The Gold Standard of Anthropology | Evaluator Conflicts of Interest, Funder Influence and Finding Independence

1. Define independence as a set of decisions

Back to contents

Separate organisational, financial, methodological and reporting independence. An evaluator may be employed by the commissioner yet retain control of analysis; an external firm may still be dependent on renewal income or restricted access. State which decisions belong to the commissioner, evaluator, community authority and oversight body.

Independence is strongest when another reader can see how the question, sample, evidence, interpretation and final report were governed. A declaration without a decision map is not enough.

2. Build a conflict-of-interest register before appointment

Back to contents

Ask proposed evaluators to disclose prior work, financial ties, advocacy roles, personal relationships, intellectual commitments and pending bids that could reasonably be perceived as relevant. Include subcontractors and specialist advisers, not only the lead evaluator.

A conflict can be actual, potential or perceived. Disclosure does not always require exclusion; it requires a recorded decision to exclude, mitigate, balance expertise or explain why the relationship does not compromise the assignment.

3. Distinguish funder interest from improper influence

Back to contents

Funders may legitimately define the programme, decision timetable, budget and broad evaluation questions. They should not preselect a favourable conclusion, suppress an inconvenient subgroup, require unsupported causal language or make payment conditional on a positive result.

Write these boundaries into the terms of reference. When a requested change improves factual accuracy, accept and log it. When it changes a supported finding to protect reputation, retain the finding and document the disagreement.

4. Protect question setting and sampling choices

Back to contents

A narrow question can make failure invisible before data collection starts. Include delivery staff, participants, affected communities and intended users when defining the evaluation matrix. Show which questions were proposed, accepted, revised or declined and why.

Sampling decisions deserve the same scrutiny. If a commissioner supplies only successful sites or cooperative participants, record the coverage gap and seek an independent route to the missing cases.

5. Create firewalls around analysis, payment and publication

Back to contents

Use milestone payments tied to completed work rather than desirable findings. Give the evaluator direct access to authorised evidence, a secure route for participant contact and control of analytic files. Predefine factual review, disclosure review and publication stages.

A firewall does not prevent challenge. It ensures that challenges concern evidence, method, confidentiality or interpretation and leave an audit trail instead of becoming private instructions attached to the final invoice.

6. Worked example: a funder asks to soften a negative finding

Back to contents

A foundation commissions an evaluation of a youth heritage programme. The evaluator finds that attendance improved but that young participants had little influence over exhibition captions. Before renewal, the funder asks to replace “limited decision authority” with “emerging leadership” and to remove two critical quotations. The evaluator also hopes to win a larger contract.

The agreed protocol separates factual correction from analytic judgement. The programme corrects one date; the community safeguarding lead reviews the quotations for disclosure risk; the evaluator retains the finding, explains the coding evidence and includes the programme response. Payment follows delivery of the report, not acceptance of its conclusion.

7. Include community and participant power in the independence test

Back to contents

Influence does not flow only from funders. Gatekeepers can select agreeable speakers, senior representatives can silence internal disagreement, and evaluators can privilege familiar professional voices. Map who can grant access, who can refuse, and who may face consequences for speaking.

Use confidential routes, purposive inclusion and transparent limits. Community governance protects collective rights, but it should not be misrepresented as proof that every member shares one view.

8. Changed-context transfer check: a regulatory inspection

Back to contents

Move from a grant evaluation to an inspection where the regulator has statutory authority and publication duties. The transferable safeguards are conflict disclosure, evidence independence, logged changes and clear reporting roles. The changed context may limit negotiated confidentiality and give the commissioner legitimate enforcement powers.

Do not copy a participatory evaluation protocol mechanically. State the legal duties, explain their limits to participants and keep analytic independence distinct from the regulator’s decision about sanctions.

9. Report constraints without performing false neutrality

Back to contents

Disclose who commissioned and paid for the work, team relationships, access restrictions, excluded questions, sample gaps, protocol changes and unresolved disagreements. Explain the mitigation used and what it cannot repair.

Transparency does not require pretending that every position is equally supported. The evaluator should make a reasoned judgement while distinguishing evidence, interpretation, stakeholder response and residual uncertainty.

10. Audit interference and independence over the full life cycle

Back to contents

Retain declarations, appointment criteria, terms of reference, question revisions, sample requests, access refusals, analytic decisions, review comments, payment milestones and publication versions. Review whether pressure accumulated at predictable points such as contract renewal or public launch.

Near misses are evidence. A proposed deletion that was later withdrawn still reveals where the process needs a stronger firewall. Record it without turning ordinary methodological debate into an allegation of misconduct.

11. Frequently asked questions

Back to contents

Must an evaluator have no prior relationship with the programme? No; relevant expertise may come with relationships, but they must be disclosed and managed. Can a funder comment on a draft? Yes, through a defined factual and methodological review. May a community request removal of harmful detail? Yes, under agreed governance and disclosure rules. Does publication prove independence? No; a published report can still reflect restricted questions, access or analysis.

12. The gold standard makes influence visible and answerable

Back to contents

Independent findings come from governed decisions, not from distance alone. Register interests, allocate authority, protect sampling and analysis, separate payment from conclusions, provide legitimate review routes and disclose what constrained the work.

The gold standard accepts that evaluation is relational. It prevents those relationships from operating invisibly, so useful challenge can improve the work without allowing money, access or status to decide what the evidence is permitted to say.