Z-Score Calculator

A z-score answers one question: how many standard deviations is this value from the mean? The formula is z = (x − μ) / σ. A student who scores 130 on a scale with a mean of 100 and a standard deviation of 15 has z = (130 − 100) / 15 = 2, so the score sits exactly two standard deviations above average. A negative z means below the mean, and z = 0 means the value equals the mean.

Standardising puts different scales on the same ruler. Suppose one student scores 680 on a test with μ = 500 and σ = 100 (z = 1.80) and another scores 27 on a test with μ = 21 and σ = 5 (z = 1.20). The raw numbers cannot be compared directly, but the z-scores show the first result is further above its group. This is the reasoning behind comparing SAT-style and ACT-style scores, and it is exactly what Common Core standard HSS-ID.A.4 asks students to do when they use the mean and standard deviation of a data set to fit it to a normal distribution and estimate population percentages.

If the data are roughly normal, a z-score converts to a percentile through the standard normal cumulative distribution function, written Φ(z). Φ(1.96) ≈ 0.9750, so a value with z = 1.96 is higher than about 97.50% of the distribution, and the two tails beyond ±1.96 together hold about 5%. That is why 1.96 appears in 95% confidence intervals.

• 68-95-99.7 rule: about 68.27% of values fall within 1σ of the mean, 95.45% within 2σ and 99.73% within 3σ. Use the Between tab with z-scores of ±1, ±2 and ±3 to check.
• Unusual values: many introductory courses, including AP Statistics, treat |z| greater than 2 as unusual and |z| greater than 3 as very unusual.
• Reverse questions: to find the cut-off for the 90th percentile, look up z = 1.2816 and compute x = μ + zσ. With μ = 70 and σ = 10 that is about 82.8.

Percentiles only follow the bell curve when the data are close to normal. Skewed data, such as incomes or reaction times, can have z-scores that are fine for describing distance from the mean but misleading as percentiles.

Use the population standard deviation σ when the numbers describe the whole group or a published scale (IQ tests use σ = 15). When you only have a sample, the sample standard deviation s (dividing by n − 1) is the usual estimate, and the result is still read as a z-score for descriptive work. For small samples in inference problems, a t-distribution is more appropriate than the normal curve. Common Core HSS-ID.A.2 covers choosing mean and standard deviation to compare the centre and spread of data sets.

Probabilities come from Hart's rational approximation of the normal CDF as published by Graeme West, which is accurate to about 1 × 10⁻¹⁴, so tail areas stay correct even beyond z = 6. Right-tail areas are computed directly rather than as 1 − Φ(z) to avoid rounding. Percentile to z conversions use Peter Acklam's inverse normal approximation followed by one Halley refinement step, which matches printed z-tables to well beyond their four decimal places.

Classroom ideas: collect class heights and standardise each one; give pairs of test results on different scales and ask who did better; or project the shaded curve and have students estimate the area before revealing the percentage.