Summary

  • An external report published on 19 August says AFRINIC launched a survey on its services, engagement, ecosystem role and future priorities, encouraged participants to share it, and said more responses would produce a more representative picture.
  • More replies can broaden feedback. They cannot, by themselves, show which member organisations or other stakeholders were covered, who selected themselves into the exercise, whether responses were unique, or how uneven participation was treated.
  • When results are released, AFRINIC should publish a compact methods receipt: separate target populations, recruitment channels, invitation or reach denominators, valid unweighted responses, exclusions, weighting, subgroup bases, limitations and a dated management response.

The launch promises input, not a completed mandate

The useful fact is modest. AFRINIC is asking members and stakeholders what works, what should improve and what deserves priority. According to the external launch report, it also wants the invitation circulated through colleagues and relevant networks. That can surface experience that tickets, meetings and elections miss.

Nothing in the reviewed evidence says the survey is a vote. Nothing shows that a result has been calculated or a priority adopted. The present object is an invitation. Its value lies in the observations it can collect, not in a representative status that has yet to be demonstrated.

The wording about response volume therefore deserves care. A larger number can reduce random variation in a properly described design. It can also add more people from the same highly engaged circles. Without a sampling or outreach frame, the reader cannot tell which happened.

One form can contain several populations

Members and stakeholders sounds like a single audience. It is at least two kinds of unit.

A member organisation may answer through one authorised representative, several employees or nobody. A network engineer, consultant, civil-society participant, researcher or other interested individual may answer in a personal capacity. Those responses can all be informative, but they do not share the same denominator and should not silently become interchangeable.

The first public table should therefore separate units before opinions. How many member organisations were in the relevant frame? How were their authorised or operational contacts reached? Did the analysis count organisations, people or both? Which other stakeholder groups were intentionally sought? If one organisation supplied several individual responses, were they retained as separate experiences, collapsed into one organisational observation or weighted in some other way?

These are not objections to openness. They are the information needed to understand what openness produced.

The methods receipt belongs beside the results

The AAPOR disclosure standards offer a practical checklist. For publicly released survey findings, they call for the population under study; the method used to generate and recruit the sample; modes, languages and collection dates; sample sizes; weighting; quality procedures; and acknowledged limitations. They also distinguish probability from non-probability recruitment and ask researchers to describe duplicate, bot and other validity controls where applicable.

AFRINIC need not adopt an American association's rulebook as law. It can use the disclosure logic because the questions travel well. A reader cannot evaluate a percentage without knowing the population, base and exclusions beneath it.

A compact AFRINIC receipt could use five linked blocks:

  1. Population: member organisations and each other analysed stakeholder population, defined separately.
  2. Reach: invitation lists or outreach channels, language availability, collection dates and any coverage gaps.
  3. Valid base: complete, partial, duplicate, ineligible and excluded responses, with the unit of response stated.
  4. Adjustment: unweighted and weighted bases, variables and benchmarks used, subgroup minimums and any precision model.
  5. Use: which finding informed which proposed action, who owns the response, and whether the action was accepted, deferred or rejected.

The receipt can remain aggregated. It need not expose a respondent, private comment or member contact list. Privacy protects people; it does not require the method to disappear.

A large count can be precisely wrong

Pew Research Center's study of online opt-in weighting is not a measurement of AFRINIC. It examined US samples, different questions and different providers. Its relevance is narrower and important: in the tested samples, increasing interviews from 2,000 to 8,000 barely improved accuracy, while bias remained. More observations tightened estimates around a result that could still be wrong.

That does not prove AFRINIC's survey will be biased. The reviewed record does not establish whether its design is probability-based, volunteer, census-style outreach to members or a mixture. The Pew result simply defeats one shortcut: response count alone cannot carry the word representative.

The stronger publication would use literal labels. Forty-two per cent of valid respondents is different from forty-two per cent of members. Most participating network operators is different from the African Internet community. Precision begins by refusing to enlarge the subject beyond the measured base.

Feedback becomes accountable when the response is traceable

The survey's practical value is not only in a chart. It is in the chain from finding to decision.

Suppose respondents ask for faster registration services, clearer transfer procedures or different training priorities. The result record should show the relevant base and then a management disposition: accepted, under analysis, deferred or not adopted; a reason; an owner; and a next date. Otherwise the institution can cite the survey when convenient and leave contrary findings without an outcome.

That disposition ledger also protects AFRINIC. It separates listening from promising. A widely supported request may be outside budget, policy or lawful authority. A smaller but operationally severe finding may deserve action. The methods record shows who spoke; the management record shows what the institution decided.

Participation is evidence before it is authority

Heng Lu's Multi-Stakeholder Mirage supplies the correct boundary: participation can be evidence, expertise, warning and objection without becoming authorization over absent parties. The same distinction makes the survey more useful, not less.

Responses should be treated as observations from identified analytical groups. They may reveal service defects, priorities and unfamiliar operating costs. They do not become the voice of a continent because an invitation was public or a response box was large.

AFRINIC can close that gap prospectively. Publish the questionnaire and methods with the findings. Name the denominators and the missing groups. Keep raw identities protected. Connect each material result to a dated institutional response. Then the survey will produce something stronger than the appearance of consultation: an inspectable body of evidence whose limits are as visible as its conclusions.

Sources

  • CADE, external report on the AFRINIC member and stakeholder survey launch
  • American Association for Public Opinion Research, Disclosure Standards
  • Pew Research Center, For Weighting Online Opt-In Samples, What Matters Most?
  • Heng Lu, The Multi-Stakeholder Mirage