An existing dataset is a credible grant baseline only when it describes the right population, place, period, and outcome using a method your organization can explain. Being official, recent, or easy to access is not enough. The baseline must be fit for the decision and comparable with the results you plan to measure later.
Start with the comparison you need
A baseline gives the starting value against which change will be judged. Before reviewing a source, define the programme population, geography, indicator, reference period, and intended follow-up method. If the endline will measure a household outcome through a survey, a broad national administrative total may provide context but may not be a defensible baseline for programme participants.
Use a seven-part source test
- Relevance: Does the source measure the same concept as the proposed indicator?
- Population: Does it cover the people your programme will serve, including important exclusions?
- Geography: Is the data available at the level where implementation and change will occur?
- Timing: Is the reference period recent enough, and has a major shock made it obsolete?
- Method: Are sampling, instruments, definitions, and calculation rules documented?
- Quality: Are completeness, missing values, response rates, revisions, or known biases disclosed?
- Comparability: Can your organization use a sufficiently similar method at follow-up?
Record each answer as suitable, suitable with limitations, or unsuitable. A source with one limitation is not automatically unusable. The question is whether the limitation could change the programme decision or misrepresent progress.
Choose one of three baseline decisions
Use the source directly
Choose this only when the definition, population, geography, timing, and method align closely with the programme indicator. Keep the source file, citation, extraction date, table or page reference, and calculation used.
Use it with a documented limitation
The source may be good enough for planning but not precise enough for an attribution claim. State the gap clearly, use cautious target language, and plan a stronger measurement activity early in implementation.
Collect a new baseline
Do this when the existing source covers the wrong population, uses a materially different definition, is too old after a major change, or cannot support comparison with the planned endline. Budget the work and complete it before significant programme exposure changes the starting condition.
Worked example: current data, wrong population
A county health dataset is current and methodologically sound, but it excludes informal settlements where the project will operate. The team should not present the county average as the participants’ baseline. It can use the dataset to explain wider context, then collect a focused starting measure in the target settlements using the same outcome definition planned for follow-up.
Document the evidence trail
For every baseline value, retain the source owner, publication and reference period, URL or file, table or page, original unit, any transformation, reviewer, approval date, and limitations. If you calculate a rate from two source fields, save the numerator, denominator, and formula—not only the final percentage.
What to do next
Use the Grant Evidence Inventory to classify the source as available, partial, outdated, or missing. Download the free Africads Monitoring and Evaluation Plan (DOCX) to record baselines, targets, data sources, quality controls, reporting dates, and follow-up responsibilities.
Frequently asked questions
How recent must baseline data be?
There is no universal age limit. Judge whether economic, policy, demographic, environmental, or service changes have made the value unrepresentative.
Can national data be used for a local project?
It can provide context, but it is rarely a substitute when the programme population or local conditions differ materially from the national average.
Is an official source automatically reliable?
No. Official data can still have coverage, timing, definition, or comparability limitations. Review the methodology and intended use.
What if no credible baseline exists?
State the gap, avoid invented precision, budget early baseline collection, and use contextual data only for what it can genuinely support.
Can a baseline be revised later?
Yes, when better evidence reveals a material error or mismatch. Preserve the original, document the reason, obtain approval, and explain the effect on targets and trend comparison.
Conclusion
Baseline quality depends on fitness for use, not convenience. Test relevance, population, geography, timing, method, quality, and comparability. Then choose deliberately: use the source, use it with limitations, or collect a new baseline.
