Direct answer
A histogram groups observations into intervals to reveal shape; skewness summarizes asymmetry and kurtosis summarizes tail-weight and peakedness conventions, but all require context and none proves a distribution model.
Visual explanation
Bin choices change the visible shape
What this calculation tells you
Shape describes how observations occupy their scale: symmetric or asymmetric, concentrated or dispersed, unimodal or multimodal, light- or heavy-tailed. Graphics and numerical coefficients expose different parts of that structure.
Inspect several defensible bin widths and the raw sample size before interpreting a histogram. Use skewness and kurtosis as supporting summaries rather than labels that override the visible data.
Where it is used
Exploratory analysis
Inspect structure before choosing a model or transformation.
Quality
Identify mixtures, boundaries, and tail observations for investigation.
Finance
Describe historical return shape without equating it to future risk.
Education
Connect visual shape to numerical moments and their limitations.
Common situations
- Selecting a histogram bin count.
- Comparing symmetry across datasets.
- Explaining a skewness coefficient with a graph.
- Investigating whether a mixture may underlie multiple peaks.
Start with the statistical question
Inspect several defensible bin widths and the raw sample size before interpreting a histogram. Use skewness and kurtosis as supporting summaries rather than labels that override the visible data.
One dataset is redrawn with three bin widths, followed by symmetric, skewed, light-tailed, and heavy-tailed reference silhouettes. The same observations remain underneath every display.
Worked example
A small dataset with most values near 10 and one value at 50 may show positive skew. Moving one bin boundary can make a sparse gap appear or disappear even though the observations have not changed.
Assumptions that carry the result
Moment-based skewness and kurtosis are sensitive to extremes and unstable in small samples. Formula conventions include population moments and finite-sample adjustments; the convention must be named.
Interpret the result without overreaching
Shape summaries do not diagnose a generating mechanism, prove normality, or justify deleting tails. Sampling variation, mixtures, censoring, rounding, and measurement limits can all shape the display.
- Selecting one flattering bin width.
- Calling positive skew ‘most values are high’ without checking the axis.
- Interpreting high kurtosis only as a sharp peak while ignoring tails.
Practical questions
Frequently asked questions
What is the best number of histogram bins?
There is no universal best count. Use a defensible rule as a starting point and inspect nearby choices.
Does zero skewness mean normal?
No. Many non-normal distributions are symmetric and therefore have zero or near-zero skewness.
Does kurtosis measure only peakedness?
No. Modern interpretation emphasizes tail weight and the influence of extreme observations as well as central shape.
Further reading
Authoritative sources
Use these primary and professional resources to check definitions, conventions, or requirements that may extend beyond this guide.