Histograms, box plots, scatter plots, and stratification

Learning objectives

• Choose histogram, box, scatter or run chart appropriately.
• Stratify by plausible contextual factors.
• Distinguish association from causation.

One graph, one question

• Histogram — shape of one variable.
• Box plot — group comparison.
• Scatter — two quantitative variables.
• Run/control chart — time ordering.

Garment-factory example

A box plot of shrinkage by fabric lot shows one lot shifted; a scatter of curing temperature versus shrinkage suggests a positive relationship that needs a controlled test to confirm.

Method

1. Plot raw data first.
2. Label units and sample sizes.
3. Stratify one meaningful factor at a time.
4. Follow anomalies back to the process.

Common mistakes

• Connecting unordered categories with lines.
• Hiding time order in aggregated summaries.