Home →
Troubleshooting →
StatTools →
**Kurtosis Different from Excel's Computation**

**Applies to:**

StatTools, all releases

@RISK, all releases

StatTools calculated a kurtosis of 3.72 for my data set, but Excel's KURT( ) function calculated 0.72. Is StatTools wrong, or Excel?

They're both right, because they're measuring slightly different things. The standard computation of kurtosis gives a value of 3 for a normal distribution, and StatTools follows that computation. But you may want to know how your distribution's kurtosis compares to a normal distribution. Subtract 3, the kurtosis of a normal distribution, from the kurtosis of your distribution, and you have the **excess kurtosis**.

(excess kurtosis) = (kurtosis) – 3

If your distribution has a kurtosis of less than 3, it has a negative excess kurtosis and is called platykurtic; if your distribution has a kurtosis greater than 3, it has a positive excess kurtosis and is called leptokurtic.

Excel's KURT( ) function computes the excess kurtosis, not the kurtosis, and so it will always return a value 3 less than the StatTools value.

The above also applies to iteration data in @RISK. @RISK reports kurtosis for inputs and outputs. If you apply Excel's KURT( ) function to the iteration data for an input or output, you'll get a value 3 less than 2hat @RISK reported, because @RISK computes kurtosis and Excel computes excess kurtosis.

Last edited: 2016-03-17

This page was: Helpful |
Not Helpful