MATH 3 • STATISTICS & PROBABILITY

Using z-Scores — I can use z-scores at an introductory level to compare values from a normal distribution.

Learn how z-scores let you compare data from completely different distributions on a single, universal scale.

Historical Context & Motivation

Imagine you scored 82 on a chemistry test and 74 on a history test. Which score is actually "better" relative to your classmates? The raw numbers alone don't tell you, because the two tests might have very different averages and spreads. This exact problem—comparing values from different distributions—motivated statisticians to develop a universal yardstick called the z-score. The idea grew out of centuries of work on understanding how data clusters around a central value and how far individual observations tend to stray.

1733
De Moivre's Bell Curve
Abraham de Moivre first described the normal (bell-shaped) curve while studying probabilities of coin flips. His work laid the groundwork for measuring how far a value lies from the center of a distribution.
1809
Gauss & the Normal Distribution
Carl Friedrich Gauss used the bell curve to model errors in astronomical measurements. The normal distribution became known as the Gaussian distribution in his honor.
1893
Karl Pearson's Standard Deviation
Karl Pearson coined the term standard deviation, giving statisticians a precise way to measure the spread of data—a concept essential for computing z-scores.
1920s
Standardization Takes Hold
Researchers began routinely converting raw scores into standard scores (z-scores) so that results from different experiments could be compared on a common scale. This practice became a cornerstone of modern statistics.

The core question that z-scores answer is deceptively simple: How unusual is a particular data value within its own distribution? By expressing every value in terms of how many standard deviations it falls above or below the mean, z-scores give us a universal language for comparison—regardless of the original units or scale.

Core Principles & Definitions

Before you can calculate or interpret a z-score, you need to understand a few foundational ideas. A z-score doesn't exist in isolation—it relies on the mean and standard deviation of the data set it comes from. Together, these two measures anchor the z-score and give it meaning.

1

Mean (μ)

The mean is the arithmetic average of all values in a data set. It serves as the center of a normal distribution—the peak of the bell curve.
2

Standard Deviation (σ)

The standard deviation measures how spread out the data values are from the mean. A small σ means data points cluster tightly; a large σ means they are widely scattered.
3

Normal Distribution

A normal distribution is a symmetric, bell-shaped curve where about 68% of data falls within 1σ of the mean, 95% within 2σ, and 99.7% within 3σ.
4

Z-Score

A z-score tells you how many standard deviations a particular value (x) is above or below the mean. Positive z = above mean; negative z = below mean; z = 0 means exactly at the mean.
KEY TAKEAWAY
Think of a z-score like a GPS coordinate for data. Just as GPS uses latitude and longitude to pinpoint any location on Earth—no matter what country you're in—a z-score uses the mean and standard deviation to pinpoint where any data value sits within its distribution. Two test scores measured in completely different units can be placed on the same "map" once you convert them to z-scores.

The Normal Curve & z-Score Regions

The diagram below shows a standard normal distribution—a bell curve with a mean of 0 and a standard deviation of 1. Every normal distribution can be transformed into this standard form using z-scores. The shaded regions illustrate the 68-95-99.7 rule (also called the Empirical Rule), which describes the percentage of data falling within 1, 2, and 3 standard deviations of the mean.

The bell curve peaks at z = 0 (the mean). The purple shaded band covers z = −1 to z = +1, capturing 68% of all data. The amber bracket extends to ±2σ (95%), and the green bracket reaches ±3σ (99.7%).

Notice that the curve is perfectly symmetric about the mean. A z-score of +1.5 and a z-score of −1.5 are equally far from the center—they just lie on opposite sides. Values with z-scores beyond ±2 are relatively rare (only about 5% of the data), and values beyond ±3 are extremely unusual (about 0.3%). This visual intuition is powerful: the further your z-score is from zero, the more exceptional your data value is compared to the rest of the distribution.

The z-Score Formula

The math behind z-scores is elegant and straightforward. You only need three values: the data point you're examining, the mean of the distribution, and the standard deviation. The formula converts any raw score into its standardized position on the bell curve.

Z-SCORE FORMULA
z = (x − μ) / σ
Where x = the individual data value, μ (mu) = the mean of the distribution, and σ (sigma) = the standard deviation of the distribution.

