Methodology
How field perception becomes a decision-grade signal
EVI captures structured civil-society perception, not direct measurement. That is its strength — field organisations observe what no satellite or model captures in real time — but it means rigour must be engineered into every layer of the design.
01 · The instrument
Six survey modules, administered annually
The questionnaire is co-designed by the academic consortium, which also conducts an independent statistical review after data collection to certify representativeness and identify bias patterns before scoring is finalised. Respondents assess specific indicators against controlled taxonomies and calibrated scales — not open-ended sentiment.
| Module | Question | Response format | Output |
|---|---|---|---|
| A · Threats | Which ten drivers pose the greatest threat to natural capital today? | Ranking from a controlled taxonomy of 14 drivers | Earth Voice Threat Ranking |
| B · Condition | How would you assess the current state of natural capital in your geography of expertise? | Scale 1–10 (1 = critical collapse; 10 = thriving ecosystems) | EVI Condition Score · 40% weight |
| C · Trend | Compared with twelve months ago, how has the trajectory of natural capital evolved? | Scale 1–10 (1 = rapid deterioration; 10 = transformational improvement) | EVI Trend Score · 30% weight |
| D · Resilience | How resilient do you believe natural ecosystems are in your geography of expertise? | Scale 1–10 (1 = near irreversible tipping points; 10 = strong recovery capacity) | EVI Resilience Score · 30% weight |
| E · Setbacks | Select the three most significant setbacks observed, then describe the most important. | Structured multi-select plus open text | Earth Voice Setback Monitor |
| F · Progress | Select the three most significant positive developments observed, then describe the most important. | Structured multi-select plus open text | Earth Voice Progress Monitor |
02 · Scoring
Composition and aggregation
The EVI Global Score is computed as a composite of the Condition Score (40%), Trend Score (30%) and Resilience Score (30%), normalised to a 0–100 scale.
National scores are computed from national NGO panel responses and aggregated to the global level using a four-factor formula: population (30%), biodiversity significance (30%), ecosystem vulnerability (25%) and national panel confidence level (15%). The academic consortium reviews and certifies this aggregation process annually.
Respondent weighting is itself structured: 50% base respondent weight, 20% field relevance, 15% geographic and ecosystem relevance, 10% evidence and transparency, and a 5% diversity correction factor. The weighting formula is reviewed and certified by the consortium each year.
03 · Geographic architecture
Local, national and global — by design
The foundational unit is the on-the-ground organisation. Each participant provides field assessments anchored to specific landscapes and ecosystems — river basins, forest patches, coastal zones, dryland corridors. This is primary observation from organisations with physical presence, not a proxy or an estimate.
Landscape-level assessments are aggregated upward: first into sub-national clusters, then national scores, then a global index. Aggregation rules — weighting by ecosystem area, NGO density and regional representativeness — form part of the methodology handbook and are subject to independent academic review.
- Project-level due diligence (TNFD LEAP, EUDR) draws on landscape-specific data.
- Country risk and portfolio reporting (CSRD, UNEP FI PRB) uses national scores.
- Investment strategy and sovereign positioning draw on the global index.
04 · The central question
How advocacy bias is prevented
The risk that NGOs systematically overstate deterioration is the most important methodological question EVI faces. Four mechanisms work in combination.
Scale
Aggregating more than 200 organisations — against a panel target of around 250 — across diverse geographies, ecosystems and organisational cultures means systematic overstatement by any sub-group is both diluted and statistically detectable.
Structure
Controlled taxonomies and calibrated scales rather than open-ended responses. Respondents assess specific indicators, not general sentiment.
Independent oversight
The North–South academic consortium is contractually responsible for bias detection and correction — reviewing response distributions, flagging outliers, and certifying the final score before publication.
Anonymous aggregation
Individual NGO responses are never published, preventing any organisation from using the survey as a lobbying instrument.
The IPBES Values Assessment — approved by 139 governments — establishes this combination of geographic diversity, structured elicitation and independent oversight as the credible standard for multi-actor knowledge aggregation.
05 · Standard of proof
What makes EVI decision-grade
"Decision-grade" means the output is sufficiently rigorous, stable and transparent to be cited in investment, policy and regulatory contexts. EVI earns this through seven design principles.
Annual comparability
Fixed methodology, published each year at the same date.
Academic certification
An independent consortium reviews questionnaire design, scoring model and bias controls before publication.
Governance independence
Methodology is governed by an independent scientific committee — not by sponsors, and not by NGO respondents.
Transparent confidence levels
Findings published with explicit uncertainty ranges and known limitations, mirroring IPBES practice.
Reproducibility
The scoring model and weightings are published in full in a methodology handbook.
Geographic representativeness
Mandatory regional balance prevents Northern-country overrepresentation.
Auditability. Individual response data is available to the academic consortium for validation, subject to NGO confidentiality protocols.
06 · Integration
Connecting a perception signal to quantitative nature frameworks
The honest starting point is an acknowledgement: EVI captures structured perception, not direct measurement. The connection to quantitative frameworks must therefore be deliberately engineered, not assumed. The methodology is designed to achieve alignment at four levels.
- Indicator mapping
- Every field-assessment question is mapped directly onto established global frameworks — TNFD core disclosure metrics, Kunming-Montreal GBF indicators, and IPBES's recognised drivers of biodiversity loss — so an EVI score can be read alongside them without translation.
- Data structuring
- Underlying data is structured using Essential Biodiversity Variables (EBVs) and SEEA Ecosystem Accounting standards, purpose-built to aggregate consistently from local observation to regional, national and global levels.
