Agriculture & Future Food Systems · Article
Review Effects and Revise the Plan
Compare a food-system trial with its stated goals, report unexpected effects and decide what the evidence supports doing next.
Preparation and related reading
Shared Foundations
Related articles
- Set a useful food-use goal and baseline
- Practise a food-flow record
- Compare food-use results and investigate causes
- Change one food-use step and review again
- Questions, Indicators and Causal Claims
- Household, Business and National Scales
- Survey, Administrative and Flow Data
- Trade-offs and Transparent Dashboards
This article helps you
- Distinguish a recorded change, target attainment and an intervention-caused effect.
- Read benefits beside resources, burdens and missing evidence.
- Revise a plan while preserving the original baseline and decision rules.
Begin with the goal that was actually set
A vegetable-box service tested evening collection. After the trial, the team wants to say whether it worked. Begin with the original goal, the intended people and the measurement rule rather than selecting whichever result looks most favourable. Check what the service actually did and whether the relevant records are complete.
A review connects the original plan with observed operation and outcomes. It can identify a useful improvement even when the target was missed, or a reached target alongside an unexpected burden. Neither achievement is a substitute for evidence on the other.
Change,target attainment and cause differ
A recorded change compares observations across the defined periods. Target attainment compares the later result with a previously chosen threshold. A causal claim asks how the outcome would have developed without the intervention. These are different comparisons and may require different evidence.
Keep the distinctions in the report. A later value can exceed the baseline yet fall below the target. It can also meet a target for reasons unrelated to the tested change. A useful operational review does not need to claim a causal effect when its design cannot establish one.
Report the achieved value without rewriting the target
In an invented follow-up, the same 80 eligible households and defined box are used for comparable weekly counts. The original baseline was 40 collecting households, or 50%. The target was at least 56, or 70%. The later observed result is 54, or 67.5%. The recorded share increased 17.5 percentage points but remained 2 households below the target.
These counts neither establish adequate diets nor isolate the evening-hours contribution. Weather, transport, eligibility and other conditions still need investigation. Retain the original target rather than lowering it retrospectively to make the result appear to have met the initial goal.
| Record | Households collecting | Share of 80 eligible households |
|---|---|---|
| Original baseline | 40 | 50% |
| Original target | At least 56 | At least 70% |
| Later observation | 54 | 67.5% |
| Result relative to target | 2 below threshold | 2.5 percentage points below |
An observed result can improve yet miss a target
Read the diagram as text
The invented later result is 54 of 80 households, against an original 56-household target and 40-household baseline. Improvement and target shortfall are separate comparisons.
Parts
- Target: 56 of 80
- Original proposed 70% collection threshold
- Observed: 54 of 80
- 67.5%; 14 more than the 40-household baseline
- Shortfall: 2 households
- 56 − 54; 2.5 percentage points on the same denominator
- Review causes and burdens
- Target comparison alone does not establish cause
Connections
- Target: 56 of 80 → Observed: 54 of 80: compare result
- Observed: 54 of 80 → Shortfall: 2 households: 56 − 54
- Shortfall: 2 households → Review causes and burdens: revise with evidence
Check burdens and missing outcomes
Suppose the trial also logged 14 additional session-hours rather than the planned 12. That is 2 hours more, approximately 16.7% above the planned opening time. It does not give a complete labour or cost overrun: staff numbers and preparation or other duties need their own account.
Read user and worker feedback alongside those records. Some households may gain access while another group still cannot attend. A process problem or unresolved safety evidence requires appropriate action, even if the collection count improves. Missing follow-up data should remain a limitation, not be classified as unchanged or favourable.
Make the next decision traceable
Identify which finding supports a revision. If transport remains a barrier, a new location or delivery option may be worth comparing. If staffing is difficult, consider the hours and available support. Record the altered scope, resources and expected mechanism instead of calling a larger unchanged trial an automatic solution.
Preserve the original baseline, target, methods and result. A revised goal is legitimate when its reason is explicit and future results use the revised version. It should not erase the earlier shortfall. Assign a next review point and explain which uncertainty the next test is meant to reduce.
State progress and uncertainty together
A clear invented-case conclusion is:'Collection increased from 40 to 54 of 80 households, below the 56-household target; the trial used more opening time than planned, and cause and wider access outcomes remain unresolved.' This states useful progress and limits without turning every observation into either success or failure.
Consider whether the conditions will hold in another season or site before expanding. Share the relevant findings with affected people and retain unexpected results. Shared system and uncertainty Foundations support the interpretation; neighbouring topics develop evaluation, resources and transition planning. Review is a continuing part of improvement, not a required course exercise.
Worked examples
Compare all three values
What does the invented 54-of-80 result show against baseline 40 and target 56?
- Compute 54/80=67.5%, compared with 40/80=50%.
- The recorded increase is 54−40=14 households, or 17.5 percentage points.
- The target shortfall is 56−54=2 households, or 2.5 percentage points.
There was a recorded increase, but the initial target was not met. The arithmetic does not isolate the intervention's effect.
Read the opening-time difference
What does 14 session-hours versus the planned 12 establish?
- Subtract 14−12=2 additional session-hours.
- Divide 2/12×100≈16.7% above planned opening time.
- Keep staff counts, other tasks, cost and causal explanation separate.
The logged opening time exceeded the plan by 2 hours. It is not a complete person-hours, cost or performance account.
Check your understanding
Should the original target be silently changed from 56 to 54 after seeing the result?
Show answer and reasoning
No.
Keep the earlier rule and result. A future revision needs an explicit reason and version.
Does an observed increase automatically prove that evening collection caused it?
Show answer and reasoning
No.
A causal conclusion needs an appropriate counterfactual and supporting assumptions.
Can a target shortfall still provide useful information for a revised plan?
Show answer and reasoning
Yes.
Observed operation, barriers and resource needs can identify the next useful investigation.
Key ideas
- Compare observations with the original goal and definition.
- Keep recorded change, target attainment and causal effect separate.
- Review resources, affected groups and unexpected outcomes together.
- Preserve prior evidence when revising a future plan.
References
Sources supporting the explanations and examples in this article.
- Impact Evaluation in Practice, second edition — selecting indicators and causal comparisonsPaul J. Gertler; Sebastian Martinez; Patrick Premand; Laura B. Rawlings; Christel M. J. VermeerschTextbook chapter · PDF
Food and Agriculture Organization of the United Nations2 sources