ISYS3438 Chap.2 Critiquing Deceptive Visualisations
Critiquing Deceptive Visualisations
A useful critique identifies whether the problem lies in the business question, the available data or the visual encoding. This chapter applies the Trifecta Check-up to baselines, scales, denominators, sorting, colour and omitted categories. It also distinguishes legitimate focused scales from distortion by asking what perceptual task the chart sets.
The final step is constructive: specify the smallest redesign that removes the misleading effect while preserving the valid analytical question.
What this chapter covers
- 01
trifecta check-up
- 02
question and data fit
- 03
axis baselines
- 04
scale intervals
- 05
denominator choice
- 06
constructive redesign
Worked example · free
Repair a magnified supplier difference
- +1Quantify the actual 0.3 percentage-point difference.
- +1Replace the truncated bars with a zero-based bar or labelled dot plot.
- +1Add sample size and the agreed performance threshold before declaring failure.
Key terms
- Trifecta Check-up
- The Trifecta Check-up is a critique frame connecting question, data and visual representation.
- Truncated baseline
- A truncated baseline begins a quantitative axis above zero and may magnify length differences.
- Chart junk
- Chart junk is decorative or redundant visual material that competes with the data.
- Constructive critique
- Constructive critique identifies a failure and specifies a defensible repair.
Critiquing Deceptive Visualisations FAQ
Are all truncated axes misleading?
No. A narrow line-chart scale may legitimately reveal small variation when the axis is clear and length from zero is not the encoded quantity. Bars usually need zero because their length represents magnitude.
Why are denominators so important?
Counts often reflect population size, opportunity or workload. A rate answers a different question by comparing events with exposure. Strong critique states which decision each version supports instead of treating rates as automatically superior.
How should I finish a critique?
End with a redesign specification: preserve the valid question, add the missing field or denominator, change the mark or scale, and state the narrower claim the revised display can support.
Assessment move
Use the same four-line format on five charts: question, data, encoding and repair. Include one positive design choice so the critique is balanced. Rebuild a single chart twice, once for exact comparison and once for pattern recognition, then explain how the intended task changes the better form. Create a critique sheet with three columns for question, data and visual representation.
In the first column, write the decision the chart appears to support and the strongest claim a reasonable reader might take away. In the second, identify observation unit, time boundary, denominator, missing groups and transformations. In the third, inspect mark type, scale, ordering, annotation and emphasis.
A criticism is complete only when it connects a visible choice to a likely interpretation and explains why that interpretation matters for the stated decision. Test axes according to the encoded quantity rather than applying slogans. Bar length normally requires a zero baseline because distance from the baseline carries magnitude.
A line chart may use a restricted, clearly labelled range to reveal small variation, provided the design does not imply a larger absolute change. Check interval regularity, log scales, dual axes and broken ranges. Recalculate percentage change from the printed values and ask whether the selected start and end dates manufacture the story. Audit denominators by writing both numerator and exposure in words.
A region with the most incidents may simply have the largest population, customer base or operating time. Compare counts when workload matters and rates when relative risk matters; show both if the audience needs capacity and fairness. Inspect category aggregation for Simpson's paradox or hidden subgroup changes.
When uncertainty or sample size affects the ranking, add it rather than decorating a fragile league table with confident colour. End every critique with a redesign brief. Preserve the legitimate purpose and data, then specify the comparison the audience must make, the encoding that supports it and the context that prevents overclaiming. Sketch a common-scale alternative, a clearer title and one annotation tied to evidence.
State what the redesign still cannot establish. This constructive close is stronger than listing faults because it demonstrates that you can translate diagnosis into a usable and ethically proportionate communication choice. Practise writing critiques in a four-part paragraph. First describe the visible choice precisely. Next explain the perceptual or statistical effect on interpretation.
Then state why that effect matters for the business question. Finally propose a redesign that retains legitimate information. Support the diagnosis by reconstructing values where possible instead of relying on labels such as misleading or cluttered. Compare the original and redesign under the same data and claim, and note any trade-off introduced by the repair.
This format turns a list of preferences into an evidence-based argument. It is especially useful when a chart is not wholly false but combines a valid pattern with a baseline, denominator or annotation that exaggerates its importance. Keep a before-and-after portfolio of three repaired visuals. Caption each pair with the original inference, diagnosed mechanism and revised comparison.
Include one case where the correct repair is additional context rather than a new chart. Revisit the portfolio before assessment writing to practise precise language about scale, denominator, aggregation and emphasis.