# Box and Violin Plot — Situational Evidence

Use to compare distributions across categories. Choose a box plot for compact robust summaries; choose a violin when distribution shape and sample size support density estimation.

## Required

- State or make available the box, whisker, and outlier definitions.
- Keep a shared quantitative scale across compared categories.
- Use the selected colourmap consistently by category or series.
- Show raw observations or sample size when a smooth violin could overstate certainty.

## Allowed choices

- Overlay lightly jittered points for small-to-medium samples.
- Use box plots when space is tight or sample sizes are small.
- Use violins when shape, multimodality, or skew is important and there are enough observations.
- Use half-violins or raincloud layouts when they improve comparison without crowding.

## Forbidden

- Do not use a violin for tiny samples that cannot support a meaningful density estimate.
- Do not hide extreme observations without disclosure.
- Do not vary violin width in a way that implies sample size unless width explicitly encodes it.
- Do not use different bandwidths across peer violins without a stated analytical reason.