# Scatter Plot — Situational Evidence

Use to show the relationship between two quantitative variables, including clusters, outliers, and model fit. Do not connect observations unless order or trajectory is real.

## Required

- Put one quantitative variable with units on each axis.
- Use the selected colourmap for categories, magnitude, or density; a single undifferentiated population uses one contrast-safe stop.
- Make overlapping observations visible through mark size, opacity, jitter, aggregation, or density—not invented separation.
- If a fit line is present, identify the model and distinguish observations from model output.
- Label or describe meaningful exclusions and transformations.

## Allowed choices

- Use 4–8px points for ordinary views, adjusted for density and final output size.
- Use transparency when overlap is material; verify the composited marks remain legible.
- Use shape in addition to color for a small number of critical categories.
- Use hexbin or density contours when raw points form an unreadable mass.

## Forbidden

- Do not imply causation from association.
- Do not add a trend line by default when the analytical question does not require one.
- Do not jitter quantitative values. Jitter is for revealing overlap on a discrete or rounded axis and must preserve interpretation.
- Do not label every point in a dense cloud. Label selected or consequential observations and provide details elsewhere.