In introductory statistics courses across US colleges and AP Statistics curricula, few ideas appear more frequently on homework assignments, midterms, and finals than the z-score and the standard normal distribution. Whether you are comparing SAT scores against ACT scores, evaluating manufacturing tolerances in an engineering lab, or testing hypotheses about student performance, standardizing data allows you to measure and compare observations on a universal scale.
A raw measurement alone rarely tells the full story. If a student scores 85 points on a college chemistry midterm, is that an exceptional score or a below-average performance? If the class mean was 70 with a standard deviation of 10, the score is well above average. But if the class mean was 90 with a standard deviation of 4, scoring 85 places that student more than one full standard deviation below the class average.
The z-score transforms raw scores into standardized units that measure relative standing. When the underlying population follows a normal distribution, standardizing converts observations into positions on the classic bell curve, allowing you to compute exact probabilities for any interval.
This guide explains how to calculate z-scores, read left, right, and between normal probabilities, and solve inverse percentile problems. Every numerical result in the three worked homework problems below was computed directly with ToolSphere's tested statistics library. You can verify every step using the interactive Z-score calculator.
What Is a Z-Score?
A z-score (also known as a standard score) measures the signed distance between an observed data value and the mean of its distribution, expressed in units of standard deviation.
The Z-Score Formula
For a population with known mean μ and standard deviation σ, the z-score of a raw observation x is:
z = (x − μ) / σ
For a sample with sample mean x̄ and sample standard deviation s, the sample z-score is:
z = (x − x̄) / s
Interpreting the Sign and Magnitude
- Positive z-score (z > 0): The observation sits above the mean. A score of z = +1.5 means the value is 1.5 standard deviations greater than the center of the distribution.
- Negative z-score (z < 0): The observation sits below the mean. A score of z = −1.0 means the value is 1.0 standard deviation below the center.
- Zero z-score (z = 0): The observation is exactly equal to the mean.
- Magnitude (|z|): The absolute value indicates how unusual or extreme the measurement is relative to the variation in the dataset.
Because the numerator (x − μ) and the denominator (σ) share the same physical measurement units (such as points, pounds, seconds, or dollars), the units cancel out completely. A z-score is a dimensionless quantity.
Example: Comparing Different Scales
Student A scores 1320 on the SAT (SAT mean = 1050, SD = 210).
z = (1320 − 1050) / 210 = 270 / 210 ≈ +1.286
Student B scores 28 on the ACT (ACT mean = 20.8, SD = 5.8).
z = (28 − 20.8) / 5.8 = 7.2 / 5.8 ≈ +1.241
Although 1320 and 28 cannot be directly subtracted, their z-scores reveal that
Student A placed slightly higher relative to their peer group than Student B.
The Standard Normal Distribution
While any dataset with a finite mean and standard deviation can be standardized into z-scores, converting a z-score into an area or probability requires assuming a specific distribution model.
The standard normal distribution is the special normal curve where the mean is zero and the standard deviation is one:
Z ~ N(0, 1)
Mean: μ = 0
Standard Deviation: σ = 1
Variance: σ² = 1
Fundamental Properties of the Bell Curve
- Symmetric Bell Shape: The curve is perfectly symmetric around the center z = 0. The left half and right half each contain exactly 50% of the total probability (area = 0.5000).
- Total Area Equals One: The total area underneath the probability density curve across all real numbers equals exactly 1.0000 (representing 100% of all possible outcomes).
- Mean Equals Median Equals Mode: In the standard normal distribution, the center, median cutoff, and highest peak all coincide at z = 0.
- Inflection Points at ±1: The curve changes concavity (from concave downward to concave upward) at exactly z = −1 and z = +1.
