A Scatter Plot displays paired data points to reveal possible relationships, patterns, clusters, outliers, or nonlinearity between two variables.
Definition
A Scatter Plot is a graph of paired observations with one variable on the horizontal axis and another on the vertical axis. It helps teams visually assess whether two variables may be related, whether the relationship is linear or curved, and whether outliers or clusters exist.
It is a visual analysis tool, not proof of causation by itself.
History
Scatter plots are a long-standing statistical visualization method and one of the classic quality tools. They became popular in improvement work because many suspected causes can be screened quickly by plotting input-output relationships.
When to Use
Use a Scatter Plot when exploring whether an input may affect an output, checking relationships before regression, reviewing measurement agreement, identifying clusters, or communicating data patterns to a team.
Step-by-Step
- Define the two variables and why a relationship is suspected.
- Collect paired observations from the same units, times, or conditions.
- Plot the data with clear axes and units.
- Look for direction, strength, shape, clusters, and outliers.
- Stratify by product, shift, supplier, or condition if needed.
- Investigate outliers and special cases.
- Use correlation, regression, or experiments for deeper analysis.
- Confirm suspected causes with process evidence.
Examples
- Manufacturing: Plot oven temperature versus bond strength.
- Service: Plot number of handoffs versus cycle time.
- Maintenance: Plot operating hours versus vibration level.
Common Pitfalls
- Assuming correlation proves cause.
- Plotting unpaired or mismatched data.
- Ignoring clusters from mixed populations.
- Hiding time sequence when drift matters.
- Deleting outliers without cause.
- Using too few points to infer a pattern.