- Bias correction and calibration
- The academic consortium applies bias-correction procedures calibrated against independent reference datasets — including Global Forest Watch, WRI Aqueduct and the IUCN Red List — to ensure cross-country comparability and suppress NGO-specific systematic error.
- Flight-level flexibility
- The resulting architecture allows financial institutions to integrate EVI at whatever level their use case requires — asset or project, portfolio or country, or the global index — without loss of methodological coherence between levels.
07 · The keystone
The academic consortium, selected competitively
Academic collaboration is not a peripheral endorsement — it is a structural component of EVI's design. Rather than negotiating bilaterally with a single institution, EVI selects its methodology consortium through a structured, competitive Request for Proposals.
Competition and rigour
Multiple academic teams compete on methodological quality, global representativeness and timeline commitment — raising the bar for the selected partner.
Mandatory North–South structure
Responses must come from a consortium of at least two institutions, including at least one based in the Global South. This is non-negotiable and embedded in the selection criteria.
Financial clarity
A dedicated CHF 85,000 budget for the MVP phase, contracted as a hard cost rather than an in-kind arrangement — clear commitments on both sides from day one.
Timeline accountability
A binding development timeline with deliverables at weeks 4, 8, 16, 32 and 44. Institutions that cannot commit self-select out during the RFP, eliminating schedule risk at the outset.
| RFP component | Specification |
|---|---|
| Consortium structure | Minimum two institutions; at least one in the Global South; at least one with published expertise in biodiversity monitoring, environmental governance, or perception survey methodology |
| Dedicated budget | CHF 85,000, included in the MVP budget as a hard cost — not in kind (comprising the consortium fee and RFP process management) |
| Methodology deliverables | Survey taxonomy review (week 4); questionnaire validation (week 8); weighting model design (week 16); pilot statistical review (week 32); final scoring certification (week 44) |
| Independence safeguards | Members must disclose funding sources; no corporate funding from entities operating in sectors covered by the EVI threat taxonomy |
| Publication rights | The consortium publishes an independent methodology paper alongside the inaugural EVI report; EVI retains the right to publish under open access |
| Evaluation criteria | 40% methodological quality; 25% consortium representativeness; 20% timeline credibility; 15% budget efficiency |
The RFP is issued in week 1 of the MVP, with results announced by week 9 and the first deliverable completed by week 13. Institutions to be solicited include the Stockholm Resilience Centre, the Oxford Martin School, ETH Zurich, the Yale Center for Environmental Law & Policy, the Stanford Natural Capital Project, Universidade de São Paulo, the University of Nairobi, the Centre de Recherches Écologiques et Appliquées in Senegal, and the Asian Institute of Technology.
08 · The panel
Who is admitted, and how they are weighted
A three-layer vetting model
- Layer 1 — Eligibility filter. Minimum criteria covering legal status, environmental mission, a track record of at least three years, evidence base, independence (disclosure of major funders and governance), integrity, and geographic relevance.
- Layer 2 — Categorisation. Each vetted organisation is tagged by geography, ecosystem focus, function (advocacy, conservation, litigation, science, restoration, community mobilisation), constituency, scale and independence profile.
- Layer 3 — Sampling architecture. The panel is actively balanced to avoid over-representation of large Western organisations, targeting four respondent types: global, national leading, local and frontline, and specialist thematic organisations.
The vetting process runs in five steps: a public call for interest, secretariat screening, an independent review committee of five members including an academic consortium representative, tier assignment, and an annual refresh with full re-vetting every three years.
The NGO Internal Consensus Tool
A single designated focal point may not reflect the breadth of expertise inside a large, geographically distributed organisation. EVI therefore replicates its own aggregation logic at the organisational level.
- The organisation designates an EVI Internal Coordinator, who selects which staff, field experts and scientific advisors to invite.
- Invited respondents complete the same questionnaire modules through a private, organisation-specific portal, confidential from other participants.
- The tool aggregates internal responses using the same weighted methodology, surfacing the degree of internal consensus or divergence on each question.
- The Coordinator reviews the aggregated positions, annotates divergences, and submits a single consolidated organisational response.
- After global publication, each participant receives a private benchmarking report comparing its internal consensus to global, regional and thematic peer groups.
Built as part of the technology work package, at an estimated additional development cost of CHF 10,000.
09 · The predictive layer
The AI Earth Board
As a forward-looking complement to the core index, the EVI Secretariat is exploring the development of an AI Earth Board: a curated set of AI agents specialised by thematic focus — forests, freshwater, marine, soil, biodiversity corridors — and by geographic context. These agents would generate one-year probabilistic forecasts on natural capital trajectories, drawing on publicly available data streams, scientific literature and policy developments.
There is no practical limit to the number of agents that could be deployed, enabling granular coverage that even a large NGO panel cannot replicate. Agent forecasts would be reproducible and continuously updatable, enabling monitoring between annual editions.
The AI Earth Board is an exploratory workstream, not a committed MVP deliverable. It would be developed in close collaboration with the academic consortium and the EVI technology team.
Not a substitute for human judgment
AI Earth Board forecasts are published alongside each annual EVI edition and submitted to the real NGO panel for reaction and calibration — creating a structured dialogue between machine-generated anticipation and field-grounded expertise.
Systematic divergences between machine anticipation and field assessment are themselves informative: they mark where expert knowledge parts company with data-inferred trends.
Phase I · Minimum Viable Product
The methodology handbook is an MVP deliverable
Everything on this page is a design commitment. The MVP is what converts it into a certified, published, reproducible methodology — and co-funders shape it while its architecture is still being set.