What Does a Z Score Determine?

Home » Blog » What Does a Z Score Determine?

A z score determines how many standard deviations a particular value lies above or below the mean of a distribution. It shows a value’s relative position, making it easier to compare observations measured on different scales or judge how unusual a value is under a statistical model.

What a z score tells you

A z score, also called a standard score, expresses the distance between a value and its mean in units of standard deviation. A positive score means the value is above the mean; a negative score means it is below; and zero means it equals the mean.

For example, a z score of 1.5 says that the observation is 1.5 standard deviations above its distribution’s mean. A score of −2 says it is two standard deviations below the mean. The score does not, by itself, say whether the value is good, bad, important, or statistically significant. Those judgments depend on the context and the assumptions behind the analysis.

How to calculate a z score

For an individual observation in a population, calculate a z score with:

z = (x − μ) / σ

  • x is the observation being evaluated.
  • μ (mu) is the population mean.
  • σ (sigma) is the population standard deviation.

Subtract the mean from the observation, then divide the difference by the standard deviation. Suppose exam scores have a mean of 70 and a standard deviation of 10. A score of 85 has a z score of (85 − 70) / 10 = 1.5. It is 1.5 standard deviations above the mean.

When working with a sample rather than an entire population, analysts commonly use the sample mean and sample standard deviation instead. Notation and conventions can differ by course or software, so check whether the calculation is intended to describe a population, a sample, or a standardized sample statistic.

Using z scores to compare values

Z scores are especially useful when raw measurements use different units or have different averages and spreads. Imagine one student scores 85 on a test with a mean of 70 and standard deviation of 10, while another scores 78 on a test with a mean of 60 and standard deviation of 12. Their scores are 1.5 and 1.5 standard deviations above their respective means. In relative terms, each performed equally far above the average, though the raw scores cannot be compared directly as if the tests were identical.

This comparison is meaningful only when the reference groups and measurements are suitable. A score standardized against one class does not automatically compare fairly with a score standardized against a different population, a different year, or a test measuring a different skill.

Connecting z scores to percentiles and probability

If values follow a normal distribution, a z score can be used with the standard normal distribution to find the proportion of observations below, above, or between particular values. For instance, z = 0 corresponds to the mean and the 50th percentile in a continuous normal distribution. A z score near 1.96 leaves about 2.5% of the distribution above it and about 2.5% below −1.96.

These probability interpretations depend on the distribution. A z score alone does not prove that a value is rare: the same score can have different percentile implications in distributions with different shapes. For a strongly skewed or heavy-tailed distribution, normal-table estimates may be misleading. Check the data’s distribution or the model’s assumptions before translating a score into a probability.

Common interpretations and limitations

A common rule of thumb treats values more than two standard deviations from the mean as unusual when the distribution is approximately normal. That is a guideline, not a universal cutoff. In a normal distribution, roughly 95% of observations fall within two standard deviations of the mean; real data may not follow that pattern.

  • It is not the raw difference: z = 2 means two standard deviations from the mean, not two measurement units.
  • It is not automatically a probability: a probability or percentile requires a distributional model or relevant empirical data.
  • It does not establish cause or importance: an unusual value may deserve investigation, but the score cannot explain why it occurred.
  • It can be affected by outliers: the mean and standard deviation are sensitive to extreme observations, which can shift or stretch the reference scale.

If the standard deviation is zero, all observations in the reference group are identical, so the usual z score formula involves division by zero and is undefined. If the data are highly skewed or contain influential outliers, robust summaries or other methods may describe relative position more appropriately.

Z scores in hypothesis tests and other statistics

In a z test, a z statistic measures how far an observed result is from a value expected under a null hypothesis, in standard-error units. This is related to the idea of a z score, but it is not necessarily the same calculation as standardizing one individual observation. The z statistic is compared with a reference distribution to assess how compatible the result is with the null hypothesis, given the test’s assumptions.

Z scores also appear in data preparation. Standardizing a variable by subtracting its mean and dividing by its standard deviation gives transformed values with a mean of zero and standard deviation of one, when calculated using the same reference data and conventional population scaling. This can help put numerical features on comparable scales in some analyses, but it does not make the data normally distributed or remove their original outliers.

Frequently Asked Questions

What does a z score of zero mean?
A z score of zero means the value is exactly equal to the mean used as its reference.

Is a higher z score always better?
No. It means the value is farther above the reference mean. Whether that is desirable depends on what is being measured; a high score could be beneficial, harmful, or simply unusual.

What does a negative z score mean?
It means the value is below the mean. For example, z = −1.2 indicates a value 1.2 standard deviations below the reference mean.

Does a z score tell me the percentile?
Not by itself. You can convert it to a percentile if you know or reasonably assume the relevant distribution, commonly the normal distribution. Without that context, the score gives distance from the mean in standard-deviation units.

Can a z score be greater than 3 or less than −3?
Yes. There is no fixed limit on z scores. Under a normal model, values beyond three standard deviations are rare, but they can occur and may be more common in non-normal data.

Related

Scroll to Top