Utiliser des enquêtes d’opinion représentatives dans le Mécanisme Africain d’Évaluation par les Pairs (MAEP)

This research paper by Robert Mattes, published by the South African Institute of International Affairs, argues that representative public opinion surveys are essential for the African Peer Review Mechanism (APRM) to achieve a comprehensive national self-assessment. The author contrasts the limitations of 'participatory' public consultations with the precision of random probability sampling and provides a technical framework for conducting credible surveys in Africa, including sampling stages, cost estimates, and the necessity of avoiding 'head of household' bias.

• Public consultations intended to foster national ownership often fail to produce representative data because they rely on self-selection, attracting primarily the most educated and politically active citizens. In a 2006 Afrobarometer survey in South Africa, only 6% of respondents had heard of the APRM process, and only 2% had filled out a questionnaire.
• Representative accuracy depends on the method of selection rather than the total number of people contacted. Random probability sampling can provide accurate estimates with relatively small samples; for instance, a sample of 300 people can produce estimates with a margin of error of approximately five percentage points.
• Due to the law of diminishing returns, increasing sample size provides marginal gains in precision at a high cost. To reduce the margin of error by one percentage point, the sample size must be doubled. Consequently, most socio-political surveys use samples between 1,000 and 2,500 people, with a minimum of 1,200 recommended for national surveys to achieve a margin of error of plus or minus three percentage points.
• Credible representative surveys require specific prerequisites: the absence of widespread civil conflict to ensure field worker safety, and the availability of accurate, recent census data detailed enough to allow for 'probability proportional to size' (PPS) sampling and stratification by urban-rural, religious, or linguistic lines.
• A critical error in survey design is using the 'head of household' as the sole informant. Because household heads are typically older, male, and more economically active, this method produces biased results. For governance and democracy assessments, the individual citizen, not the household, must be the unit of analysis.
• The design of the APRM public opinion questionnaire should be balanced between what citizens have a right to express (such as perceptions of corruption and leadership) and what they cannot realistically inform on (such as technical corporate governance). It is not necessary for ordinary citizens to answer every question in the official APRM instrument.
• Executing a credible national citizen survey in Africa requires significant planning and budget. The author estimates a timeline of five to six months from the decision to launch to receiving results, with costs ranging between 85,000 and 125,000 dollars for 1,200 respondents, depending on the country and provider.

APA

Mattes, R. (2008). Utiliser des enquêtes d’opinion représentatives dans le Mécanisme Africain d’Évaluation par les Pairs (MAEP). South African Institute of International Affairs. https://saiia.org.za/wp-content/uploads/2008/10/Occasional-Paper-13-FR.pdf

Chicago

Mattes, Robert. Utiliser des enquêtes d’opinion représentatives dans le Mécanisme Africain d’Évaluation par les Pairs (MAEP). South African Institute of International Affairs, 2008. https://saiia.org.za/wp-content/uploads/2008/10/Occasional-Paper-13-FR.pdf.

Wikipedia

{{cite report |last1=Mattes |first1=Robert |title=Utiliser des enquêtes d’opinion représentatives dans le Mécanisme Africain d’Évaluation par les Pairs (MAEP) |publisher=South African Institute of International Affairs |date=August 2008 |url=https://saiia.org.za/wp-content/uploads/2008/10/Occasional-Paper-13-FR.pdf |access-date=17 August 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{mattes2008utiliser, author = {Mattes, Robert}, title = {{Utiliser des enquêtes d’opinion représentatives dans le Mécanisme Africain d’Évaluation par les Pairs (MAEP)}}, institution = {South African Institute of International Affairs}, year = {2008}, month = aug, url = {https://saiia.org.za/wp-content/uploads/2008/10/Occasional-Paper-13-FR.pdf}, urldate = {2026-08-17}, note = {Indexed by Climate Insights Directory} }