The short answer
Program evaluation answers practical questions about a specific program: Is it being delivered as planned? Is it reaching the people it was designed for? Are they better off? What should change? It uses evidence, such as program records, surveys and interviews, rather than impressions, and it ends with findings people can act on.
Evaluation is not the same as research
Research aims to produce knowledge that applies beyond one setting, often through controlled studies. Evaluation serves a particular program and the people who run, fund and use it. That difference matters in practice:
- Evaluation works within real constraints. It uses the data a program can realistically collect, on the program's timeline and budget.
- It is judged by usefulness. A good evaluation helps someone make a decision, such as to expand, change or end a program.
- It is honest about limits. Most programs cannot prove that they caused an outcome, and a good evaluation says so plainly.
The main types of evaluation
- Formative or process evaluation looks at how a program is delivered: whether activities happen as planned, who takes part, and what helps or gets in the way. It is most useful early in a program, when there is still time to adjust.
- Outcome evaluation looks at what changes for participants or communities, such as knowledge, behavior, access to services or health status, over the short and medium term.
- Impact evaluation asks whether the program caused the changes observed. This usually needs a comparison group and a formal design, and it is closer to research.
Many evaluations combine process and outcome questions. A program that is not delivered as planned is unlikely to achieve its outcomes, so it helps to know both.
A recognized framework: CDC's six steps
In September 2024, the Centers for Disease Control and Prevention published an updated CDC Program Evaluation Framework. It is widely used in public health and offers a sensible order of work for any program. Its six steps are:
- Assess context. Understand the program's setting, the people involved and why the evaluation is needed.
- Describe the program. Set out its activities, the population it serves and the outcomes it expects, often in a logic model.
- Focus the evaluation questions and design. Choose the questions that matter most and the methods that can answer them.
- Gather credible evidence. Collect data that is good enough to trust for the decisions it will inform.
- Generate and support conclusions. Analyze the data and judge what it shows, with the limits stated.
- Act on findings. Share the results and use them to make decisions and improvements.
The framework also sets out five standards for good evaluation: relevance and utility, rigor, independence and objectivity, transparency, and ethics. And it describes three cross-cutting actions: engage collaboratively, advance equity, and learn from and use insights.
What a logic model is
A logic model is a one-page picture of how a program is supposed to work. It usually shows:
- Inputs: staff, funding, partners and materials.
- Activities: what the program does, such as home visits, classes or referrals.
- Outputs: the direct products of activities, such as the number of visits completed.
- Outcomes: the changes the program expects, in the short, medium and long term.
Writing a logic model with your team often reveals assumptions that nobody has tested, which makes it one of the most useful early steps in any evaluation.
Questions to answer before you start
- Who will use the findings, and for what decision? A funder renewal, a board decision and a staff improvement effort need different evaluations.
- What are the two or three most important questions? Fewer, sharper questions produce better evaluations than a long list.
- What data already exists? Program records and existing reports may answer more than you expect.
- What can you realistically collect? Consider staff time, participant burden, privacy and ethics approvals.
- When are the findings needed? Work back from grant deadlines and board meetings.
Example: turning a broad question into evaluation questions
| Broad question | Sharper evaluation questions |
|---|---|
| Is the program working? | Are participants completing the planned number of sessions? Do participants report better access to care after six months? |
| Is it reaching the right people? | What share of participants live in the target neighborhoods? Which groups are under-represented? |
| What should we change? | Which barriers do staff and participants report most often? |
Common mistakes to avoid
- Starting with a survey. Choose the questions first, then the methods.
- Measuring only activity. Counting sessions held says little about whether anyone is better off.
- Claiming too much. Changes among participants are not proof that the program caused them.
- Leaving staff out. Program staff know where the data is weak and which recommendations are practical.
- Writing for the shelf. Short, plain reports with clear recommendations get used.
Where to go next
If your program has not yet defined its population or priorities, read our guide on how to conduct a needs assessment. If you need help planning or carrying out an evaluation, see our program evaluation service.
Sources and further reading
Product capabilities and guidance change. These are the primary sources this article relies on, checked on the review date above.
- CDC Program Evaluation Framework, 2024, Centers for Disease Control and Prevention, MMWR Recommendations and Reports 73(6)
This article is general information, not legal, accounting or security advice for your specific situation. Examples are hypothetical unless stated otherwise.