 3 (Simplify) goes to elements that will survive. The order also matches the cognitive workflow: the decision to delete is binary (keep or not), the decision to lighten is a matter of degree, and the decision to simplify is the most detailed and context-dependent. Moving from coarse decisions to fine decisions is how most design processes work.

18. Name and briefly describe each of the five categories of chart-junk. For each, give one sentence on what the "fix" looks like.

Answer **(1) Decorative:** elements added for aesthetic reasons that do not encode or aid interpretation (drop shadows, gradients, decorative icons). **Fix:** delete them; professional charts are cleaner, not flashier. **(2) Structural:** elements representing the chart's own frame (top and right spines, figure borders, legend borders). **Fix:** systematically remove or lighten; top/right spines usually delete. **(3) Redundant:** elements duplicating information already shown (legends on charts with direct labels, value labels on charts with axes). **Fix:** choose the best single encoding and delete the backups. **(4) Default:** elements that appear because of software defaults (default gridline colors, default tick densities, default background colors). **Fix:** override defaults deliberately for every publication-quality chart. **(5) Dimensional:** elements that add false dimensions (3D bars, 3D pies, perspective effects). **Fix:** delete them always; there is no legitimate use case for 3D in quantitative visualization.

Applied Scenarios (2 questions)

19. You are reviewing a draft chart from a colleague. The chart is a bar chart of Meridian Corp's five product lines by annual revenue. The chart has: a thick black border around the figure, a top and right spine enclosing the plotting area, horizontal and vertical gridlines in medium gray, a default title "Chart1", each bar labeled with its exact numerical value on top, a y-axis with numerical tick labels, a legend box identifying "Revenue" (the only series), and a subtle drop shadow on each bar.

(a) For each of the elements above, classify it as data ink, useful non-data ink, or wasted non-data ink. (b) Apply the Step 1 (Remove) pass of the declutter procedure and list what you would remove and what you would keep. (c) Explain why you would delete either the per-bar value labels or the y-axis (not both), and justify your choice.

Answer **(a)** Data ink: the five bars themselves. Useful non-data ink: the x-axis category labels (product line names), the y-axis numerical labels (or alternatively, the per-bar value labels). Wasted non-data ink: the figure border, the top and right spines, the vertical gridlines (redundant with bars), the "Chart1" title (useless default), the legend box (redundant with a single-series chart), the drop shadows. **(b)** Remove: figure border, top spine, right spine, vertical gridlines, "Chart1" title, legend box, drop shadows. Keep: the bars, the x-axis category labels, and one of (y-axis numerical labels / per-bar value labels). Also rewrite the title to be an action title that states the finding. **(c)** Either the y-axis numerical labels or the per-bar value labels are redundant; the other is sufficient for quantitative reading. Choose the per-bar value labels if the number of bars is small (5 qualifies) and precise values are important to the audience — this lets you delete the y-axis entirely, producing a dramatically cleaner chart. Choose the y-axis labels if the focus is on relative comparison and precise values are less important. For five executive-facing product lines, the per-bar value labels are probably the better choice: executives care about the specific numbers, and deleting the y-axis gives the visual space back to the bars themselves.

20. A public health team is producing an infographic for a general audience to explain vaccination coverage trends over time. The infographic will appear on social media, where viewers glance at it for 3-5 seconds and scroll on. The team's draft chart is a simple line chart of vaccination rates from 2000 to 2024, but the team's design lead wants to add: (a) a background illustration of a hospital, (b) cartoon figures of people next to the line, (c) a color gradient behind the plotting area, (d) a large, stylized decorative frame.

(a) Which elements would Tufte's declutter principle tell you to reject, and why? (b) Which elements (if any) could be defended under the Bateman "thematic embellishment" principle, and under what conditions? (c) Given the specific context — social media, 3-5 second glance, general audience — what is your recommendation to the design lead?

Answer **(a)** The strict Tufte principle would reject all four elements. Background illustrations, cartoon figures, color gradients behind the plotting area, and decorative frames are all decorative chart-junk by the chapter's taxonomy — they add visual weight without encoding data or aiding interpretation. They compete with the data for attention, slow comprehension, and (in the case of the background illustration) may introduce luminance contrast that makes the line harder to read. **(b)** The cartoon figures and possibly the hospital background could potentially be defended under Bateman's principle if (1) they are thematic — clearly connected to the topic of vaccination — rather than arbitrary, (2) they do not interfere with the encoding accuracy of the line itself, and (3) memorability is more important than reading speed in the target context. The color gradient and decorative frame cannot be defended under Bateman — they are not thematic and offer no memorability benefit. **(c)** In the specific context of social media with a 3-5 second glance, the recommendation is to reject most of the design lead's suggestions. A 3-5 second glance leaves very little time for the viewer's visual system to parse complex decoration, so the chart must be extremely simple and the data must be the most salient feature. Allow at most one thematic element (perhaps a small vaccination icon next to the title) as a Bateman-style memorability aid. Replace the rest with: a clean line chart, an action title stating the finding, direct labels for the key data points, and a source attribution. The chart should read in 3 seconds and still be memorable a week later — but it gets there through clarity, not through illustration. If the design lead insists on decoration, push back with the specific argument that decoration steals attention from the data in a 3-second window.

Review your results against the mastery thresholds at the top. If you scored below 14, revisit Sections 6.1 through 6.5 — especially the declutter procedure and the five categories of chart-junk — before starting Chapter 7 on typography and annotation. The next chapter assumes you can already look at a chart and identify what does not earn its place.