Course contentsShow
← Data Science
Lesson 65 of 2,1452. Descriptive Statistics FoundationsFree lesson

Calculating Skewness

Learn the formula for skewness and how to compute it from sample data using standardized moments.

Calculating Skewness

What you'll learn: How to compute skewness using a standardized formula based on the third moment of your data.

Now that you understand what skewness means (asymmetry in a distribution), let's see how to actually calculate it from sample data.

The Skewness Formula

Skewness uses standardized moments—a way of measuring how far data points deviate from the mean, scaled by the spread. The sample skewness formula is:

Skewness = (n / ((n-1)(n-2))) × Σ((xᵢ - mean) / s)³

Where:

  • n = number of observations
  • xᵢ = each individual data point
  • mean = the arithmetic average
  • s = the standard deviation (sample version)
  • Σ = sum across all observations

Breaking Down the Calculation

Think of this formula in three layers:

  1. Cube the z-scores: For each value, subtract the mean, divide by standard deviation, then cube it. Cubing preserves direction (positive stays positive, negative stays negative) while emphasizing extreme values.

  2. Average those cubes: Sum them all up—this is your "third moment."

  3. Apply the correction factor: The n/((n-1)(n-2)) part adjusts for sample size, making the estimate more accurate for smaller datasets.

Why Cube?

Squaring (like in variance) loses information about direction—everything becomes positive. Cubing keeps the sign, so values below the mean contribute negatively and values above contribute positively. This lets us detect which tail is longer.

A distribution with a long right tail accumulates large positive cubed values, resulting in positive skewness. A long left tail does the opposite.

Key Takeaway: Skewness is calculated by cubing standardized deviations from the mean, then averaging them with a sample-size correction—giving you a single number that quantifies asymmetry in your distribution.