Evaluation begins before the outcome analysis
A common mistake is to jump directly to “Did it work?” before defining what the program actually did, who it reached, and whether implementation matched the plan. If delivery failed, a null outcome has a different meaning than if the intervention was delivered well and still had no effect.
CDC organizes evaluation as a sequence of decisions
The CDC’s 2024 framework includes assessing context, describing the program, focusing questions and design, gathering credible evidence, generating conclusions, and acting on findings. The sequence is practical because each step constrains what can responsibly be concluded later.
Process and outcome measures answer different questions
| Question | Example |
|---|---|
| Reach | Who actually participated? |
| Fidelity | Was the program delivered as intended? |
| Experience | How did participants and staff experience it? |
| Outcome | Did the target measure change? |
| Efficiency | What resources were required for the change? |
Evaluation and research can overlap without being identical
Both can use surveys, interviews, experiments, observational data, and statistical analysis. Research often seeks more generalizable knowledge; evaluation is frequently organized around decisions about a specific program, policy, or organization.
The distinction depends on purpose and context more than on whether the work “looks scientific.”
Negative findings can still be useful
A well-designed evaluation can reveal that a program reached the wrong audience, was too burdensome to deliver, produced an unintended effect, or had no detectable outcome. Those are actionable findings when the evaluation was designed to support decisions rather than to certify success.
Evaluation starts by making the program's causal logic explicit
Before measuring success, describe how the program is supposed to create it. Inputs support activities; activities produce immediate outputs; those outputs are expected to influence short- and long-term outcomes. A logic model or similar framework makes those assumptions visible. If the chain is vague, a disappointing result is hard to diagnose.
Different evaluation questions require different measures. Process evaluation asks whether the program reached the intended population and was delivered as planned. Outcome evaluation asks whether relevant outcomes changed. Economic evaluation can compare costs with consequences. Implementation measures can explain why the same program performs differently across settings.
Choose indicators that can change for the reason you care about
A convenient metric is not automatically a useful one. Attendance may show reach but not behavior change. Satisfaction may reflect acceptability but not effectiveness. A downstream health outcome may matter greatly but take too long to change for a short evaluation. A strong evaluation uses a set of measures that correspond to different points in the program's logic.
