HKSM

A histogram is a chart that shows how often numeric values fall within defined ranges (bins). It helps teams understand variation, shape, and central tendency in process data to support analysis and decisions.

Key Points

• Displays the distribution of a continuous or discrete numeric variable using adjacent bars that represent bins.
• Reveals shape (normal, skewed, bimodal), spread, center, and possible outliers.
• Bin width and number of bins strongly influence the story the chart tells.
• Requires enough data to be reliable; small samples can be misleading.
• Useful for quality, cycle time, cost, or risk impact frequency analysis.
• Not for categorical data; use a bar chart for categories.

What the Diagram Shows

• How observations are distributed across value ranges.
• Where most values cluster (mode) and the overall central tendency.
• Variability and range, including potential outliers or gaps.
• Skewness and whether the distribution is symmetric or not.
• Portion of values within targets or specification limits, if shown.

How to Construct

• Define the purpose and choose a single numeric variable (for example, cycle time, cost variance, defect size).
• Collect a consistent set of measurements from a defined time window or population.
• Clean the data: handle missing values, apply inclusion criteria, and note any outlier rules.
• Select binning (number and width) using a simple rule of thumb (for example, square-root of sample size) and adjust for readability.
• Compute bin edges, count how many observations fall into each bin, and choose frequency or percentage scale.
• Draw contiguous bars with equal widths, label axes and units, and optionally add vertical lines for targets or limits.
• Review the chart with stakeholders and refine binning if the picture is unclear.

Inputs Needed

• Raw numeric data for the variable of interest.
• Measurement units, definitions, and data collection period.
• Rules for binning (number of bins, bin width, handling of boundary values).
• Optional targets, specification limits, or service-level thresholds.
• Data quality criteria and outlier treatment approach.

Outputs Produced

• A histogram chart showing the distribution of the data set.
• Bin counts or relative frequencies for each interval.
• Basic descriptive statistics (min, max, mean, median, mode) to summarize the dataset.
• Insights about spread, skew, and potential multimodality to guide further analysis.
• Action cues, such as the need to stratify data, investigate outliers, or adjust process steps.

Interpretation Tips

• Right-skew often indicates occasional long delays or high values; left-skew suggests a lower tail.
• Multiple peaks can signal mixed populations; stratify by source, team, product, or shift.
• Use percentages when comparing histograms with different sample sizes.
• Too few bins can hide patterns; too many bins can create random noise. Try several settings.
• Bars should touch; if you need gaps between categories, you likely need a bar chart instead.
• Do not infer trends over time from a histogram; use a run or control chart for temporal behavior.

Example

A project team measures task completion times for 200 work items. They build a histogram with 12 bins and see a right-skewed shape with a long tail. Most tasks finish within the target, but a small portion takes much longer. The team stratifies by task type and finds the tail is driven by rework items, prompting a root cause analysis and a new review checkpoint.

Pitfalls

• Using categorical labels instead of numeric ranges, which turns the chart into a bar chart misuse.
• Choosing uneven or inappropriate bin widths that distort the visual story.
• Drawing conclusions from very small samples that are not representative.
• Mixing different populations without stratification, which can hide multimodal patterns.
• Comparing histograms with different sample sizes without normalizing to percentages.
• Reading a histogram as a time trend or as proof of process stability.

PMP Example Question

A team plots a histogram of cycle times and sees two distinct peaks. What should the project manager do next?

1. Increase the number of bins until the peaks disappear.
2. Stratify the data by relevant factors, such as task type or team, and analyze each group.
3. Replace the histogram with a scatter diagram to show trends over time.
4. Conclude the process is stable because the shape is consistent.

Correct Answer: B — Stratify the data by relevant factors, such as task type or team, and analyze each group.

Explanation: Two peaks often indicate mixed populations. Stratification helps isolate sources of variation so the team can investigate root causes.