At a Glance
| Baseline | Midline | Endline | |
|---|---|---|---|
| When | Before or at the very start of the program | Midway through (typically 12-18 months in) | At or near the end of the program |
| Purpose | Establish starting values for all indicators | Check progress, enable course corrections | Measure final results against baseline |
| Required? | Almost always (donors expect it) | Optional (recommended for 3+ year programs) | Almost always |
| Typical cost | 15-25% of total M&E budget | 10-15% of M&E budget | 15-25% of M&E budget |
| Who uses it | Program team (to set targets), donors (to approve indicators) | Program team (to adjust), donors (to track progress) | Everyone (final reporting, evaluation, learning) |
What Each One Collects
All three surveys should measure the same indicators using the same methodology. This is critical. If your baseline uses face-to-face interviews and your endline uses phone surveys, you cannot compare the results with confidence. Lock down your survey design before the baseline and do not change it.
Baseline: Collects current values for every outcome and output indicator in your logframe or MEL plan. Also captures contextual information (demographics, livelihoods, access to services) that helps you understand your target population and interpret later results. Baselines are not only surveys. Qualitative baseline data from focus groups or interviews provides context that surveys miss, such as community power dynamics, perceived barriers, and local definitions of well-being.
Midline: Collects the same indicator values as the baseline to check progress. The midline is your chance to identify problems early: if indicators are not moving, you can adjust your approach before it is too late. A good midline also captures implementation data (who has been reached, what activities have been delivered) to explain why results are or are not emerging.
Endline: Collects final values for all indicators. The difference between baseline and endline values is your program's measured result. The endline should also capture unintended effects (positive and negative) and sustainability indicators.
Baseline Timing
Aim to complete baseline data collection before program activities begin, or within the first four to six weeks at most. Every week of delay risks contaminating your starting values with early program effects. Build baseline design and data collection into your project inception plan so it does not get pushed back by procurement delays or slow staff recruitment. If field conditions make early collection impossible, document exactly which activities had already started and in which locations, so you can account for this when comparing to the endline.
When You Need a Midline
A midline is worth the investment when:
- Program duration is 3+ years. Waiting until the endline to discover problems wastes years.
- The donor requires it. Some contracts mandate midline data collection.
- You are implementing adaptive management. The midline provides evidence for mid-course corrections.
- Early indicators suggest problems. If routine monitoring data is flat or declining, a midline can diagnose why.
A midline is likely not worth it when:
- Program is under 2 years. By the time you collect, analyze, and act on midline data, the program may be nearly over.
- Budget is constrained. A strong baseline and endline is more valuable than a mediocre baseline, midline, and endline.
- Routine monitoring data is sufficient. If your regular monitoring system tracks key indicators quarterly, a formal midline survey may be redundant.
Panel vs Repeated Cross-Section
A key design decision: do you survey the same people at baseline and endline, or draw a new random sample each time? Use the sampling calculator to determine the right sample size for either approach.
| Panel (same people) | Repeated Cross-Section (new sample) | |
|---|---|---|
| Advantage | Measures individual-level change; smaller sample needed for same statistical power | No attrition problem; simpler logistics |
| Disadvantage | People move, die, refuse (attrition); tracking costs money | Cannot measure individual change; larger sample needed |
| Best for | Programs under 3 years, low-mobility populations, when you need to show individual change | Programs over 3 years, high-mobility populations, when population-level change is sufficient |
| Attrition risk | High if mobile population (aim for under 20% attrition) | None |
What To Do If You Missed Your Baseline
It happens. The program started before M&E was set up, or the baseline was planned but delayed past the point of usefulness.
Option 1: Retrospective baseline. Add recall questions to your next data collection: "Before the program started, did your household...?" This works for factual questions (did you have a latrine?) but is unreliable for subjective or numerical questions (what was your income 18 months ago?).
Option 2: Secondary data. Use existing data sources as proxy baseline values: national surveys (DHS, MICS), facility records, school enrollment data, government statistics. State the limitations clearly in your reporting: these sources may cover different time periods, geographies, or populations.
Option 3: Comparison group. If you can identify a similar population that has not received the program, their current values can serve as an estimate of what your target group looked like before the intervention.
Option 4: Delayed baseline with adjusted targets. If the program just started and activities have not yet reached most beneficiaries, a survey conducted in the first two to three months can still serve as a reasonable baseline for most indicators.
Whatever approach you use, document the limitation in your analysis and explain how it affects the confidence of your findings.
Common Mistakes
Mistake 1: Changing the methodology between surveys. If your baseline used household visits and your endline used phone surveys, any differences in results could be caused by the method change, not by your program. Use the same sampling approach, same questionnaire, and same data collection method.
Mistake 2: Baseline collected too late. If your baseline happens six months into the program, after activities have already started, your "baseline" values may already reflect early program effects. Collect baseline data before or during the first month of activities.
Mistake 3: No comparison group. Without a comparison group, you cannot distinguish program effects from external factors. If school attendance improved nationwide because of a government policy, your endline will show improvement that has nothing to do with your program.
Mistake 4: Endline collected too early. Some outcomes take time to materialize. If your program ends in December and you collect endline data in October, you may miss results that emerge in the final implementation period.
Mistake 5: Different sample sizes. If your baseline surveyed 500 households but your endline only reached 300 (due to budget cuts), the endline has less statistical power. Plan and budget for consistent sample sizes across all rounds.
Decision Guide
- Running a 2-year program? Skip the midline. Invest that budget in a stronger baseline and endline with adequate sample sizes.
- Program is 3+ years with adaptive management goals? Plan a midline at month 18. Allow three months for analysis and response before the next implementation cycle.
- Working with a mobile population (pastoralists, urban migrants)? Use a repeated cross-section design. Panel tracking in high-mobility settings leads to expensive attrition problems.
- Budget covers only two data collection rounds? Baseline and endline, always. A midline without a baseline is useless. A baseline without an endline is almost as bad.