The Empirical Rule (68–95–99.7 Rule)
For any approximately bell-shaped normal distribution:
- Approximately 68.27% of all observations lie within 1 standard deviation of the mean:
P(−1 ≤ Z ≤ 1) ≈ 0.6827. - Approximately 95.45% of all observations lie within 2 standard deviations of the mean:
P(−2 ≤ Z ≤ 2) ≈ 0.9545. - Approximately 99.73% of all observations lie within 3 standard deviations of the mean:
P(−3 ≤ Z ≤ 3) ≈ 0.9973.
In statistical inference and confidence intervals, instructors frequently use z = ±1.96 rather than z = ±2, because exactly 95.00% of the normal probability sits between −1.96 and +1.96: P(−1.96 ≤ Z ≤ 1.96) ≈ 0.9500.
Reading Normal Probabilities
Normal distribution questions on exams and homework typically fall into three distinct geometric regions under the curve:
Left Tail Right Tail Between Two Bounds
(At or Below) (At or Above) (Interval [z1, z2])
████ ████ ████████
██████ ██████ ████████████
████████ ████████ ████████████████
──████████──────── ────────████── ────████████████████──
z z z1 z2
P(Z ≤ z) = Φ(z) P(Z ≥ z) = 1 − Φ(z) P(z1 ≤ Z ≤ z2) = Φ(z2) − Φ(z1)
1. Cumulative Left-Tail Probability: P(Z ≤ z)
The cumulative distribution function (CDF), traditionally written as Φ(z), gives the probability that a standard normal random variable Z is less than or equal to a given cutoff z:
P(Z ≤ z) = Φ(z)
In standard textbook tables (the printed z-table in the back of your statistics book), the row provides the sign, whole number, and tenths digit of z, while the column provides the hundredths digit. The intersection gives Φ(z). For example, looking up z = 1.00 gives 0.8413.
2. Right-Tail Probability: P(Z ≥ z)
A question asking for the proportion "above," "greater than," or "exceeding" a value targets the upper tail:
P(Z ≥ z) = 1 − P(Z ≤ z) = 1 − Φ(z)
Because the total area is 1, the right-tail probability is simply the complement of the left-tail probability. By symmetry, the right tail at +z is also identical to the left tail at −z:
P(Z ≥ z) = P(Z ≤ −z) = Φ(−z)
3. Between Probability: P(z1 ≤ Z ≤ z2)
To find the probability that an observation falls inside an interval between two values z1 and z2 (where z1 < z2):
P(z1 ≤ Z ≤ z2) = Φ(z2) − Φ(z1)
You calculate the cumulative area up to the higher bound z2 and subtract the unshaded cumulative area below the lower bound z1.
Continuous Distribution Note: Strict vs. Non-Strict Inequalities
In a continuous probability distribution like the normal distribution, the probability of obtaining any single exact point is zero: P(Z = c) = 0. Consequently, strict and non-strict inequalities yield identical probabilities:
P(Z ≤ z) = P(Z < z)
P(Z ≥ z) = P(Z > z)
P(z1 ≤ Z ≤ z2) = P(z1 < Z < z2)
Including or excluding the boundary line does not change the calculated area.
Working Backward: Percentiles and Inverse Normal
Homework problems often ask you to reverse this workflow: given a desired percentage or percentile rank, what raw score or z-score serves as the cutoff?
Defining a Percentile
The k-th percentile (denoted P_k) is the value below which k percent of observations fall:
P(Z ≤ z) = p = k / 100
For example, the 90th percentile corresponds to a left-tail cumulative probability of p = 0.9000.
The Inverse Normal Process
- Identify the cumulative left area p: If the problem specifies the "top 10%," the left area is
p = 1.00 − 0.10 = 0.90. - Find the standard normal quantile z: Look inside the body of the z-table for the probability closest to p, then read outward to find the corresponding z-score
z = Φ⁻¹(p). - Un-standardize to raw units x: Rearrange the z-score formula to solve for x:
x = μ + z × σ
Three Worked Homework Problems
Every calculation below is computed using ToolSphere's validated normal.ts mathematics engine. Both the exact double-precision values and standard four-decimal textbook rounding are provided so you can verify your homework answers.
