Learn
Interactive·

Skewness & Kurtosis

Distributions with the same mean and standard deviation can still have different tail directions and tail thickness.

Income data often contain many middle values and a few very large ones, creating a long right tail. An easy exam may show the opposite pattern: scores pile up near the top while a tail stretches toward lower values.

The mean and standard deviation do not reveal which way a distribution is asymmetric or how much weight lies in the tails.

Key question

How can we summarize the direction of asymmetry and the thickness of a distribution's tails?

1

Concept

What does skewness describe?

Skewness indicates the direction of the longer tail

SkewnessTypical shapeExample
Near 0roughly symmetricheight or measurement error
Positivelong right tailincome or waiting time
Negativelong left tailscores on an easy exam

The sign of skewness refers to the direction of the tail, not the side where most observations are concentrated.

2

Why It Matters

Is kurtosis just peak height?

Kurtosis is more about tails and extreme deviations than peak height

Describing kurtosis only as “peakedness” can be misleading. Higher kurtosis more directly reflects heavier tails and a greater tendency for large deviations relative to a normal distribution.

Excess kurtosis

Excess kurtosis defines the normal distribution as 0. Some software reports ordinary kurtosis, where the normal distribution equals 3, so always check the convention.

3

How It Works

Why do measures of center move?

A long tail can pull the mean in the tail's direction

DistributionCommon pattern
Right-skewedmode < median < mean
Symmetric and unimodalmean ≈ median ≈ mode
Left-skewedmean < median < mode

Not an absolute law

These are common patterns for unimodal distributions. They may fail when a distribution has multiple peaks or a more complex shape.

4

Example

How should the numbers be read safely?

Do not diagnose a distribution from skewness and kurtosis alone

Financial returns

Even when the average return is near zero, high kurtosis may indicate that large gains or losses occur more often than a normal model would suggest.

  • In small samples, skewness and kurtosis estimates can be unstable.
  • A single outlier can change both statistics substantially.
  • Values near zero do not prove that a distribution is normal.
  • Inspect a histogram, boxplot, and Q–Q plot as well.
5

Interactive

Change the distribution shape

Change the distribution shape

Begin with a symmetric distribution and then change the skew direction. Next increase the tail-thickness setting and compare what changes in the center and in the extremes.

Skew setting

Symmetric

Tail emphasis

1.0

This is an illustrative shape control; the slider values are not computed sample skewness or kurtosis.

What to notice

Changing skewness shifts the tail direction and the relative position of the mean. Increasing kurtosis makes the tail behavior and extreme values more prominent relative to the center.

Key takeaways

  • Skewness describes the direction of the longer tail.
  • Kurtosis reflects tail thickness and the tendency for large deviations relative to a normal distribution.
  • The ordering of mean, median, and mode is a common pattern, not a universal rule.
  • Interpret skewness and kurtosis together with graphs and sample size.
Next: Frequency Distributions & Histograms