Method isn’t an appendix. It’s the product.

1.1

No methodology is chosen before the decision it must inform is written down and agreed with the client. Method follows the question — never the reverse.

1.2

No methodological framework is imposed by a parent company, network or shareholder. Every protocol is built for the study at hand, not carried over because a global template requires it.

1.3

The methodological proposal states the method(s) chosen, their known limits, and what they will not allow you to conclude.

2.1

Any claim presented as national rests on a minimum sample of n = 384 respondents, the statistical threshold for a ±5% margin of error at a 95% confidence level.

2.2

No sub-group is published below n = 100 without a small-base warning printed directly on the chart — never relegated to a footnote.

2.3

The quota structure (age, gender, region, socio-professional category, or sector-specific criterion) is documented before fieldwork and shared with the client alongside the final report, quotas achieved included.

3.1

A self-administered online instrument does not exceed a 25-minute median duration, measured in the pilot before main fieldwork launches.

3.2

Every questionnaire is piloted (a minimum of 10 to 15 interviews) before main fieldwork launches; adjustments arising from the pilot are documented.

3.3

Any translation (Arabic, French, English, Darija) undergoes an independent back-translation before validation.

4.1

Respondent selection criteria are fixed before fieldwork opens and are not changed mid-collection without the client's written agreement.

4.2

Any off-quota respondent substitution is justified in writing and available on request.

4.3

The source of every respondent (proprietary panel, fieldwork partner, direct recruitment) is tracked and can be reported to the client by sample stratum.

5.1

No interviewer goes into the field without 3-day certification covering the questionnaire, neutral administration, and refusal scenarios.

5.2

A refresher session is held before every new wave of a repeated study, even when the fieldwork team is unchanged.

5.3

Certification is renewed annually; an interviewer without current certification is not assigned to MindMap fieldwork.

6.1

A minimum of 15% of each interviewer's interviews undergo an independent back-check, by a third party not involved in the original collection.

6.2

The back-check is carried out within 72 hours of the original interview, before the respondent's memory fades.

6.3

Any interviewer who fails a back-check has their entire fieldwork redone, at MindMap's cost — never billed to the client.

7.1

Any weighting applied is documented: calibration variable(s), source of the reference structure, minimum and maximum weights obtained.

7.2

The statistical significance threshold used (typically 95%, sometimes 90% for exploratory studies) is stated in the report, never implicit.

7.3

No sub-group is isolated after the fact to produce a more favourable result without that cut being flagged as exploratory.

Full policy at /en/methode/ia.

8.1

No synthetic respondent — a human simulated by a language model — is ever used to fill a quota, however difficult fieldwork proves.

8.2

AI is used for assisted, human-audited tasks: transcription, verbatim coding, first-draft translation. Never for the autonomous production of results.

9.1

Every deliverable carries its method, sample size, margin of error and fieldwork dates — on the document itself, not in a separate annex.

9.2

Raw data tables (top-line and cross-tabulated) are made available to the client on request, at no extra cost, for the duration of the contract.

9.3

Where a MOSAÏQUE benchmark exists for the category studied, the result is presented against that benchmark rather than in isolation.

Full policy at /en/methode/donnees.

10.1

Processing of personal data follows Law 09-08 and the requirements of the CNDP, from consent collection through to data destruction.

10.2

Identifying data is separated from response data as soon as fieldwork closes; analysis is only ever run on the anonymised file.

11.1

Respondent data is hosted in Morocco. Any transfer outside the country is subject to a specific contractual agreement and documented safeguards.

11.2

Access to respondent-level data is restricted to the project team; bulk export requires named authorisation.