Skewness & Kurtosis Calculator

Measure the asymmetry (skewness) and tail heaviness (excess kurtosis) of any dataset — with a histogram and a smooth shape overlay so the numbers have a visual anchor.

Shape statistics explained, step by step

This tool reports two sample moments — Fisher–Pearson adjusted skewness (G₁) and excess kurtosis (G₂). Each card below covers one piece of what the calculator actually computes.

1. Skewness — asymmetry of the distribution

Skewness measures how lopsided a distribution is. Symmetric data (like a normal curve) has skewness near 0. A long right tail gives positive skew; a long left tail gives negative skew. The tail pulls the mean toward itself, so mean vs median already hints at direction.

G₁ = n / ((n−1)(n−2)) · Σ((xᵢ − x̄)/s)³
nsample size·mean·ssample standard deviation
Example
Given
2, 4, 4, 5, 5, 5, 6, 6, 7, 9, 12, 18
Substitute
mean ≈ 6.92
median = 5.5
Answer
G₁ ≈ +1.55 (right-skewed)

2. Kurtosis — tail heaviness vs a normal curve

Kurtosis is about the tails, not the peak's height. Excess kurtosis subtracts 3 so a normal distribution reads 0. Positive (leptokurtic) means fat tails and frequent extreme values; negative (platykurtic) means light tails and values cluster more evenly.

G₂ = (n(n+1)) / ((n−1)(n−2)(n−3)) · Σ((xᵢ − x̄)/s)⁴ − 3(n−1)² / ((n−2)(n−3))
G₂sample excess kurtosis·ssample SD
Example
Given
n = 30
Σz⁴ ≈ 180
Substitute
G₂ = (30·31)/(29·28·27)·180 − 3·29²/(28·27)
Answer
G₂ ≈ 0.8 (mildly leptokurtic)

3. Plain-language classification

The calculator turns the raw numbers into a label so results are easier to read. Skew is grouped as approximately symmetric, moderately skewed, or highly skewed; kurtosis is grouped as platykurtic, mesokurtic, or leptokurtic.

|G₁| < 0.5 symmetric · 0.5–1 moderate · > 1 high
|G₁| < 0.5 → approximately symmetric
0.5 ≤ |G₁| ≤ 1 → moderately skewed
|G₁| > 1 → highly skewed
G₂ > 0 leptokurtic · G₂ ≈ 0 mesokurtic · G₂ < 0 platykurtic
Example
Given
G₁ = +1.55
G₂ = +2.1
Substitute
|1.55| > 1
G₂ > 0
Answer
highly right-skewed
leptokurtic

4. Sample-size caveats and outlier sensitivity

Deviations are cubed for skewness and raised to the fourth power for kurtosis, so a single outlier can dominate the result. With fewer than ~30 values the estimates are noisy — treat them as rough shape indicators, and always cross-check with the Outlier Detector when values look extreme.

need n ≥ 3 for G₁
n ≥ 4 for G₂ · trust weakens below n ≈ 30
Example
Given
9 typical values + 1 outlier
Substitute
outlier contributes ≈ 80% of Σ deviation⁴
Answer
kurtosis inflated by outlier

Features of this calculator

  • Sample skewness using the Fisher–Pearson adjusted G₁ formula
  • Sample excess kurtosis (G₂) — normal distribution = 0
  • Automatic plain-language classification (symmetric, right/left-skewed, leptokurtic, platykurtic)
  • Histogram of your data with a smooth distribution-shape overlay
  • Full step-by-step working: mean, standard deviation, and formula substitution
  • Robust input parser — commas, spaces, tabs, currency symbols and thousand separators
  • Copy the summary or download the whole result panel as an image

Frequently asked questions

+What does 'excess' kurtosis mean?

Kurtosis is often reported relative to the normal distribution, whose kurtosis value is 3. Excess kurtosis subtracts that 3, so a normal distribution has excess kurtosis of 0. Positive means heavier tails and a sharper peak than normal (leptokurtic); negative means lighter tails and a flatter peak (platykurtic). This calculator reports excess kurtosis.

+How does skewness relate to outliers?

Skewness depends on the cube of each deviation, so a single extreme value on one side can push skewness sharply positive or negative. If a dataset has strong skew, check for outliers with the Outlier Detector before drawing conclusions.

+What sample size do I need?

Skewness needs at least 3 values and excess kurtosis needs at least 4, but both estimates are unstable with small samples. With fewer than ~30 values the numbers swing wildly from sample to sample — treat them as rough shape indicators, not precise population parameters.

+How do I interpret the skewness value?

Rough guide: |skew| < 0.5 is approximately symmetric, 0.5–1 is moderately skewed, and above 1 is highly skewed. Positive = long right tail (mean > median); negative = long left tail.

Related calculators