Skewness and Kurtosis Calculator

Paste a list of numbers, separated by commas, spaces or line breaks. The calculator updates instantly and reports all three skewness formulas, both kurtosis measures, standard errors and a skew gauge.

Your data

Kurtosis needs at least 4 values (n ≥ 4). Skewness needs at least 3 values. Constant data (all values the same) has a standard deviation of 0, so skewness and kurtosis are undefined.

Shape verdict

Nearly symmetric

Adjusted skewness G1 = 0.000 | Excess kurtosis (Excel) = -1.200

Moment skewness (g1)0.000
Adjusted skewness (G1)0.000
Pearson median skewness0.000
Excess kurtosis (Excel)-1.200
Moment excess kurtosis-1.224
SE of skewness0.687
SE of kurtosis1.334
Sample size (n)10
Mean11.000
Median11.000
Sample SD (s)6.055

This skewness calculator measures the shape of your data, including kurtosis. It reports moment skewness, the adjusted Fisher-Pearson measure used by Excel, and Pearson median skewness, plus excess kurtosis and standard errors, so you can see at a glance whether your data is symmetric, right-skewed or left-skewed, and how heavy its tails are.

Skew gauge

The marker shows your adjusted skewness (G1) on a scale from −2 (strongly left-skewed) to +2 (strongly right-skewed). Values near 0 sit in the symmetric middle.

−2 · Left-skewed0 · Symmetric+2 · Right-skewed

G1 = 0.00: the tails balance, so the data is nearly symmetric.

Calculation ledger — every step shown

  1. Count and sort: n = 10. Sorted data: 2, 4, 6, 8, 10, 12, 14, 16, 18, 20.
  2. Mean = 110 / 10 = 11.000. Median = 11.000.
  3. Sample SD s = 6.055.
  4. Moment skewness g1 = 0.000; adjusted G1 = 0.000; Pearson median skewness = 0.000.
  5. Excess kurtosis (Excel) = -1.200; moment excess kurtosis = -1.224.
  6. SE of skewness = 0.687; SE of kurtosis = 1.334.
Is my data normal enough for the empirical rule? Skewness and kurtosis are a quick shape check, not a verdict. If the gauge sits near 0 and the standard errors are large relative to the estimates, your data may still be close enough to normal. For a fuller answer, check the same dataset for normality on the same numbers.

What Does a Skewness of 0, Positive or Negative Mean?

A skewness of 0 means the data is symmetric: the left and right tails balance, and the mean and the median are equal. Positive skewness means the right tail is longer and the bulk of the values sits on the left, so the mean is pulled above the median. Negative skewness is the mirror image: the left tail is longer, most values sit on the right, and the mean sits below the median.

Adjusted skewness (G1)What it means
−0.5 to +0.5Approximately symmetric
+0.5 to +1 (or −0.5 to −1)Moderately skewed
Above +1 (or below −1)Highly skewed
Above +2 (or below −2)Very strongly skewed — check for outliers

These bands are rules of thumb, not hard cut-offs. Always compare the estimate with its standard error: if G1 is smaller than twice its standard error, the data is still consistent with a symmetric population. Because skewness depends on the spread of the data, it also helps to check your sample standard deviation with our standard deviation calculator before drawing conclusions.

How Do You Calculate Skewness and Kurtosis?

Three skewness formulas are in common use, and this calculator reports all three so you can see where they agree and where they differ. For the background behind the definitions, see what skewness and kurtosis are.

MeasureFormulaNotes
Moment skewness (g1)g1 = m3 / m2^(3/2)The ratio of the third and second central moments. The baseline definition.
Adjusted Fisher-Pearson (G1)G1 = √(n(n−1)) / (n−2) × g1The version Excel's SKEW function returns. It corrects the small-sample bias in g1.
Pearson median skewness3 × (mean − median) / sUses the median instead of the third moment, so one wild value moves it less.
Excess kurtosis (Excel KURT)KURT, with normal = 0Needs n ≥ 4. The value for a normal distribution is 0.
Moment excess kurtosism4 / m2^2 − 3The unadjusted version of the same idea.

All five measures work on raw values, not standardised ones. If you need standardised values first, use our z-score calculator.

Why the three skewness values can disagree

The formulas measure the same idea in different ways. The moment formulas (g1 and G1) cube every deviation, so a single large outlier can dominate them. Pearson's version only uses the mean, the median and the standard deviation, so it reacts less to extremes. G1 is always a little larger than g1 in absolute value for small samples — the correction factor √(n(n−1))/(n−2) shrinks towards 1 as n grows — so the three values converge for large datasets. If all three agree on the sign, the direction of skew is clear even when the numbers differ.

Worked examples

Symmetric data: 2, 4, 6, 8, 10, 12, 14, 16, 18, 20 (n = 10, mean = 11, median = 11). All three skewness measures are 0. Excess kurtosis is −1.20 by Excel's method (−1.22 by the moment method), and the standard error of skewness is 0.69. The flat, even spacing gives light tails.

Right-skewed data: 1, 2, 2, 3, 3, 3, 4, 5, 9 (n = 9, mean 3.556, median 3). Moment skewness g1 = 1.42, adjusted G1 = 1.73 and Pearson median skewness = 0.71. All three are positive, and the single 9 pulls the mean above the median.

