Why use this calculator
A raw number means nothing without its distribution. Eighty-five is excellent on one test and mediocre on another, and the only way to compare them is to express each as a distance from its own mean measured in its own standard deviations. That is exactly what a z-score does, which is why it underlies grading curves, quality control limits, clinical reference ranges and every standardised test score.
This calculator gives the z-score and then does the part people usually look up in a printed table: the area under the curve to either side, the percentile, and the p-values. It also states plainly whether the result clears the conventional five per cent threshold, so the significance question is answered rather than left as an exercise.
How to use it
- Enter the single observation you want to place, such as one exam mark, one measured part or one patient reading.
- Enter the mean of the distribution it came from, which should be the published or historical average for that population rather than the average of a handful of nearby values.
- Enter the standard deviation of that same population; if you only have raw numbers, run them through the standard deviation calculator first and bring the result back here.
- Keep all three in the same units, because a mean in centimetres with a value in millimetres produces a confident and completely wrong answer.
- Then read the percentile for position and the p-values for how unusual the observation is.
What each result means
- Z-score
- How many standard deviations the value sits above or below the mean; negative means below.
- Percentile
- The percentage of the distribution that falls below this value, assuming normality.
- Area to the left
- Probability that a randomly drawn value is smaller than yours, as a decimal from 0 to 1.
- Area to the right
- Probability that a randomly drawn value is larger than yours; the two areas sum to one.
- One-tailed p-value
- Chance of a result this extreme in one specified direction if the null hypothesis is true.
- Two-tailed p-value
- Chance of a result this far from the mean in either direction under the null hypothesis.
- Distance from the mean
- The gap between your value and the mean in the original units, not in standard deviations.
- Significant at the 5% level
- Whether the two-tailed test clears the conventional 0.05 threshold, needing a z of 1.96.
The formula
Z = (X − μ) / σ
- X
- the observed value
- μ
- the population mean
- σ
- the population standard deviation, all in the same units
The z-score itself is dimensionless because the units cancel.
The area to the left of z under the standard normal curve
Φ(z) = ½[1 + erf(z / √2)]
The one-tailed p-value is the smaller tail area, and the two-tailed p-value is 2 × [1 − Φ(|z|)]. Critical values are 1.645 for a one-tailed 5% test and 1.96 for a two-tailed 5% test.
How this number is used in practice
Quality control turns z-scores into limits on a wall chart. A filling line with a target of 500 g and a standard deviation of 2 g places control limits three standard deviations out, at 494 g and 506 g, because under a normal distribution only about three parts in a thousand should fall outside on chance alone.
When a measurement produces a z-score beyond three, the operator stops the line and looks for a cause rather than shrugging at bad luck. The same logic sets clinical reference ranges: a bone density score is literally reported as a standard deviation count against a young adult reference, and the treatment threshold is a z-score, not a raw number.
In testing and analysis the z-score is what makes different scales comparable. Admissions offices standardise marks from schools with different grading habits before combining them, and analysts standardise returns before comparing assets with different volatilities. The important caution is what a small p-value licenses you to say.
It is the probability of data this extreme if the null hypothesis were true; it is not the probability that the null hypothesis is true, and it says nothing about how big or important the effect is. A trivial difference measured on a huge sample produces a tiny p-value, which is why effect sizes and confidence intervals are reported alongside it.
Worked examples
A score of 85 where the mean is 75 and the standard deviation is 10
Z-score of 1.0000, sitting at the 84.13th percentile.
An IQ of 130 against a mean of 100 and standard deviation of 15
Z-score of 2.0000, the 97.72nd percentile, two-tailed p-value 0.0455.
A reading of 68 where the mean is 72 with a standard deviation of 8
Z-score of −0.5000, which is the 30.85th percentile.
When people use it
- Comparing marks from tests that use different scales or different graders
- Setting statistical process control limits at two or three standard deviations
- Deciding whether a single measurement is unusual enough to investigate
- Converting a measurement into the percentile a report needs to quote
- Reading off the critical values for a one-tailed or two-tailed hypothesis test
Tips
- A z-score near zero is ordinary. Beyond two it is uncommon, and beyond three it is rare enough to be worth investigating.