Problem 1: Single Bound (Left and Right Tail)
Scenario: Scores on a university general chemistry midterm are normally distributed with a mean of μ = 75 points and a standard deviation of σ = 10 points. A student scores x = 88 points.
- Part A: What is the student's z-score?
- Part B: What proportion of the class scored at or below 88 points?
- Part C: What proportion of the class scored higher than 88 points?
Solution Step by Step
Step 1: Calculate the z-score
z = (x − μ) / σ
z = (88 − 75) / 10
z = 13 / 10 = +1.30
The student scored 1.30 standard deviations above the class mean.
Step 2: Find the left-tail probability P(X ≤ 88)
Convert the raw score inequality into standard normal notation:
P(X ≤ 88) = P(Z ≤ 1.30) = Φ(1.30)
Looking up z = 1.30 in a standard normal table gives 0.9032. The exact numerical value evaluated by the library is:
Exact library value: 0.9031995154143897
Textbook table value: 0.9032 (or 90.32%)
Interpretation: Approximately 90.32% of the students scored at or below 88 points on the chemistry midterm.
Step 3: Find the right-tail probability P(X ≥ 88)
Use the complement rule:
P(X ≥ 88) = 1 − P(X ≤ 88)
P(X ≥ 88) = 1 − 0.9031995154143897 = 0.09680048458561025
Textbook table value: 1 − 0.9032 = 0.0968 (or 9.68%)
Interpretation: Only about 9.68% of the students scored higher than 88 points.
Problem 2: Interval Between Two Bounds
Scenario: An express courier service monitors delivery times for packages within an urban zone. Delivery times follow an approximately normal distribution with a mean of μ = 48 minutes and a standard deviation of σ = 6 minutes.
- Question: What is the probability that a randomly chosen package is delivered between 42 minutes and 57 minutes (
P(42 ≤ X ≤ 57))?
Solution Step by Step
Step 1: Standardize both boundary values
Lower bound (x1 = 42 minutes):
z1 = (x1 − μ) / σ
z1 = (42 − 48) / 6 = −6 / 6 = −1.00
Upper bound (x2 = 57 minutes):
z2 = (x2 − μ) / σ
z2 = (57 − 48) / 6 = +9 / 6 = +1.50
The question is equivalent to finding P(−1.00 ≤ Z ≤ 1.50).
Step 2: Find the cumulative probability for each boundary
From the library's normal CDF:
Lower bound area: Φ(−1.00) = 0.15865525393145702 (table value: 0.1587)
Upper bound area: Φ(1.50) = 0.9331927987311419 (table value: 0.9332)
Step 3: Subtract the lower cumulative area from the upper cumulative area
P(−1.00 ≤ Z ≤ 1.50) = Φ(1.50) − Φ(−1.00)
P(−1.00 ≤ Z ≤ 1.50) = 0.9331927987311419 − 0.15865525393145702
P(−1.00 ≤ Z ≤ 1.50) = 0.774537544799685
Textbook table value: 0.9332 − 0.1587 = 0.7745 (or 77.45%)
Interpretation: There is an estimated 77.45% probability (approximately 77 out of every 100 packages) that delivery takes between 42 and 57 minutes.
Problem 3: Inverse Normal and Percentile Cutoffs
Scenario: A state university department administers a standardized engineering readiness examination where scores are normally distributed with a mean of μ = 500 points and a standard deviation of σ = 100 points.
- Part A: The department awards a Dean's Scholarship to students scoring in the top 5% of all test takers. What minimum score is required to qualify?
- Part B: Students scoring in the lower quartile (bottom 25%) are assigned a peer tutor. What is the score cutoff for tutorial support?
