Make the purpose measurable.

Specify the intervention and question.

Record the artist, work, dimensions, material, location, sightlines, lighting, installation date, funding and selection process. State one primary question: for example, whether a sculpture helps visitors recognise a destination.

Choose a measure that fits the encounter.

For immediate experience, consider a brief question about the space, observed behaviour, or an appropriate measure of state anxiety. For longer-term staff wellbeing, use instruments with a recall period suited to that purpose.

WHO-5 and WEMWBS concern wellbeing over a period of time; they should not be interpreted as measures of the immediate effect of walking past a work. Check instrument permissions, validated translations, scoring and population suitability. Altering wording can alter validity.

Plan a credible comparison.

Where feasible, compare measurements before and after installation with a similar space that has not changed. Keep recruitment, times, locations and conditions consistent. Record concurrent changes in staffing, refurbishment, planting, waiting times or clinical activity.

For a causal research question, consider randomisation, repeated baseline measurements or another design with a research partner. A simple before-and-after difference cannot rule out other explanations.

Match the sample to the question.

There is no universal minimum of 30 people. A formal study needs a sample-size rationale based on its primary outcome and design. A small service evaluation can inform local decisions but should report its uncertainty and response rate.

Analyse patients, staff and visitors separately when their experiences and exposure differ. Record how people were recruited and who did not take part.

Arrange proportionate oversight.

Ask the institution to determine whether the work is service evaluation, quality improvement or research, and what review and consent processes apply. Protect patient information and avoid collecting identifiable clinical data without an agreed purpose.

Report the result in full.

State the intervention, methods, sample, comparison, timing, measures, missing data and interests. Report effect sizes and uncertainty where appropriate, including null and unfavourable findings. Distinguish perception, behaviour and clinical outcomes.