A baseline study does more than produce a set of numbers before an NGO begins work. Done properly, it establishes what conditions look like, which communities face the greatest constraints and how progress will be measured fairly. Done poorly, it creates a misleading starting point that later reports struggle to correct. In Nigeria, where programmes can operate across very different locations, languages and service environments, that distinction matters.
Whether an organisation is preparing a livelihoods intervention in Kano, a maternal health project in Lagos or an education programme in rural Benue, the baseline should answer a practical question: what needs to be known now to make better programme decisions later?
Start with decisions, not questionnaires
Before drafting any survey questions, define the decisions the findings must inform. A programme team may need to establish the proportion of eligible households with access to safe drinking water, identify barriers to antenatal care or understand how youth currently obtain work-related skills. Each decision points towards different information and different respondents.
Translate the theory of change into a short measurement framework. For every expected outcome, specify an indicator, its definition, its numerator and denominator where relevant, the target population, the data source and the intended measurement frequency. Separate indicators that describe activities, such as number of training sessions, from those that describe outcomes, such as changes in knowledge or behaviour. A baseline is particularly valuable for the second category.
Define the study population and geography
A Nigeria-wide ambition does not automatically require a nationally representative survey. If an NGO serves selected wards in three local government areas, its baseline should usually represent the eligible population in those programme areas, not an abstract national population. Document the eligibility criteria, geographic boundaries and groups that may otherwise be overlooked.
Build a geographic listing of states, LGAs, wards, settlements and implementation sites, checking names and accessibility with local partners. Urban, peri-urban and rural populations may require different field procedures. Internally displaced people, residents of informal settlements and mobile populations need particular attention because a convenient household list can miss them altogether.
Choose a sample that supports the intended claims
Sample size is a design decision, not simply a budget line. It depends on the precision required, expected variation, subgroup comparisons, available sampling frame and clustering. If households are selected through communities or enumeration areas, account for the design effect. Allow for non-response without treating replacement interviews as a licence to select whoever is easiest to find.
A common approach is multistage sampling: select geographic clusters, list or verify households within them, then randomly select eligible households or individuals. Record the selection probabilities and plan for weighting where appropriate. If probability sampling is not feasible, use purposive or quota-based methods transparently and limit statistical claims accordingly. A sample of 500 people is not representative merely because 500 sounds substantial.
Combine the right methods
Quantitative surveys answer questions about prevalence and distribution. Qualitative methods explain processes, experiences and meanings that a questionnaire may miss. For example, a household survey may show low utilisation of a health service, while interviews reveal transport costs, inconvenient opening hours or fears about confidentiality.
Consider key informant interviews with service providers, focus group discussions with carefully selected participants, facility assessments and direct observation. Use existing administrative data, but verify definitions and completeness before treating them as baseline measures. Triangulation is useful when independent sources address the same question; agreement strengthens an interpretation, while disagreement signals a need to investigate.
Design questions around the local reality
A strong questionnaire sounds natural to respondents and measures one idea at a time. Avoid vague phrases such as 'regularly attends', 'adequate income' or 'improved access' unless they have precise operational definitions. Set recall periods that participants can reasonably remember. Asking about the last seven days may suit recent behaviour; asking about the last twelve months may introduce recall error for frequent small transactions.
Translate instruments into the languages genuinely needed in each study area and check conceptual equivalence through review and practice. A literal translation may preserve words while changing meaning. Pilot with people similar to the intended respondents, revise confusing skip patterns and assess the length of interviews. Sensitive subjects should never be added casually because a donor might find the data interesting.
Prepare field teams before data collection
Training should cover the study objectives, consent procedures, neutral probing, device use, sample selection, safety and realistic interview simulations. Supervisors need clear rules for unsuccessful visits, call-backs, refusals and escalation. Fieldwork planning should account for road conditions, local calendars, language coverage and security assessments, without assuming access is uniform across Nigeria.
For many household studies, computer-assisted personal interviewing can provide skip logic, validation checks and timestamps. Yet digital tools do not correct a bad question or an interviewer who prompts an answer. Build daily review procedures that examine completion rates, interview durations, missing values, GPS information where appropriate and unusually repetitive responses. Verify a proportion of interviews through ethically conducted back-checks without exposing respondents to unnecessary risk.
Protect participants and their data
Explain who is conducting the study, what participation involves, whether questions can be skipped and how information will be used. Participation should be voluntary, with special safeguarding arrangements where children or vulnerable groups are involved. Obtain the relevant ethics review or permissions where applicable; community permission does not replace individual informed consent.
Nigeria's Data Protection Act 2023 provides an important legal framework for handling personal data. Collect only necessary identifiers, define a lawful basis for processing, restrict access, encrypt storage where appropriate and set retention and deletion rules. Consent to a survey is not automatically permission to share names, photographs or recordings with every project partner.
Analyse against the measurement framework
Data cleaning should be planned before fieldwork ends. Check duplicates, outliers, inconsistent responses, coverage gaps and the treatment of missing observations. Apply weights and account for cluster design when producing estimates if the sample design requires them. Show denominators alongside percentages so readers can distinguish a percentage of all eligible respondents from a percentage of a small subgroup.
Disaggregate only where the sample can support meaningful analysis. Comparing outcomes by sex, age, disability status or geographic area can expose inequities, but tiny cells should not be portrayed as definitive differences. Interpret findings in the programme context and state the limitations plainly, including access restrictions, potential non-response bias and seasonal effects.
Turn findings into programme choices
A useful baseline report should contain an executive summary, clear methodology, indicator table, key findings, subgroup analysis, limitations and recommendations linked to actual management decisions. For instance, if communities report awareness of a service but limited physical access, another awareness campaign may be less useful than adjusting service locations or operating hours.
Agree in advance how baseline values will be preserved for endline comparison. Retain indicator definitions, question wording, sampling documentation and data dictionaries. If the intervention changes significantly, record the change rather than quietly redefining the original measure. Importantly, a before-and-after difference does not by itself prove programme impact; attribution requires an appropriate evaluation design.
Common mistakes NGOs can avoid
Three mistakes appear repeatedly in baseline work: collecting data without a decision framework, allowing convenient respondents to stand in for the target population and treating every numerical change as a programme effect. Other warning signs include unpiloted instruments, weak field supervision and reports filled with tables but little interpretation. Addressing these risks early is less costly than explaining them after the endline.
Conclusion
The strongest baseline is not necessarily the largest survey. It is the study that measures the right indicators in the right population, protects participants and produces findings programme teams can act on. In Nigeria, credible baseline research combines careful sampling, local knowledge, fieldwork discipline and honest analysis. That foundation makes future monitoring more useful and subsequent evaluation more defensible.
Surveysphere Africa supports organisations with baseline assessments, survey design, fieldwork and monitoring and evaluation research across African markets. For a discussion about an upcoming study, contact info@surveysphereafrica.com or visit surveysphereafrica.com.
Related guidance
For the continent-wide view, read our guide to baseline, midline and endline studies.
Our services for research in Nigeria
We run research across African markets. Start with these services for work in Nigeria:
- Monitoring, evaluation and learning in Nigeria: Baseline, midline, endline and programme evaluation.
- Social and development research in Nigeria: Qualitative and quantitative studies for social and development programmes.
We also offer: Outsourced fieldwork, Market and consumer research, Public health research, Policy, economic and institutional research, Data analytics and insights visualisation.