Let's break this formula into two logical steps. First, the numerator (x − μ) calculates how far the data value is from the mean—this is the raw deviation. Second, you divide that deviation by the standard deviation σ, which rescales the distance into "number of standard deviations." Dividing by σ is the key step that makes different distributions comparable.

INTERPRETING THE SIGN
z > 0 → above the mean | z < 0 → below the mean | z = 0 → at the mean
The sign of the z-score instantly tells you which side of the center the value falls on. The magnitude tells you how far.
RECOVERING THE RAW VALUE
x = μ + z × σ
If you know the z-score and want to find the original data value, rearrange the formula: multiply the z-score by the standard deviation and add the mean.
💡 Quick Check
If a test has μ = 75 and σ = 5, what z-score does a student with a score of 75 receive? Since 75 − 75 = 0, we get z = 0 / 5 = 0. A score equal to the mean always has a z-score of zero, regardless of the standard deviation.

Using z-Scores to Compare Across Distributions

The real power of z-scores shows up when you need to compare values from different distributions. Raw scores are often misleading because each distribution has its own mean and spread. Converting to z-scores places every value on the same standardized scale, so you can make a fair, apples-to-apples comparison.

An SAT score of 620 converts to z = +1.2, while an ACT score of 29 converts to z = +1.6. On the unified z-score scale at the bottom, the ACT 29 (z = +1.6) sits further to the right, indicating a relatively stronger performance compared to peers—even though the raw numbers look very different.

In the diagram above, raw scores of 620 (SAT) and 29 (ACT) are meaningless to compare directly because the scales differ dramatically. However, once you compute each z-score, the comparison is simple: the student with z = +1.6 performed further above average relative to test-takers than the student with z = +1.2. The rule is straightforward—the higher the z-score, the better the relative performance (assuming higher is better in context).

Comparison of SAT and ACT scores using z-scores
ScenarioRaw ValueMean (μ)Std Dev (σ)z-Score
SAT Math620500100+1.20
ACT Math29215+1.60

Worked Example: Who Ran Faster?

Two athletes from different schools both claim they're the fastest sprinter in their league. In League A, Mia ran the 100-meter dash in 12.1 seconds (League A mean = 13.0 s, σ = 0.6 s). In League B, Carlos ran it in 12.4 seconds (League B mean = 13.5 s, σ = 0.8 s). Who performed better relative to their competition?

Comparing Sprint Times with z-Scores
1
Step 1 — Identify Given ValuesFor Mia (League A): x = 12.1 s, μ = 13.0 s, σ = 0.6 s. For Carlos (League B): x = 12.4 s, μ = 13.5 s, σ = 0.8 s. Note that in sprinting, lower times are better, so a more negative z-score (further below the mean) indicates a stronger performance.
2
Step 2 — Calculate Mia's z-ScoreApply the formula: z = (x − μ) / σ = (12.1 − 13.0) / 0.6 = (−0.9) / 0.6
Mia's z-score: z = −1.50
3
Step 3 — Calculate Carlos's z-ScoreApply the formula: z = (x − μ) / σ = (12.4 − 13.5) / 0.8 = (−1.1) / 0.8
Carlos's z-score: z = −1.375
4
Step 4 — Compare and InterpretSince lower race times are better, the more negative z-score represents a stronger relative performance. Mia's z = −1.50 is further below the mean than Carlos's z = −1.375. This means Mia's time was 1.5 standard deviations below her league's average, while Carlos's was about 1.38 standard deviations below his.
Mia performed relatively better compared to her league, even though Carlos had a faster raw difference from his mean in seconds.
⚠️ Context Matters!
When interpreting z-scores, always consider whether a higher or lower raw value is "better." For test scores, a high z-score is good. For race times, costs, or error rates, a low (more negative) z-score is better. The z-score itself just measures distance from the mean—you supply the context.

Strengths & Limitations of z-Scores

Z-scores are one of the most versatile tools in introductory statistics, but like any tool, they work best when you understand both what they can and cannot do. The table below outlines the key strengths and limitations you should keep in mind.

