Distinguish the claims.
| Evidence | What it can describe |
|---|---|
| Number of sessions or participants | An activity was delivered and people attended. |
| Materials or services produced | An output was created. |
| Participant satisfaction | Respondents’ experience of the activity. |
| A comparable learning score | Knowledge or skill observed by the chosen measure. |
| Later use of learning | Reported or otherwise observed application in practice. |
| Causal impact | A stronger claim requiring a design that supports attribution. |
Build indicators from the project question.
Specify the people, the expected change, the time frame and the measurement. “Participants use the new task at work within three months” is more informative than “the project is successful”.
Plan baseline and follow-up recruitment together. Keep wording and scales comparable. Decide what evidence comes from a survey and what requires another source.
Keep selection and missingness visible.
People who complete a follow-up may differ from those who do not. A high satisfaction rate among attendees does not describe people unable to attend. Report the usable base and likely gaps.
Weighting, an index or a confidence interval can support a defined analysis. None independently solves an unsuitable sample or supplies a causal design.
Use the report to support a decision.
Explain the finding, method and limitation beside the chart. In Nuvopoll, components from baseline, feedback and assessment studies can appear in one report without merging records.
Reader filters apply within their defined scope and do not recalculate model results. State which group each component describes. Finish with what the team will change or investigate, rather than only a positive headline.