Research does not finish when the statistical software produces a table. The work is only partly complete until the findings are interpreted, communicated to the right people and connected to decisions that the evidence can reasonably support. A technically correct report may still have little impact if its main message is buried in jargon or if its recommendations do not reflect the realities of the people expected to act on them.
Begin with the decision, not the chart
Before preparing a report or presentation, ask who will use the findings and what decision they need to make. A county programme manager may need to know which service barriers are most urgent. A school leader may want to understand whether an intervention is reaching the intended students. A research audience may need detailed methods, uncertainty estimates and limitations to assess the strength of the evidence. These audiences can use the same study, but they need different levels of detail and different ways into the results.

Separate what the data show from what you recommend
A useful report makes a clear distinction between results, interpretation and action. The result might be that participants who received more support reported lower average distress. Interpretation considers whether the difference is large, precise and consistent with the design. A recommendation might be to test a support programme more systematically or to investigate which elements are most helpful. Keeping these steps separate prevents a plausible explanation from being presented as if it were directly proven by the data.
Statistical significance is not the same as practical importance
A small difference can be statistically significant in a very large sample but have little practical value. A larger difference may be uncertain when the sample is small. Decision-makers need to understand both the estimated size of an effect and how precisely it has been estimated. Explain confidence intervals in plain language, compare results with meaningful benchmarks where available, and consider costs, feasibility, equity and possible unintended consequences.
For example, if a new training programme is associated with a small improvement in scores, the next question is not only whether the p-value is below a threshold. Ask how much improvement occurred, whether it matters to participants, how long it lasts, whether it reaches different groups equally and what resources are needed to deliver it.
Make visualisations serve the message
Good charts help readers see a pattern without making them work through a page of numbers. Use bar charts to compare categories, line charts to show change over time and scatterplots to explore relationships between two numeric variables. Label axes clearly, state the units, show the relevant sample size where useful and avoid decorative effects that obscure comparisons. Use a sensible scale and never crop an axis in a way that exaggerates a difference without making the choice clear.
Tables are still valuable when readers need exact estimates, subgroup detail or confidence intervals. The goal is not to replace every table with a picture. It is to choose a format that makes the important information easy to understand while preserving the details needed to judge it.
Show uncertainty and limitations honestly
Every study has limits. The sample may not represent everyone, some participants may not have responded, measures may be imperfect, or the design may not establish cause and effect. A report that acknowledges these limits is more credible, not less. Explain what the study can answer, what it cannot answer and what additional evidence would help. Avoid burying limitations at the end if they materially change how the main finding should be understood.
Make recommendations specific and realistic
Recommendations are most useful when they name an action, a responsible person or team, a reasonable timeframe and a way to assess progress. “Improve awareness” is broad. “Pilot a short staff briefing in three sites over the next quarter and assess attendance, comprehension and referral patterns” is more actionable. Recommendations should follow from the evidence and should not promise that one study has solved a complex problem.
Where evidence is preliminary, recommend a pilot, further measurement or a carefully designed evaluation rather than immediate large-scale implementation. Where the evidence is strong, identify what is needed to implement it well and how outcomes should be monitored.
Bring the people affected by the findings into the conversation
Research is more likely to be useful when participants, community representatives, practitioners and decision-makers can respond to the interpretation. They may point out contextual factors that the dataset cannot show, identify practical barriers or help prioritise recommendations. Engagement should not mean changing results to satisfy stakeholders; it means explaining the evidence clearly and understanding how it fits the real setting.
Researchers should also think about returning findings to participants and communities in an accessible form when appropriate. A short summary, community meeting or plain-language brief can make the work more accountable than a technical report that only the research team reads.
Close the loop
Plan dissemination early rather than waiting until the last week of the project. Decide which audiences need a technical report, policy brief, presentation, infographic or community summary. Track whether recommendations are adopted and whether they lead to the expected results. Sometimes the most useful finding is that a programme did not work as expected; that knowledge can prevent wasted resources and guide a better next step.
The purpose of research is not simply to produce numbers. It is to create dependable evidence that helps people ask better questions, make more informed choices and recognise where uncertainty remains. That is how data begin to make a difference.
