Skip to content
MasterMath

Z-Score Calculator

A z-score says how many standard deviations a value sits from the mean, and from that, what percentile it occupies if the distribution is normal.

Z-score

—

Z-score—
Percentile—
Falls above it—
Falls below it—
Distance from the mean—
Reading—

How this was worked out

    The formula

    z = (x − μ) / σ

    What it means

    Subtracting the mean and dividing by the deviation puts any value on a common scale where zero is the mean and one unit is one standard deviation. That is what makes it possible to compare things measured in different ways: a 7 on a hard exam can beat a 9 on an easy one, and the z-score says so. If the distribution is also roughly normal, each z translates straight into a percentile.

    How to work it out by hand

    1. Subtract the mean from the value
    2. Divide that difference by the standard deviation
    3. The result is how many deviations separate you from the mean
    4. If the distribution is normal, look that z up in the table for the percentile

    What is worth knowing

    The percentile shown here assumes the distribution is normal, and that assumption has to be earned: for incomes, which are heavily skewed, a z of 2 is nowhere near the 97.7th percentile. The z-score itself assumes nothing and is always correct as a measure of distance; it is the translation into a percentile that needs normality. In a normal distribution, two thirds of the data fall between −1 and 1, 95 % between −2 and 2 and 99.7 % between −3 and 3.

    Frequently asked questions

    Is a negative z bad?

    It only means the value is below the mean. If you are measuring waiting times, below is good.

    What z counts as an outlier?

    By convention, more than 3 in absolute value: in a normal distribution that leaves out 0.3 % of cases.

    Do my data need to be normal?

    Not for the z-score itself. For the translation into a percentile, yes: that conversion assumes a normal distribution.

    Can I compare two different exams with it?

    That is exactly what it is for. It puts both marks on the same scale and makes them comparable.

    What if the deviation is zero?

    It cannot be worked out: every value would be identical and there is no spread to measure distance against.