Survey Sample Sizes and Quotas in the UAE: A Planning Guide

Use quotas deliberately

Quotas can help organise recruitment by agreed characteristics, such as location or user type. They do not, by themselves, establish that the sample represents the whole population. Consider the recruitment sources and the people those sources may miss.

Distinguish proportional quotas from deliberately boosted groups. Boosting a small audience may make its results more useful, but its share of the recruited sample is not necessarily its share of the market. Explain any weighting when producing combined figures, and retain the achieved unweighted bases.

Be careful with claims about precision

Do not attach a conventional random-sampling margin of error to an opt-in or convenience sample without an appropriate statistical basis and explanation. For probability samples, precision calculations must reflect the design and any relevant weighting effects. A familiar calculator cannot compensate for missing coverage or biased recruitment.

AAPOR distinguishes probability and nonprobability sampling and emphasises transparent methodology. Consult AAPOR disclosure standards when specifying the information that should accompany published survey findings.

What to request in a sample proposal

• A clear target population and eligibility definition.
• The recruitment sources and selection method.
• Proposed and achieved quotas by priority group.
• Expected usable bases for the main comparisons.
• Exclusions, validation rules and replacement procedures.
• The reporting approach for weighting and small subgroups.

A worked example: designing around a regional comparison

Consider a hypothetical brand that wants to compare recent users in Dubai and Abu Dhabi. It also wants a separate view of customers using a newly launched product. A single total of interviews does not answer whether all three comparisons will have an adequate basis. The sample plan should list the intended tables before allocating interviews.

Start with eligibility and the recruitment source. Determine whether the new-product group will occur naturally in the recruited audience or require a separate boost. If it is boosted, record those interviews distinctly. The report can show the group’s experience without assuming its recruited share equals its share of the market.

Review the proposed intersection of location and product group. It may be feasible to report the product group overall while avoiding a further location split. Making that decision before fieldwork helps prevent a report full of tiny cells. The minimum usable basis depends on the intended analysis and should be agreed with the analyst, rather than chosen as a universal rule.

This example uses hypothetical groups and does not recommend a fixed number of interviews. Its purpose is to connect sample allocation to the decisions the report must support. The final proposal should explain which comparisons are primary, which are exploratory and which will not be supported.

Incidence, screening and usable interviews

Incidence describes how frequently the target profile occurs within the people being approached or screened under a defined process. It can affect recruitment effort, but the figure depends on the recruitment source and the eligibility definition. An incidence assumption drawn from one audience should not be applied uncritically to another.

Distinguish screened people, eligible people, started interviews and usable completed interviews. A fieldwork plan should allow for the difference between those stages. Some people may decline, abandon the survey or fail a documented quality review. The required usable total should not be confused with the number of invitations sent.

When asking for a proposal, request the assumptions behind recruitment feasibility. A rare combination of role, experience and location may need more time than a broad consumer profile. If eligibility is relaxed to improve feasibility, record exactly what changed and whether the resulting audience still answers the original question.

Plan a pilot or initial recruitment check when the assumptions are uncertain. Use it to refine operational estimates while keeping the target population definition explicit. Do not quietly broaden the audience after launch merely to reach the headline number of completed interviews.

Weighting and the reporting base

Weighting changes the contribution of individual responses according to a defined analytical approach. It may be used to align selected characteristics or account for deliberate allocation choices. Explain the variables, benchmarks and method used. A weighted percentage should still be accompanied by the achieved unweighted respondent base.

Weighting cannot create evidence from a group that was not reached. It also cannot guarantee that people recruited through one source behave like people missing from that source. Review the coverage and selection process before interpreting a weighted result as representative. Strong claims require a defensible design and transparent explanation.

Agree how small cells will be reported. Options include suppressing a table, combining groups where substantively appropriate or labelling a result as exploratory. Keep the decision consistent across the report. Avoid displaying an exact-looking percentage without explaining that it rests on very few respondents.

A sample review meeting before launch

Ask the team to review a sample specification with four columns: target group, reason it matters, planned allocation and intended reporting use. The reason should relate to the business decision. If a group has no reporting purpose, decide whether it still needs a quota or can be recorded as a profiling variable instead.

Review the questionnaire routes against the allocation. A subgroup may have enough interviews overall but fewer eligible answers to a particular section. For example, only recent support users may answer resolution questions. The basis for that measure is the relevant route, not everyone recruited into the study.

Agree how replacements, exclusions and any boosts will be tracked. At the end of fieldwork, compare planned and achieved coverage before analysis starts. If the allocation changed materially, revise the reporting plan and disclose the limitation. The final report should describe the sample that was actually achieved, not only the sample originally proposed.

Frequently asked questions

Does a larger sample remove recruitment bias?

A larger sample can provide more responses, but it does not automatically fix missing coverage or biased selection. Review who had an opportunity to participate and how they were recruited. A large convenience sample may still systematically miss relevant people, so the methodology and claim need to remain aligned.

Are quotas the same as random selection?

No. Quotas organise the achieved profile by specified characteristics. Random selection describes how people are chosen under a probability design. A quota study may be useful for a research purpose, but reaching its quota targets does not by itself establish known selection probabilities or population representation.

Why do tables show different respondent bases?

Eligibility routes, missing answers and question-specific exclusions can change the base. A question about a recent purchase should use respondents who made one within the defined period. Every percentage needs a clear denominator so readers can understand which people contributed to that result.

When should an audience be deliberately boosted?

Consider a boost when a small but important group needs separate analysis and would otherwise have too few interviews. Keep the boost identifiable, explain its purpose and agree how combined results will be produced. The recruited share of a boosted audience should not be presented as its market prevalence.

Continue your research planning

• Quantitative research services
• UAE respondent recruitment
• Research cost planning

Cover image: AI-generated editorial illustration. It does not depict an actual client project.