Chapter 10. Quantitative and Survey Research

Sampling. Probability sampling requires a good sampling frame. In some settings, population registers or household lists may be outdated, incomplete or unavailable, so researchers may use multistage cluster sampling with census enumeration areas and field listing. Document how samples are drawn and any exclusions (for example conflict zones, mobile populations or informal settlements) [46][54].

Sample size and power. Determine sample size from the research question, expected effect size, variability and intended precision; account for design effects in cluster samples and expected non-response [61].

Survey instruments. Use validated scales where possible, but check validity in each language and culture. Translation should follow a rigorous procedure such as forward translation, independent back-translation and reconciliation, followed by cognitive pretesting [55][56].

Common measurement issues

Acquiescence (tendency to agree) and extreme or midpoint response styles, which differ across cultural groups.

Social desirability, which may be stronger for sensitive topics or when interviewers hold higher status.

Concept equivalence: terms such as “household,” “family,” “religion” or “income” may mean different things in different contexts.

Literacy: self-completion forms may be unsuitable; use interviewer administration [56].

Modes of data collection. Face-to-face interviews remain important in many settings; telephone and online modes have expanded, with unequal access by income, age, gender and region. Mixed-mode designs need care to avoid mode effects [54].

Enumerators. Recruit, train and supervise local enumerators carefully. Match language and, where appropriate, gender; provide fair pay and safety measures; check data quality through supervision and back-checks.

Data entry and electronic collection. Tablet-based collection improves accuracy but requires secure devices, backup and compliance with data protection laws.

Using secondary data. Large datasets from national surveys, censuses, regional barometer surveys and international programmes are valuable; read documentation on sampling, weighting and definitions.

Analysis and inference. Use appropriate methods for complex samples (weights, clustering), report uncertainty, and avoid over-interpreting small subgroups. Present results with clear descriptions of limitations [45][46].

Ethics in quantitative research. Obtain consent, protect confidentiality in small populations (small cell sizes can identify individuals) and share data responsibly.

Reflection: How will you ensure that each key survey concept means the same thing to respondents in each language?