Saved

Press the star next to a tool's name to save it.

All my tools

Empirical Rule Calculator

Use this empirical rule calculator to find the ranges that hold 68, 95 and 99.7 percent of normal data, Chebyshev's guaranteed share for any data, and a value's z-score.

Restored your last values ·

Enter the mean and the standard deviation of your data. You get the ranges that hold about 68 percent, 95 percent and 99.7 percent of values if the data is roughly normal. Add a number of standard deviations, k, to see the share for normal data and Chebyshev's guaranteed minimum for any data.

Everything is worked out in your browser. Nothing you type is sent anywhere.

–

–

Enter the mean and the standard deviation, then press Calculate.

See full result

An empirical rule calculator turns a mean and a standard deviation into the ranges that contain about 68, 95 and 99.7 percent of the values of a normal distribution. It also gives Chebyshev's theorem, which guarantees a minimum share for any data, whatever its shape. Enter the mean and the standard deviation, and optionally a number of standard deviations or a value to check. It runs in your browser.

How to use the empirical rule calculator

  1. Enter the mean and the standard deviation.
  2. Add a number of standard deviations, k, or a value if you want them checked.
  3. Press Calculate.

The 68–95–99.7 rule

The Wikipedia article on the 68–95–99.7 rule says that, for a normal distribution, the chance that a value lies within one standard deviation of the mean is about 68.27 percent, within two standard deviations about 95.45 percent, and within three about 99.73 percent.[1] It describes the rounded figures, 68, 95 and 99.7 percent, as approximations that are meant for a normal population.[1] With a mean of 100 and a standard deviation of 15, about 68 percent of values fall between 85 and 115, 95 percent between 70 and 130, and 99.7 percent between 55 and 145. The calculator gives these ranges.

What lies outside

The article's table of the complements gives the share beyond each range: 31.73 percent beyond one standard deviation, about 1 in 3, 4.55 percent beyond two, about 1 in 22, and 0.27 percent beyond three, about 1 in 370.[1] A value that is more than three standard deviations from the mean is therefore rare, which is why such values are often called outliers. The calculator reports these shares beside the ranges.

Chebyshev's theorem

The empirical rule needs a normal distribution. Chebyshev's inequality does not. The article states that the chance that a value is at least k standard deviations from the mean is at most 1 ÷ k², for any distribution whose mean and variance are defined, and that only the case k above 1 is useful.[2] So at least 1 − 1 ÷ k² of the values lie within k standard deviations. The page's table gives 75 percent for k = 2, 88.89 percent for k = 3 and 93.75 percent for k = 4.[2] The calculator works out the guarantee for any k that you enter, such as 55.56 percent for 1.5 or 99 percent for 10.

A worked example

The page's example is a journal whose articles have a mean of 1,000 words and a standard deviation of 200. Chebyshev's inequality says that at least 75 percent of the articles have between 600 and 1,400 words. If the lengths were normal, the page says that there would be a 75 percent chance that an article has between 770 and 1,230 words.[2] Enter a mean of 1,000, a standard deviation of 200 and a k of 1.15 to see this range, and the normal share, 74.99 percent. The comparison shows how much more the normal assumption tells you.

Checking a value

If you enter a value, the calculator gives its z-score, which is the number of standard deviations from the mean, and says which band it is in. A value of 130 with a mean of 100 and a standard deviation of 15 is 2 standard deviations above the mean. A z-score of 2 or more happens in about 2.3 percent of cases on each side of a normal distribution, so the value is unusual but not extreme. The z-score uses the simple formula (value − mean) ÷ standard deviation.

When the rule works

The article warns that the percentages hold for normal data, and that distributions with fat tails put fewer values within the three ranges.[1] Heights, measurement errors and many natural quantities are close to normal, and the rule works well for them. Incomes, waiting times and counts of rare events are not, so use Chebyshev's guarantee, or look at the actual data, for those.

Using the ranges

The ranges help to judge whether a value is normal or surprising, to set limits for quality control, and to check a result for sense. For a process with a mean of 50 and a standard deviation of 2, nearly all products should fall between 44 and 56, and a measurement of 60 is five standard deviations away, which signals a problem. They are also a quick way to describe the spread of data to someone without statistics training.

Choosing between the rule and the theorem

If you know that your data is close to a bell curve, use the empirical rule, because it gives a much tighter answer. If you do not know the shape, or the data is skewed, use Chebyshev's theorem, because it is always safe. A good habit is to calculate both. For k = 2, the normal share is 95.45 percent and the guaranteed share is 75 percent, so the difference shows how much the assumption of a normal shape is worth. If the real share is nearer to 75 percent, the data is far from normal.

Common mistakes

  • Using the rule for any data. It needs a roughly normal distribution.
  • Mixing up the percentages. 68 percent is within one standard deviation, not two.
  • Using the variance. Enter the standard deviation, which is its square root.
  • Expecting Chebyshev to be tight. It is a minimum, and the real share is often much higher.

What next

Use this often? Add Empirical Ru to your Home screen.

Frequently asked questions

What is the empirical rule?

For a normal distribution, about 68 percent of values lie within 1 standard deviation of the mean, 95 percent within 2 and 99.7 percent within 3.

What is Chebyshev's theorem?

For any distribution with a mean and variance, at least 1 − 1 ÷ k² of the values lie within k standard deviations. It gives 75 percent for k = 2.

What is an empirical rule calculator used for?

It finds the ranges around a mean that hold about 68, 95 and 99.7 percent of normal data, and Chebyshev's guaranteed share for any data.

What is the difference between the two?

The empirical rule is more precise but needs normal data. Chebyshev's theorem holds for any data but gives only a minimum.

How many values fall more than 3 standard deviations away?

About 0.27 percent for normal data, or 1 in 370.

Can I use it with a variance?

Take the square root first. The calculator needs the standard deviation.

What went wrong?

We only receive the reason you pick, never what you typed into the tool.

Thanks for your feedback.

More Math Tools