Solution Step by Step
Part A: Top 5% Cutoff (Dean's Scholarship)
Determine the cumulative probability p:
Being in the top 5% means 95% of test takers scored below you.p = 1.00 − 0.05 = 0.9500Find the critical z-score:
In standard z-tables, p = 0.9500 sits halfway between z = 1.64 (0.9495) and z = 1.65 (0.9505), commonly noted as z ≈ 1.645.
Evaluating the inverse normal CDF directly yields:z = Φ⁻¹(0.95) = 1.6448536269514715Convert back to the raw score x:
x = μ + z × σ x = 500 + (1.6448536269514715 × 100) x = 500 + 164.48536269514715 x = 664.4853626951472 pointsRounding to two decimal places gives 664.49 points (or a whole test score of 665 points).
Interpretation: A student must achieve at least 664.49 points to qualify for the Dean's Scholarship.
Part B: Bottom 25% Cutoff (Lower Quartile Q1)
Determine the cumulative probability p:
The lower quartile represents the 25th percentile:p = 0.2500Find the critical z-score:
Because p < 0.50, the z-score must be negative. By symmetry with the 75th percentile (z ≈ +0.6745):z = Φ⁻¹(0.25) = −0.6744897501960814Convert back to the raw score x:
x = μ + z × σ x = 500 + (−0.6744897501960814 × 100) x = 500 − 67.44897501960814 x = 432.55102498039184 pointsRounding to two decimal places gives 432.55 points (or a whole score of 432 points and below).
Interpretation: Any applicant scoring 432.55 points or lower falls into the bottom 25% and receives peer tutoring.
Five Common Exam Pitfalls to Avoid
When completing normal distribution homework or exam questions, watch out for these frequent mistakes:
1. Forgetting to Subtract from 1 on "Greater Than" Questions
Standard normal tables display cumulative areas to the left: P(Z ≤ z). When an exam question asks for the probability that a value "exceeds," "is more than," or "is at least" a threshold, you must compute 1 − Φ(z). Failing to subtract from 1 gives the exact opposite proportion.
2. Confusing the Top k% with the k-th Percentile
"Top 10%" does not mean looking up p = 0.10. An observation in the top 10% has 90% of the distribution below it, meaning you must look up p = 0.90 (yielding z ≈ +1.282). Looking up p = 0.10 yields a negative z-score (z ≈ −1.282), which represents the bottom 10%.
3. Confusing a Z-Score with a Probability
A z-score is a distance measured in standard deviations; a probability is an area under the curve between 0 and 1. A calculated z-score of 1.45 is not a probability of 1.45% or 145%. You must pass the z-score through the normal CDF function (or look it up in a table) to obtain the corresponding probability (Φ(1.45) ≈ 0.9265).
4. Assuming Standardization Makes Skewed Data Normal
Calculating z-scores shifts the mean to 0 and scales the standard deviation to 1, but it preserves the original relative spacings between data points. If a dataset is heavily right-skewed (such as household income or software defect counts), the standardized z-scores will remain right-skewed. Normal curve probabilities are valid only when the underlying distribution is reasonably normal.
5. Using Individual Standard Deviation Instead of Standard Error for Means
If a problem asks for the probability regarding a single randomly chosen individual, use the standard deviation σ. But if the problem asks for the probability regarding the average of a sample of n individuals (x̄), you must use the standard error of the mean:
σ_x̄ = σ / √n
Failing to divide by √n on sampling distribution questions is one of the most common point deductions in introductory statistics.
Verify Your Calculations Online
You can verify any z-score, find left, right, or between normal probabilities, and compute exact percentile thresholds with ToolSphere's free Z-Score Calculator. The tool displays a shaded SVG normal curve, updates probabilities in real time, and supports both original values and standardized units.
For additional probability distributions and data analysis tools, explore the Probability and Statistics Lab.
Use the Statistics Calculator to summarize a dataset before interpreting it. For a related walkthrough, read Normal, Binomial or Poisson: Which Applies.