ISYS3438 Chap.1 Why Visualise: Purpose, History and Ethics
Why Visualise: Purpose, History and Ethics
Data visualisation turns observations into marks that an audience interprets and may use to make a business decision. This chapter treats that route as an analytical and ethical chain. It explains what visual displays contribute beyond tables, why integrity and privacy must be considered before styling, and how claim strength should remain proportional to the evidence.
The worked example distinguishes a visible exception from a persistent trend and shows why a missing denominator can change the conclusion.
What this chapter covers
- 01
data, encoding, claim and decision
- 02
visual pattern recognition
- 03
data integrity and privacy
- 04
denominators and exposure
- 05
ethical annotation
- 06
claim boundaries
Worked example · free
Interpret a spike without inventing a trend
- +1Plot all weeks on one scale and label the final value.
- +1Describe one exception rather than a persistent trend.
- +1Request sales volume and product mix before interpreting return risk.
Key terms
- Data visualisation
- Data visualisation is the graphical representation of information to support interpretation and communication.
- Claim boundary
- A claim boundary states the strongest conclusion that the displayed evidence can support.
- Exposure denominator
- An exposure denominator is the population or opportunity against which an event count is compared.
- Data integrity
- Data integrity is the accuracy, consistency and trustworthy handling of data across its lifecycle.
Why Visualise: Purpose, History and Ethics FAQ
Why visualise instead of showing a table?
A visual display can reveal comparison, trend, clustering or exception faster than scanning many individual values. A table remains useful for exact retrieval. The choice depends on whether the audience needs pattern recognition, precise lookup or both.
When does a chart become an ethical problem?
A chart becomes ethically problematic when collection, privacy, selection, scale or annotation leads the audience toward a belief the evidence cannot justify. Intent is not required; an accidental distortion still needs correction.
How should uncertainty appear?
Use wording, intervals, scenario ranges or visible missing-data states appropriate to the analysis. At minimum, distinguish observed pattern from interpretation and state what additional evidence is needed before acting strongly.
Assessment move
Pick one chart from business news and trace its full chain: source data, encoding, implied claim and likely decision. Identify the denominator, any excluded group and the strongest defensible wording. Redraw only after you can explain what the original view helps the reader see and where it risks overclaiming. Build a weekly evidence diary rather than a gallery of attractive charts.
For each visual encountered in class, work or media, record its intended audience, the decision it invites and the observation unit behind one mark. Rewrite its main claim at three strengths: a direct description, a cautious association and a causal explanation. Mark which version the data can actually support. Then identify the denominator, comparison period and excluded records.
This routine makes the boundary between persuasion and evidence visible before it becomes an assessment problem. Run a privacy and harm review on the same example. Ask whether labels, filters, precise locations or small groups could identify a person even when names are absent. Consider who supplied the data, whether that use was reasonably expected and which people are missing.
Test whether the category scheme or ordering embeds a value judgement. Propose an aggregation, suppression rule or alternative framing that reduces harm while preserving the analytical purpose. Explain the trade-off instead of claiming that anonymisation removes every risk. Practise historical comparison by rebuilding one claim with a table and one with a visual form.
Note what becomes faster to perceive, what exact values become harder to retrieve and how the technology constrains the design. Avoid treating historical examples as a parade of inventions; connect each form to the question, available data and audience. This prepares you to explain why a contemporary dashboard needs both overview and lookup rather than assuming visualisation is always superior.
Before submitting work, trace the memorable sentence backwards to the data. Every quantity and comparison should point to a visible mark, a defined field and a reproducible transformation. Separate a workload count from a rate, a temporal coincidence from an explanation and a forecast from an observed value. Finish with a proportionate action and one piece of evidence that would change it.
Keep that audit beside the chart so ethical restraint is part of the communication, not a disclaimer added after the design is complete. Use a final peer review to test communicative responsibility. Give the chart and its main sentence to someone who has not seen the dataset. Ask what population, period, comparison and cause they infer, then compare those inferences with the documented evidence.
Revise any title or emphasis that invites a stronger conclusion than the analysis supports. Check that sources can be traced, confidential detail is absent and uncertainty appears near the affected claim. Record the reviewer's most serious alternative interpretation and explain whether new data, a different comparison or narrower wording resolves it. This produces a concrete ethical audit trail for a report or presentation.
It also separates genuine disagreement about action from preventable ambiguity in the design. As a retrieval exercise, explain one chart to a colleague without showing it. Include its population, encoding, main pattern, uncertainty and ethical risk. Then reveal the chart and ask which features encouraged a different interpretation. Use the discrepancy to revise the title or context.
The exercise tests whether visual rhetoric and spoken explanation communicate the same bounded claim.