Standardization and z-Values
When working with normally distributed data, raw values alone can be hard to compare — a score of 75 on one test means something very different from a score of 75 on another if the tests have different means and spreads. Standardization solves this problem by converting any normal variable into a universal scale. The resulting value, called a z-score (or standard score), tells you exactly how many standard deviations a data point lies above or below the mean of its distribution. The z-score formula z = (x − μ)/σ is one of the most powerful tools in statistics. Once a value is standardized, you can look up probabilities using the standard normal distribution Z ~ N(0, 1), which has mean 0 and standard deviation 1. This works because every normal distribution, regardless of its mean μ and standard deviation σ, transforms perfectly onto this single reference distribution. Tables of the standard normal (often called z-tables or Φ-tables) and calculator functions like invNorm allow precise probability calculations. Beyond finding probabilities from known values, z-scores enable inverse normal calculations: starting from a known probability, you work backwards to find a missing x-value, mean, or standard deviation. This makes standardization essential for problems involving percentiles, quality control thresholds, and setting cut-off scores — wherever you know how extreme you want a value to be and need to find the corresponding measurement.
- Understand what a z-score measures
- Apply the standardization formula
- Find probabilities using the standard normal table or calculator
- Perform inverse normal calculations
- Handle two-condition problems to find μ and σ simultaneously
Understand what a z-score measures
A z-score is a signed number. Positive z means the value is above the mean; negative z means it is below. For example, z = 2 means the value is exactly 2 standard deviations above the mean, while z = −1.5 means it is 1.5 standard deviations below. Because the standard normal distribution is symmetric about 0, about 68% of data falls between z = −1 and z = 1, about 95% between z = −2 and z = 2, and about 99.7% between z = −3 and z = 3.
Apply the standardization formula
To convert a raw value x from a normal distribution X ~ N(μ, σ²) into a z-score, use z = (x − μ)/σ. The numerator shifts the distribution so its centre lands at 0, and dividing by σ rescales it so one unit equals one standard deviation. The resulting Z ~ N(0, 1). For instance, if heights in a population are distributed as X ~ N(170, 8²) cm, a height of x = 186 gives z = (186 − 170)/8 = 16/8 = 2.00.
Find probabilities using the standard normal table or calculator
The cumulative distribution function Φ(z) = P(Z ≤ z) gives the probability that the standard normal variable is at most z. Most z-tables list Φ(z) for z from −3.49 to +3.49. For example, Φ(2.00) ≈ 0.9772, so P(X ≤ 186) ≈ 97.72% for the height example. For P(X > x), use 1 − Φ(z). For a range P(a < X < b), compute Φ(z_b) − Φ(z_a) after converting both endpoints to z-scores.
Perform inverse normal calculations
Sometimes you are given a probability and must find the corresponding x-value. Use invNorm(p) (on a calculator) or the z-table in reverse to find the z-score corresponding to cumulative probability p, then rearrange the formula: x = μ + zσ. If instead the unknown is the mean μ, rearrange to μ = x − zσ. If the unknown is the standard deviation σ, use σ = (x − μ)/z. In all cases, first identify the z-score from the probability, then substitute known quantities to isolate the unknown.
Handle two-condition problems to find μ and σ simultaneously
Advanced problems provide two probability statements, creating a system of two equations. For example: 'P(X < 50) = 0.20 and P(X < 80) = 0.90.' From the z-table, z₁ = invNorm(0.20) ≈ −0.842 and z₂ = invNorm(0.90) ≈ 1.282. This gives the system: 50 = μ + (−0.842)σ and 80 = μ + (1.282)σ. Subtracting the first equation from the second: 30 = (1.282 + 0.842)σ = 2.124σ, so σ ≈ 14.12. Then μ = 50 + 0.842 × 14.12 ≈ 61.89.
The Key Equation
- z — z-score (standard score) (dimensionless): The number of standard deviations the value x is from the mean; positive above the mean, negative below.
- x — observed value (same as the original data): The specific data point being standardized.
- \mu — population mean (same as the original data): The arithmetic average of the normal distribution from which x is drawn.
- \sigma — population standard deviation (same as the original data): The square root of the variance; measures the spread of the distribution. Must be positive.
Worked Example
Problem. The masses of bags of flour produced by a factory are normally distributed with mean μ = 1000 g and standard deviation σ = 12 g. (a) Find the probability that a randomly chosen bag has mass less than 982 g. (b) The factory rejects the lightest 5% of bags. Find the minimum acceptable mass, giving your answer to 1 decimal place.
Answer. P(X < 982) ≈ 0.0668 (6.68%); minimum acceptable mass ≈ 980.3 g
Origins · 1 minCarl Friedrich Gauss and how this was discovered
Gauss formalized and popularized the normal distribution (often called the Gaussian distribution in his honour) in the early 19th century, using it to describe measurement errors in astronomical data. His method of least squares and his analysis of normally distributed errors laid the statistical foundation that makes z-score standardization possible and meaningful. Gauss was a child prodigy who reportedly corrected his father's payroll arithmetic at age 3. At school, he famously summed all integers from 1 to 100 in seconds by pairing numbers (1+100, 2+99, …) — giving 50 pairs each summing to 101, for a total of 5050.
Related Concepts
- The Normal Distribution — Standardization and z-scores only apply within normal distributions; understanding the bell-curve shape, its parameters μ and σ², and its symmetry properties is the essential prerequisite for interpreting z-scores correctly.
- Continuous Probability Models — Z-scores are used to compute probabilities for continuous random variables. Understanding probability density functions and cumulative distribution functions gives z-table look-ups their mathematical meaning.
- Mean and Standard Deviation — The z-score formula requires both μ (mean) and σ (standard deviation). A solid grasp of how these measures of centre and spread are calculated and interpreted is necessary before standardization makes sense.
- Hypothesis Testing with z-Tests — The z-score is the test statistic in one- and two-sample z-tests. Standardization is the direct bridge between observed data and the p-values used to accept or reject statistical hypotheses.