Sustainability data in sport: uncertainty, methods and fair comparison
How to distinguish measured and estimated data, explain the main uncertainties and compare venues without pretending unlike figures are identical.
Measured versus estimated data
Distinguish meter readings, invoices, survey data and modeled values so the reader knows what was directly observed. Name the variables that move the result most: travel distance, mode share, waste destination weights, electricity factors or other drivers. Store unit, method, baseline and limitations beside the metric name. Use a range when it can be defended and a qualitative confidence class when it cannot.
Show uncertainty where it enters the chain
Some values come from invoices or meters; others come from surveys, estimates and emission factors. Labeling those inputs is often more useful than forcing the whole result into one confidence percentage. The reader can then see whether the largest uncertainty sits in activity data, a conversion factor or the chosen boundary.
Consistency is not the same as comparability
A venue can apply the same method perfectly for three years and still be a poor comparator for another venue with a different climate, event mix or building function. Use year-on-year consistency for internal trend management and add normalization and context before making external league-table claims.
A range can be more honest than a precise-looking number
Event sustainability data often combines direct meters with estimates, supplier information and survey responses. Pretending those inputs carry the same certainty makes a final figure look cleaner than the evidence supports.
A short note on assumptions can solve much of the problem. Identify which categories are measured, which are estimated and which factors would change the conclusion most. Readers can then judge whether two venues are genuinely comparable or only presented in the same unit.
Create a simple data-quality register beside the KPI. For each major input, state the source, coverage period, whether it was measured or estimated, and the main reason it could be wrong. Metered electricity may have high confidence; spectator travel based on a small survey may not. The register lets readers see where additional data collection would actually improve the result.
When a method changes, preserve both versions for at least one overlap period where possible. Recalculate the old data using the new factor or boundary and show the difference. That bridge distinguishes methodological change from real operational change and avoids a false “improvement” created by a spreadsheet update.
Comparisons between venues need an extra test: are the services genuinely similar? Climate, event mix, floor area, attendance, hospitality share and non-event uses can all distort a ranking. If the answer is no, present the numbers as case context rather than a league table.
When two values use different boundaries, estimation methods or periods, show the difference before showing the ranking. A short uncertainty note can be more informative than an extra decimal place because it tells the reader whether the comparison is decision-safe.
Ask what would have to be true for the comparison to be fair: same boundary, similar services, stable denominator and compatible data quality. If several conditions fail, a rank is likely to create more certainty than the evidence deserves. Use ranges, context or separate case descriptions instead. Good reporting does not avoid comparison; it chooses the comparisons that the underlying data can actually support.
A precise number can still be weak evidence
Utility meters may be accurate to a fraction of a percent while spectator travel is estimated from a survey and emission factors. Adding them into one total does not give every component the same confidence. Mark which inputs are measured, calculated or estimated, and keep the largest uncertainty visible near the result.
The first defence against false precision is a stable baseline. If the comparison period or denominator keeps changing, statistical refinement cannot rescue the trend.
Compare only after checking four conditions
- Are the reporting boundaries materially the same?
- Are the denominators defined in the same way?
- Do the periods cover a similar event mix?
- Are major estimates based on comparable factors and survey methods?
If one condition fails, a side-by-side table can still be useful, but it should be described as contextual rather than a ranking. This is especially important for water intensity and energy intensity, where climate, area definitions and non-event use can shift the result without any change in operational quality.
Ranges are sometimes more honest than decimals. When a travel footprint depends heavily on car occupancy or flight origin, show the sensitivity. Decision-makers then see which assumption is worth improving instead of receiving a single number that looks more certain than the evidence.