How to use this skewness calculator
Paste your dataset above and press Calculate. You get the skewness coefficient with an interpretation, alongside the mean and median whose disagreement is the skewness made visible.
What skewness measures
Skewness measures lopsidedness. A symmetric dataset (a bell curve, exam marks in a well-set test) balances around its centre and has skewness near 0. A dataset with a long tail of unusually high values (household incomes, city sizes, YouTube view counts) is right-skewed and scores positive; a long tail of low values (ages at retirement, scores on an easy test) is left-skewed and scores negative.
Eight monthly salaries (thousands): 22, 25, 28, 30, 32, 35, 38, 85. Seven ordinary salaries and one executive. Skewness ≈ +1.98: strongly right-skewed. The mean (36.9) is dragged far above the median (31) by the single 85, which is exactly the distortion the number is reporting.
Reading the coefficient
- −0.5 to +0.5: roughly symmetric.
- 0.5 to 1 (or −1 to −0.5): moderately skewed.
- Beyond ±1: strongly skewed. Consider the median and IQR as your headline summary, and think twice before using tests that assume symmetry.
Frequently asked questions
What does a skewness of 0 mean?
The data is symmetric around its centre: the left and right sides mirror each other, and mean and median coincide (or nearly).
What is the difference between positive and negative skew?
Positive (right) skew has a long tail of high values, like incomes. Negative (left) skew has a long tail of low values, like scores on an easy exam.
Does skewness affect which average I should report?
Yes. Strong skew pulls the mean toward the tail; the median then represents a typical value more honestly. Report both when |skewness| exceeds about 1.