Left-skewed data: 1, 5, 6, 7, 8, 8, 9, 9, 10 (n = 9, mean = 7, median = 8). Moment skewness g1 = −1.16, adjusted G1 = −1.41 and Pearson median skewness = −1.10. All three are negative, and the single 1 pulls the mean below the median.

Try each set with the example buttons above the results. For the spread behind these examples, see how to calculate standard deviation.

What Is Excess Kurtosis, and What Do Heavy or Light Tails Mean?

Kurtosis measures how heavy the tails of a distribution are and how peaked its centre is. Raw kurtosis for a normal distribution equals 3, which is an awkward reference point, so statisticians report excess kurtosis instead: raw kurtosis minus 3. On that scale, normal data scores 0.

Excess kurtosisShapeWhat it means
Above 0LeptokurticHeavy tails and a sharp peak; extreme values are more common than in a normal distribution.
Around 0MesokurticNormal-like tails.
Below 0PlatykurticLight tails and a flat peak; extreme values are rarer than in a normal distribution.

The symmetric example above scores −1.20, a platykurtic, flat shape — exactly what evenly spaced values produce. Treat kurtosis with care in small samples: its standard error is large when n is small, so a value of 0.8 from 10 observations is weak evidence of heavy tails.

Does Skewness Prove My Data Is Not Normal?

No. Skewness and kurtosis alone do not prove that data is not normal. They are useful warning signs, but a skewed sample can still come from a normal population, especially when the sample is small or contains one outlier.

Use the standard errors as a reality check: if the adjusted skewness is less than about twice its standard error, the skewness is consistent with normal data. Then run the same dataset through our normality check tool for a fuller answer — it combines several checks instead of relying on one number. If the data does pass as close enough to normal, the empirical rule probability finder and the normal distribution calculator are safe to use; if it fails, consider a transformation or a method that does not assume normality. You can also plot the shape with our bell curve generator to see the asymmetry, and if one value looks far from the rest, check it with our empirical rule outlier detector.

How to Use This Calculator

  1. Paste your data into the dataset box, separated by commas, spaces or line breaks. Use at least 3 values for skewness and at least 4 for kurtosis.
  2. Read the headline answer: the shape verdict and the adjusted skewness (G1). Positive is right-skewed, negative is left-skewed, and near 0 is nearly symmetric.
  3. Check the results grid for all three skewness measures, both kurtosis values and the standard errors, then read the skew gauge.
  4. Open the calculation ledger to see every step, and if the shape looks off, run the full normality check on the same data.

Frequently Asked Questions

What is skewness in simple terms?

Skewness measures how lopsided a dataset is. If the values pile up on the left with a long tail stretching right, the data is right-skewed; if they pile up on the right with a long tail to the left, it is left-skewed. A symmetric dataset has a skewness of 0.

What does a skewness of 0 mean?

A skewness of 0 means the data is symmetric around its centre: the mean equals the median and the two tails balance. It does not guarantee a normal distribution, though — evenly spaced values such as 2, 4, 6, 8 and 10 also score 0 but have flat, light tails.

What is a good skewness value?

As a rule of thumb, an adjusted skewness between −0.5 and +0.5 counts as approximately symmetric, and values up to about 1 are often acceptable for methods that assume normality. Above 1, and especially above 2, investigate the cause. Context matters: always compare the value with its standard error.

What is the difference between moment skewness and adjusted Fisher-Pearson skewness?

Moment skewness (g1) is the raw ratio of the third central moment to the second central moment raised to the power 3/2. Adjusted Fisher-Pearson skewness (G1) multiplies g1 by the correction factor square root of n(n−1) over (n−2). The correction removes small-sample bias and is the value Excel's SKEW function returns; the two agree closely for large samples.

What is excess kurtosis, and why is normal equal to 0?

Excess kurtosis is raw kurtosis minus 3. A normal distribution has a raw kurtosis of 3, so subtracting 3 puts normal at 0, heavy-tailed data above 0 and light-tailed data below 0. Excel's KURT function reports excess kurtosis, and so does this calculator.

Can skewness and kurtosis prove my data is not normal?

No. They are warning signs, not proof. Small samples can look skewed by chance, and one outlier can inflate both measures, so always check against the standard errors and run a dedicated check — see our normality-check tool — before concluding that data is not normal.

How many values do I need for kurtosis?

At least 4 values. Excel's kurtosis formula divides by (n−3), so with 3 or fewer values it has no answer, and this calculator shows a message instead. Skewness needs at least 3 values. In practice, treat kurtosis from fewer than 20 values with caution — its standard error is large.

Why do my three skewness values disagree slightly?

Because they are different formulas. The moment measures cube every deviation, Pearson's version uses only the mean, median and standard deviation, and the adjusted version corrects g1 for small samples. Small differences are expected; what matters most is whether all three agree on the sign and the rough size.

Reference: The definitions and formulas on this page follow the NIST/SEMATECH e-Handbook of Statistical Methods, section 1.3.5.11, Measures of Skewness and Kurtosis.

The bottom line: report all three skewness measures rather than one, always read them beside their standard errors, and remember that a single number cannot prove non-normality. When you want a second opinion on the same data, take the normality check, or browse all our statistical tools for the next step in your analysis.

More Tools to Explore

Continue with these related tools, or browse the full hub from the heading above.