AGRI90093 Chap.8 Evaluation Purpose and Program Logic
Evaluation Purpose and Program Logic
Evaluation supports accountability, learning and improvement by testing a program's causal chain rather than only its final result. This chapter links inputs, activities, outputs, capacity, practice and higher outcomes, then places key evaluation questions and indicators at the transitions. It also treats intended use and feedback as part of design so evidence can change targeting, delivery or the theory of change.
What this chapter covers
- 01
Accountability, learning and improvement as distinct evaluation purposes
- 02
Inputs, activities, outputs, capacity, practice and higher outcomes
- 03
Assumptions and contextual conditions inside every causal arrow
- 04
Key evaluation questions compared with indicators and individual items
- 05
Baselines, timing, ethics and participant involvement in evaluation design
- 06
Feedback loops that connect findings to a named decision and redesign
Worked example · free
Evaluate a field day with advisory follow-up
- 1Use attendance and completed calls as output evidence without treating them as outcomes.
- 1Use a practical demonstration to test sampling and identification capacity.
- 1Use logs and interviews to examine routine use, adaptation and remaining constraints.
- 1Specify how repeated skill or access problems will change the next season's design.
Key terms
- Program Logic
- A causal account linking resources and activities to outputs, capacity, practice and higher outcomes.
- Key Evaluation Question
- A high-level inquiry that organises evidence, analysis and intended use.
- Indicator
- An observable sign used to assess performance, change or a condition in the program logic.
- Feedback Loop
- A process that returns evaluation evidence to targeting, delivery, resources or program theory.
Evaluation Purpose and Program Logic FAQ
Why evaluate more than the final outcome?
A final result cannot show where the program chain worked or broke. Evidence about delivery, capacity and practice can distinguish weak implementation from a poor mechanism or an external condition. That diagnosis supports improvement even when higher outcomes take longer to appear.
What is the difference between a key question and an indicator?
A key evaluation question is a high-level inquiry about implementation, mechanism, distribution or outcome. An indicator is a specific observable sign that contributes evidence. Survey items and interview prompts sit further below the question and should not replace it.
What makes evaluation evidence useful?
Useful evidence answers a named question at a proportionate level of effort and reaches a decision maker at the right time. A plan should state the review point, intended user and the finding or pattern that could change targeting, delivery, resources or the theory of change.
Assessment move
Draw the program logic from inputs to higher outcomes and place an assumption at every transition. Then write one key evaluation question for output, capacity, practice and outcome levels. Under each question, add an indicator, source and intended use. Practise distinguishing a satisfaction item from a question about mechanism or use. When results disagree, investigate the difference instead of averaging it away.
Close each evaluation design with the decision that evidence can change and the person or group who will act on it. Begin by clarifying evaluation purpose and intended users. Separate accountability, learning, program improvement and judgement because each purpose shapes timing, questions and evidence. Name the decision that the evaluation is expected to inform and when that decision occurs.
If no one can explain how a finding would be used, reconsider whether the proposed question justifies the burden of data collection. Draw the logic model with distinct columns for inputs, activities, outputs, capacities, practices and higher outcomes. Use arrows only when you can articulate the causal proposition connecting adjacent elements.
Write assumptions beside the arrows, including contextual conditions and contributions from other actors. Check that outputs describe immediate products or participation rather than smuggling in learning or behaviour change. Treat the model as a hypothesis that can be revised, not as proof that the program caused every later result. Develop key evaluation questions from the logic rather than from available survey items.
Include questions about implementation, reach and quality, the operation of proposed mechanisms, changes in capacity or practice and contribution to wider outcomes where feasible. Keep high-level questions broad enough to matter but specific enough to guide evidence choices. For each question, identify users, criteria for interpretation and the action a finding could trigger.
Make an evidence matrix with question, indicator, source, timing, responsibility, limitation and intended use. Combine sources when one method cannot capture both breadth and explanation. A satisfaction rating may reveal participant reaction, but it does not establish learning, use or impact.
Pair it with evidence suited to the claim, such as demonstrated capability, practice records, observation or qualitative accounts of how change occurred. Note ethical and access implications before deciding that more data is automatically better. Stress-test the logic with rival explanations and negative cases.
Ask what else could produce the observed result, which groups may experience a different pathway and where implementation varied. Plan how discrepant evidence will be examined rather than averaged away. Finish by scheduling a sense-making point where stakeholders review findings, revisit assumptions and agree on a response.
This keeps evaluation connected to learning and adaptation while avoiding claims stronger than the evidence permits.
Working through Evaluation Purpose and Program Logic in AGRI90093? Sia is AskSia’s AI Education tutor — ask any AGRI90093 Evaluation Purpose and Program Logic question and get a clear, step-by-step explanation grounded in how AGRI90093 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