Strengths vs. limitations of using z-scores
StrengthsLimitations
Allow comparison of values from different distributions with different units and scales.Only meaningful when the data is approximately normally distributed (bell-shaped).
Provide an immediate sense of how unusual a value is—z beyond ±2 is rare.Sensitive to outliers that can distort the mean and standard deviation.
Simple formula that only requires the mean and standard deviation.Do not tell you about causation—a high z-score doesn't explain why a value is extreme.
Foundation for more advanced concepts like hypothesis testing and confidence intervals.For highly skewed distributions, percentiles or other measures may be more appropriate.
KEY TAKEAWAY
Z-scores are like converting temperatures between Fahrenheit and Celsius—the underlying reality (the temperature) doesn't change, but the scale you express it on does. A z-score converts your data into the "standard deviation scale." This conversion is incredibly useful, but it only makes physical sense when the data is roughly bell-shaped. If your data distribution is heavily lopsided, the z-score may give a misleading picture of how unusual a value really is.

Connection to Advanced Statistics

The z-score you learn now is the foundation for many powerful statistical methods you'll encounter in college-level courses and beyond. Understanding how it connects to more advanced ideas will give you a head start and help you see why mastering z-scores really matters.

From introductory z-scores to advanced statistical methods
What You Learn NowWhere It Leads
z = (x − μ) / σ for a single data valuez-test statistic: z = (x̄ − μ) / (σ / √n) to test claims about population means using sample data
Using the 68-95-99.7 rule for rough probability estimatesUsing the standard normal (z) table to find exact probabilities and percentiles
Comparing two values from different distributionsHypothesis testing: deciding whether observed differences are statistically significant
Understanding that z = 0 means "at the mean"Confidence intervals: building ranges around estimates to capture the true population value

In AP Statistics and college courses, you'll also encounter t-scores, which work similarly to z-scores but account for the added uncertainty when you estimate the standard deviation from a small sample instead of knowing the true population σ. The t-distribution is slightly wider (heavier tails) than the standard normal distribution, but as the sample size grows, t-scores converge toward z-scores. Everything you learn about z-scores transfers directly to this more advanced context.

Practice Problems

PROBLEM 1CONCEPTUAL
A student scores exactly at the mean on a normally distributed exam. What is their z-score, and what does it tell us about their performance relative to the class?
PROBLEM 2BASIC CALCULATION
On a biology exam, the class mean is 72 and the standard deviation is 8. If Priya scores an 88, what is her z-score?
PROBLEM 3INTERMEDIATE
A student scored 680 on the SAT Critical Reading (μ = 500, σ = 100) and 31 on the ACT English (μ = 21, σ = 6). On which test did the student perform relatively better, and by how much (in terms of z-score difference)?
PROBLEM 4APPLIED
A quality-control engineer knows that the mean diameter of bolts produced by Machine A is 10.00 mm with σ = 0.05 mm, while Machine B has a mean of 10.00 mm with σ = 0.12 mm. A bolt from Machine A measures 10.08 mm and a bolt from Machine B measures 10.15 mm. Which bolt is more unusual for its machine? Should either be flagged for inspection (assume flagging at |z| ≥ 2)?
PROBLEM 5CRITICAL THINKING
Suppose a data set is heavily right-skewed (most values are low, but a few are extremely high). A value has a z-score of +2.5. Can you reliably say that approximately 99.4% of values fall below it, as you would with a normal distribution? Explain your reasoning and suggest what might be a better approach.

Lesson Summary

A z-score measures how many standard deviations a data value lies above or below the mean of its distribution, and it is calculated using the formula z = (x − μ) / σ. A positive z-score means the value is above the mean; a negative z-score means it is below. A z-score of zero indicates the value is exactly at the mean. The 68-95-99.7 rule provides quick benchmarks: about 68% of normally distributed data falls within z = ±1, 95% within z = ±2, and 99.7% within z = ±3.

The greatest power of z-scores is their ability to place values from different distributions onto a single standardized scale, enabling fair comparisons even when the original units, means, and spreads differ. However, z-scores are most reliable when the data follows an approximately normal distribution. For highly skewed data, direct percentile calculations or data transformations may be more appropriate. Mastering z-scores now prepares you for future topics like z-tables, hypothesis testing, and confidence intervals.

Varsity Tutors • Math 3 • Using z-Scores