Become a chart detective
How many different impressions can the same numbers create?
20-30 minutes
Start with these sources
Your first move: Choose a numeric time series in Our World in Data and save its data and metadata.
Without AI
- Draw the same years and values twice: first with the full vertical range, then with a narrower labeled range.
- Add the units, dates, source, and axis labels to both versions; keep all underlying values identical.
- Write a caption explaining what each design emphasizes and which comparison is fairest for your question.
With GenAI
Bring your own AI tool and the source material. This atlas does not send your data anywhere or run an AI model.
- Supply the actual data, units, definition, and source link to an AI tool and ask for two chart specifications.
- Ask it to critique what a narrow axis or selected start year could make a reader overlook.
- Create the charts in your spreadsheet, verify the plotted points, and rewrite the captions in your own words.
A prompt to make your own
Act as a chart editor using only my supplied dataset and metadata. Propose two views of the same values: a full-range view and a clearly labeled zoomed view. Explain the impression each creates without calling every zoom dishonest. Cite the supplied source for numerical claims. Preserve missing data gaps and do not fabricate numbers or interpolate them. List the axis settings and date range so I can reproduce both charts. Data, metadata, and source URL: [Paste here.]
Make something: A pair of annotated charts and a short explanation of their different impressions.
Keep your footing
- A zoomed axis can be useful when it is explicit and appropriate.
- Keep values and units identical between views.
- Do not silently connect across missing observations.
Why follow this question?
Why this is interesting
A chart contains choices: where an axis begins, which years appear, and whether a large change gets a large area on the page. Making those choices yourself helps you see them in the wild. The goal is to explain a visual effect without accusing an unknown author of deception.
The Our World in Data FAQ explains how to retrieve charts, data, and source information. Choose a short series so each plotted point remains easy to check. Your annotations should make the experiment visible to a reader.
Follow another thread
Try the same question with a World Bank series, or compare a total with a per-person measure. Now the definition changes as well as the display. Label that as a different experiment: a new denominator can answer a new question rather than simply make a chart look different.
Go a little further
Invite a friend to write a headline for each chart before showing that the data are identical.