Definition
Kurtosis is the standardized fourth central moment of a distribution. It measures standardized fourth-moment extremity; it does not identify peak shape. Some authors report raw kurtosis, where a normal distribution has kurtosis 3, and others report excess kurtosis, where normality is 0.
How it is calculated
Excess kurtosis gamma-two = expected fourth central moment divided by standard deviation to the fourth power, minus 3; normal excess kurtosis is zero.
| Symbol | Meaning | Unit |
|---|---|---|
| gamma | excess kurtosis | dimensionless |
| E | expectation | fourth-power X units |
| X | random variable | X units |
| mu | population mean | X units |
| sigma | population standard deviation | X units |
Interpretation
Kurtosis gives extra weight to observations far from the mean because deviations are raised to the fourth power. It can flag tail behavior that variance misses. Always check whether a quoted value is raw or excess kurtosis before comparing it with a benchmark.
Assumptions and conventions
-
The fourth central moment exists.
-
The population standard deviation satisfies sigma>0.
-
Raw versus excess kurtosis and any sample correction are stated.
Dimensionless.
Variants
- The population standard deviation satisfies sigma>0.
Limitations and common misreadings
-
High kurtosis is not just a statement about a sharp peak.
-
Normal kurtosis is 3 raw or 0 excess, depending on convention.
References
- Larry Wasserman. (2004). All of Statistics: A Concise Course in Statistical Inference. Springer Texts in Statistics. Chapter 3, Expectation. doi:10.1007/978-0-387-21736-9. Source

