Free Z-Score Calculator with Normal Distribution Graphs

A Z-score (standard score) measures how many standard deviations a particular value is above or below the mean of a distribution. Calculated as Z = (x - mu)/sigma, it transforms any normally distributed variable into the standard normal distribution (mu=0, sigma=1). A Z of +2 means the value is 2 standard deviations above the mean; Z = -1.5 means 1.5 SDs below. This standardisation allows comparison of values from different distributions with different means and units.

A percentile tells you what percentage of the population falls below a given value. To convert Z to percentile, calculate the CDF: P = Phi(Z) = (1/2)[1 + erf(Z/sqrt(2))]. Common conversions: Z=0 is the 50th percentile; Z=1 is the 84th; Z=1.645 is the 95th; Z=1.96 is the 97.5th; Z=2.576 is the 99.5th; Z=-1 is the 16th. This is exactly what a Z-table looks up.

Left-tail P(X <= x) = Phi(Z) is the area to the LEFT of z on the bell curve. Right-tail P(X >= x) = 1 - Phi(Z) is the area to the right. Two-tail P(|Z| >= z) = 2(1 - Phi(|z|)) is the total area in both tails. For hypothesis testing, two-tail tests use alpha/2 in each tail; one-tail tests use the full alpha in one tail. Z = 1.96 gives a 5% two-tail significance level.

The inverse Z (quantile function) finds the Z-value for a given cumulative probability. For example: the 90th percentile is Z = 1.282; the 97.5th percentile is Z = 1.960; the 99th percentile is Z = 2.326. This is used to find z* for confidence intervals and the critical z-value for a given alpha in hypothesis testing. Computed as Z = sqrt(2) * erf_inverse(2p - 1).

A confidence interval (CI) is a range of values that contains the true population mean with a specified probability. The formula is x-bar +/- z* * (sigma/sqrt(n)). The z* is the critical Z value: 1.645 for 90%, 1.960 for 95%, 2.576 for 99%. The margin of error ME = z* * SE where SE = sigma/sqrt(n). A wider CI gives more confidence but less precision. A 95% CI does not mean 95% probability that this specific interval contains mu; rather, 95% of intervals built this way contain the true mean.

A one-sample Z-test checks whether a sample mean differs significantly from a known population mean, using Z = (x-bar - mu0) / (sigma/sqrt(n)). Use it when the population standard deviation sigma is known and n > 30 (or data are normal). A two-sample Z-test compares means from two independent groups. If sigma is unknown and n is small, use a t-test instead. The p-value is derived from the standard normal CDF.

The normal (Gaussian) distribution is a symmetric bell-shaped distribution defined by its mean mu and standard deviation sigma. Many natural phenomena approximate normality: heights, test scores, measurement errors, blood pressure. The Central Limit Theorem guarantees that sample means approach normality for large n (n > 30), even when individual observations are skewed, which is why Z-score methods work broadly.

Common thresholds: |Z| > 2 for a moderate outlier, |Z| > 3 for an extreme outlier. Values with |Z| > 3 occur only 0.27% of the time in a normal distribution (about 3 in 1,000). In Six Sigma quality control, processes target +/-6 sigma from the specification target, allowing only 3.4 defects per million opportunities. The appropriate threshold depends on context.

Standard deviation (sigma or s) measures the spread of individual data values around the mean. Standard error (SE = sigma/sqrt(n)) measures how much the sample mean varies across different samples. SE is always smaller than sigma for n > 1 and shrinks as 1/sqrt(n) as sample size increases, which is why larger samples give more precise estimates of the population mean.

For a confidence interval with desired margin of error ME: n = (z* * sigma / ME) squared, rounded up. For a hypothesis test with desired power (1 - beta) and minimum detectable effect delta: n = ((z_alpha/2 + z_beta) * sigma / delta) squared. Larger n gives narrower CIs and higher power at increased cost.

Z-scores can be calculated for any distribution, but the probability interpretations from the Z-table only apply precisely to normal distributions. For non-normal data, Chebyshev's inequality gives a distribution-free lower bound: at least (1 - 1/k^2) * 100% of data lies within +/-k sigma. For large samples (n > 30), the Central Limit Theorem ensures the sample mean is approximately normal, so Z-based confidence intervals remain valid.

Both Z and t-scores measure distance from the mean in standard deviation units. Z-scores are used when the population sigma is known or n > 30. T-scores are used when sigma is unknown and estimated by s; the t-distribution has heavier tails than the normal, especially for small n. As n increases, the t-distribution converges to the standard normal. Rule of thumb: use Z when sigma is known or n > 30; use t when sigma is unknown and n <= 30.

Process Z (sigma level) measures capability: a Six Sigma process has +/-6 sigma between the mean and specification limits, corresponding to 3.4 DPMO. Process capability indices Cp = (USL - LSL)/(6 sigma) and Cpk indicate how well the process fits within spec limits. Control charts use +/-3 sigma limits as standard control limits; points beyond these trigger investigation.

Z-score normalisation: for each value x, compute Z = (x - mu) / sigma. The result has mean = 0 and std = 1. This is essential before PCA, k-means clustering, support vector machines, or any distance-based algorithm where features with different scales would otherwise dominate. Min-max normalisation is a different technique that maps values to [0, 1].

The CLT states that the sampling distribution of the mean approaches N(mu, sigma^2/n) as n increases, regardless of the population shape. For n >= 30, x-bar is approximately normal even if individual observations are skewed. This is why Z-based confidence intervals and hypothesis tests are valid for large samples from any distribution, not just normally distributed populations.