Methodology & Sources
How we collect, interpret and visualise data.
At Numera Atlas, we aim to make our data stories transparent, understandable and reproducible wherever practical.
Every visualisation involves choices: which data to use, how to compare it, what to include and how to present the results. Those choices can influence how information is understood.
This page explains the principles behind our work.
1. Data sourcing
We prioritise identifiable, credible and publicly accessible data sources.
Depending on the subject, these may include:
- National statistical offices and government agencies.
- International organisations such as the United Nations, World Bank, OECD and Eurostat.
- Academic institutions and peer-reviewed research.
- Specialist research organisations and recognised industry bodies.
- Publicly available datasets published by companies or other organisations, where appropriate.
We aim to identify the original source of a statistic rather than relying solely on secondary reporting.
Where possible, articles include source references, reporting periods and links to the underlying data.
2. Data preparation and analysis
Before visualisation, datasets may require cleaning, restructuring or calculation.
This can include standardising country names, aligning geographic identifiers, converting units, calculating percentages, or combining related datasets.
We aim to distinguish original reported values from figures that we have calculated or derived.
Where data is incomplete, estimated or not directly comparable, we seek to make those limitations clear.
We do not intentionally replace missing observations with zero values or present estimates as directly observed measurements.
3. International comparisons
Comparing countries or regions requires particular care.
Different countries may use different definitions, reporting systems, reference periods and measurement methods.
Where possible, we use harmonised datasets designed for international comparison.
We also consider whether absolute figures or population-adjusted measures provide the more meaningful comparison.
For example, the country with the largest total emissions may not have the highest emissions per person.
When comparisons involve different reporting years or definitions, we aim to disclose that distinction.
4. Visualisation and graphic integrity
Our visualisations are designed to communicate data accurately while remaining accessible and engaging.
We use maps, bar charts, line charts, treemaps, proportional symbols and other visual formats according to the nature of the data.
Our guiding principles include:
- Preserving proportional relationships when chart area, length or size represents a numerical value.
- Using consistent units, legends and scales.
- Avoiding misleading axis manipulation or unnecessary visual distortion.
- Distinguishing unavailable data from genuine zero values.
- Labelling the relevant data period and geographic coverage.
- Selecting colours that support interpretation rather than exaggerate differences.
Maps may use different geographic projections depending on the region being displayed. Projection choice can affect the appearance of land areas and distances.
Decorative graphic elements should not alter the underlying statistical relationships.
5. Estimates, uncertainty and revisions
Not all published statistics are final.
Economic indicators may be revised, population figures may be estimated, and scientific datasets may change as new observations become available.
Where relevant, we identify estimates, provisional results and significant methodological limitations.
A chart reflects the data available for the stated period and source version; it should not automatically be interpreted as a current, real-time measurement.
6. Editorial interpretation
Our articles distinguish between what the data demonstrates and possible explanations for the observed patterns.
Correlation does not necessarily establish causation.
Where a story discusses the reasons behind a trend, we aim to support those explanations with additional evidence rather than treating the visualisation alone as proof.
We also recognise that a single metric rarely captures every aspect of a complex issue.
7. Artificial intelligence and analytical tools
Numera Atlas may use software, programming tools and AI-assisted workflows to support research, data preparation, visual design and editorial drafting.
These tools do not constitute independent evidence.
Published factual claims, calculations, source references and visual representations should be assessed against the underlying data. AI-generated suggestions are not treated as verified sources.
8. Corrections and updates
Despite care in preparing our work, errors can occur.
When a material error is identified, we aim to investigate it, correct the relevant content and explain significant changes where appropriate.
Readers can report suspected inaccuracies through our Corrections page.
Questions about our data?
If you would like to understand how a particular figure was calculated, suggest a better source or raise a concern about a visualisation, please contact us through our Contact page.
Transparency is essential to making data useful — and to earning readers’ trust.