- Use the population standard deviation. If you estimated it from a small sample, the t-distribution is the correct reference, not the normal one.
- Percentiles only mean what you think if the distribution is roughly normal. Skewed data gives misleading percentiles.
- Decide one-tailed or two-tailed before you see the data. Switching afterwards to get under 0.05 is a well-known way to fool yourself.
- A p-value is not the probability that the hypothesis is false, and 0.049 is not meaningfully different from 0.051.
- Keep the value, mean and standard deviation in identical units. Mixing units is the most common cause of an absurd z-score.
What it assumes, and where it stops being accurate
- The underlying distribution is normal. For skewed or heavy-tailed data the percentile and p-value will be wrong even though the z-score is still computable.
- The mean and standard deviation are population parameters, not estimates from a small sample; with an estimated σ and a small n, use a t-score.
- The value is a single independent observation, not an average. For a sample mean, divide by the standard error rather than the standard deviation.
- The normal cumulative distribution is evaluated with a numerical approximation to the error function, accurate to better than one part in a million.
- Statistical significance is not practical importance, and this calculator cannot tell you whether an effect is large enough to matter.
Terms used on this page
- Standard score
- Another name for a z-score: a value expressed in standard deviations from the mean.
- Standard normal distribution
- The normal distribution with a mean of zero and a standard deviation of one.
- P-value
- The probability of data at least as extreme as observed, assuming the null hypothesis holds.
- Critical value
- The z-score a test statistic must exceed to reject the null hypothesis at a chosen level.
Questions people ask
- What does a z-score of 2 actually mean?
- That the value sits two standard deviations above the mean. In a normal distribution about 97.7 per cent of values fall below it, so only around 2.3 per cent are as high or higher. It says nothing about why the value is high, only how unusual that position is within the distribution you supplied.
- When should I use a t-score instead?
- When the standard deviation is estimated from a small sample rather than known for the population. The t-distribution has heavier tails to allow for that extra uncertainty, and the difference matters most below about thirty observations. Above that, t and z converge closely enough that the normal approximation is acceptable for most work.
- What does the p-value tell me, and what does it not?
- It gives the probability of seeing data at least this extreme if the null hypothesis were true. It is not the probability that the null hypothesis is true, not the probability your result is a fluke, and not a measure of effect size. Always report a confidence interval or an effect size alongside it.
- Should I use the one-tailed or the two-tailed p-value?
- Use one-tailed only when a difference in the opposite direction would be meaningless or would lead to the same decision as no difference at all. Otherwise use two-tailed, which is the conservative default. The choice must be made before looking at the data, because a one-tailed test is easier to pass.
- Can a z-score be negative?
- Yes, and it simply means the value falls below the mean. The sign carries the direction while the magnitude carries how unusual it is, so −2.5 and +2.5 are equally extreme. Percentiles differ though: −2.5 sits at roughly the 0.6th percentile and +2.5 at about the 99.4th.
- Why does my percentile disagree with a printed z-table?
- Printed tables usually round z to two decimal places and the area to four, so small differences appear. Some tables also list only the area between zero and z rather than the whole area to the left, which changes the reading by exactly 0.5. Check the caption on the table before comparing.
Use this with
Z-Score Calculator answers one part of the question. These pick up where it stops: Standard Deviation Calculator, Sample Size Calculator, Probability Calculator and Mean, Median, Mode Calculator.
Where these figures come from
This calculation is jurisdiction-neutral arithmetic. No country-specific rules are applied.
- NIST/SEMATECH e-Handbook of Statistical Methods. Reference treatment of descriptive statistics and hypothesis testing
- Australian Bureau of Statistics. Statistical language, survey methods and published standard errors
Z-Score Calculator produces an estimate for general information. It is not financial, tax, medical, legal or engineering advice. read the full disclaimer. Spotted a wrong number? Report it and it gets fixed. The form picks up this calculator and the inputs you used.