Overview
Skewness measures how lopsided a distribution is. Positive skew has a long right tail, as with incomes, and negative skew has a long left tail.
Enter at least three numbers. With only a few values the result is a rough guide, so use more data for any real conclusion.
Worked examples
| Inputs | Result |
|---|---|
| Your numbers12, 16, 14, 21, 19 | Sample skewness (G1): 0.1361Moment skewness (g1): 0.0913Pearson median skewness: 0.329Shape: roughly symmetric |
Show 2 more examples
| Your numbers2, 3, 3, 8, 20 | Sample skewness (G1): 1.7653Moment skewness (g1): 1.1842Pearson median skewness: 1.6733Shape: highly skewed to the right |
| Your numbers50, 48, 47, 45, 10, 49, 51, 46 | Sample skewness (G1): -2.7066Moment skewness (g1): -2.1701Pearson median skewness: -0.9387Shape: highly skewed to the left |
How it works
Moment skewness (g1) is the average cubed distance from the mean, scaled by the standard deviation cubed. Sample skewness (G1) adjusts it for small samples, and matches Excel’s SKEW function. Pearson’s median skewness compares the mean with the median instead. As a rule of thumb, under 0.5 in size is roughly symmetric, 0.5 to 1 is moderate, and over 1 is high.
The formula
| g1 | Moment skewness: mean cubed deviation ÷ SD³ |
| n | Number of values |
Practical tips
- 2, 3, 3, 8 and 20 have a sample skewness of 1.77, highly skewed to the right.
- If the mean is well above the median, the data are usually right-skewed.
- One outlier can dominate skewness in a small sample, because deviations are cubed.
Frequently asked questions
Which skewness formula should I use?
G1 is the usual choice for samples, and it matches Excel. Check what your course or software uses.
Is zero skewness always normal?
No. It means the data are balanced, but they could still have an unusual shape.