A Histogram displays the distribution of numeric data so teams can see center, spread, shape, skew, multimodality, and potential outliers before making process decisions.
Definition
A Histogram is a bar chart for continuous or count data grouped into intervals called bins. It shows how often observations fall within each range, making the distribution of process output visible.
In Lean Six Sigma, histograms help teams understand variation before using capability analysis, hypothesis testing, control charts, or improvement experiments. They answer practical questions such as whether data are centered, spread widely, skewed, mixed from multiple sources, or affected by outliers.
History
Histograms come from statistical graphics and quality-control practice. They became one of the classic quality tools because they let non-statisticians see distribution shape quickly.
Quality practitioners use histograms to move beyond averages. Two processes can have the same mean but very different spread, tails, and customer risk.
When to Use
Use a histogram when analyzing cycle time, dimensions, strength, weight, defect counts, waiting time, transaction amounts, or any numeric process measure. It is useful during Measure and Analyze phases, before capability studies, and when comparing before-and-after improvement.
It is less useful for time-order behavior. If sequence matters, use a run chart or control chart alongside the histogram.
Step-by-Step
- Define the measure, unit, data source, and process scope.
- Collect enough representative numeric observations.
- Check measurement definitions and obvious data errors.
- Choose bins that reveal the shape without over-smoothing or over-fragmenting.
- Plot the frequency or percent of observations in each bin.
- Review center, spread, skew, gaps, tails, and multiple peaks.
- Stratify by shift, machine, product, supplier, customer, or time period if the shape suggests mixed populations.
- Connect the pattern to process knowledge and decide the next analysis or action.
Examples
- Machining: Diameter data show a centered but wide distribution that cannot meet specification consistently.
- Service: Order-entry cycle times show a long right tail caused by missing information.
- Healthcare: Patient wait times reveal two peaks, one for scheduled visits and one for walk-ins.
- Supplier quality: Incoming material thickness shows a shift after a supplier tooling change.
Common Pitfalls
- Using too few data points to infer a distribution.
- Choosing misleading bin widths.
- Ignoring time sequence and special causes.
- Mixing unlike products, shifts, or machines without stratification.
- Assuming normality from a rough bell shape without checking context.
- Using the chart without comparing to customer requirements or specifications.