What is a z-score, and how is it used to compare observations from different normal distributions?
Calculate z-scores and use them to compare values from different normal distributions and find probabilities
A focused answer to the HSC Maths Standard 2 dot point on z-scores. The formula , standardising a value step by step, interpreting z-scores as standard-deviation distances, comparing observations from different normal distributions, reversing to find a value from a percentile, and worked Australian examples from exam marks and salary data.
Reviewed by: AI editorial process; not yet individually human-reviewed
Have a quick question? Jump to the Q&A page
What this dot point is asking
NESA wants you to do four things. Compute a z-score for any value from a normal distribution (a bell-shaped spread of data). Read that z-score as "how many standard deviations above or below the mean". Reverse the process to find a value from a given z-score or percentile. And use z-scores to compare observations that come from different normal distributions. The z-score is the bridge between a raw measurement and one common scale on which every normal distribution looks the same.
The answer
The z-score formula
For a value from a normal distribution with mean and standard deviation :
The z-score is the number of standard deviations lies from the mean. It is negative when is below the mean and positive when above. Subtracting shifts the data so the mean sits at . Dividing by then rescales it so one standard deviation becomes one unit. The result is a pure number with no units. A height in cm and a salary in dollars both become plain z-scores, and that is exactly why they can then be compared.
Standardising a value, step by step
Standardising means carrying a raw value across to the common z-scale. The diagram below maps a test mark of (on a test with , ) down to its z-score. The two parts of the formula are two moves on the picture. Subtracting the mean slides the marked value across relative to the centre. Dividing by then measures that gap in standard-deviation steps.
Here , so the mark sits standard deviations above the mean.
Reverse: finding from a z-score
Rearranging the formula gives the raw value back:
This is the move for percentile questions. Given a target z-score (often supplied, or read from the table), it returns the actual height, mark or salary at that position. The two forms are a matched pair, one going each way. The first formula, , turns a value into a z-score. The second, , turns a z-score back into a value.
Interpretation
- : the value equals the mean.
- : one standard deviation above the mean (top of a normal distribution).
- : two standard deviations below the mean (bottom ).
- : a very extreme value; under of data sit this far from the mean.
A z-score immediately tells you how rare a value is. Anything past is uncommon, and past is rare. That reading is the same for every normal distribution. This works because the empirical rule percentages (the , and figures) attach to the z-scale, not to the raw values.
Comparing values from different distributions
This is the headline use of z-scores and a favourite exam question. Two raw scores from different tests cannot be compared head to head, because the tests have different means and spreads. Convert each to a z-score and the comparison becomes easy. The higher z-score is the better result relative to its own group. Watch the comparison happen in three stages below.
Stage 1, the raw marks are not comparable. Anika scores on a test with , ; Ben scores on a test with , . Ben's raw mark is higher, but the two tests sit on different scales, so the raw numbers cannot settle who did better.
Stage 2, standardise each within its own distribution. Convert each mark using its own mean and standard deviation. Anika: . Ben: . Each z-score says how many standard deviations the mark is above its own mean.
Stage 3, compare on one common z-scale. Place both z-scores on a single axis. Anika at sits to the right of Ben at , so despite the lower raw mark, Anika performed better relative to her cohort. On the z-scale, further right always means better relative standing.
The same logic compares salaries across industries, sports results against different reference groups, or a child's height against age-specific charts. In each case, standardise within each distribution first, then compare the z-scores.
Z-scores and percentiles
Common percentile-to-z-score values from the standard normal table:
| Percentile | z-score |
|---|---|
| 50th | |
| 75th | |
| 90th | |
| 95th | |
| 97.5th | |
| 99th |
Because the bell curve is symmetric, a lower percentile uses the negative of the matching upper z-score. The 10th percentile is at , the 5th at , and the 2.5th at . So once you can do the top tail, the bottom tail is just a sign change.
Z-scores and probabilities
To find the probability that an observation is less than some value :
- Compute .
- Look up , the area to the left of , in the standard normal table provided in the HSC paper.
- .
For the other directions:
- (the right tail).
- (subtract the two left-areas).
When the endpoints are whole standard deviations from the mean (), you can skip the table and read the percentage straight off the empirical rule.
How exam questions ask about z-scores
The wording varies; the method does not. Translate the question:
- "Calculate the z-score / standardise the value." Apply directly and state the answer with its sign.
- "Who performed better relative to their group?" A comparison: compute a z-score for each, then say the larger z-score is better and why ("further above its own mean").
- "Find the value at the th percentile" or "... the top cutoff". Reverse the formula: get the z-score for that percentile (from the table or given), then .
- "What percentage / probability lie above (or below) ?" Standardise, then use from the table, or the empirical rule if is a whole number.
- "Between what two values do the middle lie?" Symmetric percentiles: the middle runs from to , so compute .
- "Is this value unusual?" Read the size of : beyond is uncommon, beyond is rare.
Exam-style practice questions
Practice questions written in the style of NESA exam questions on this dot point, with worked answer explainers. The year tag is the paper they imitate, not the source.
2022 HSC-style4 marksTwo students sit different tests. Anika scores on a test with and . Ben scores on a test with and . Which student performed better relative to their cohort?Show worked answer →
Compute each z-score: .
Anika: .
Ben: .
Anika's z-score is higher, so she performed better relative to her cohort (further above the mean in standard-deviation units), even though Ben's raw score is higher.
Markers reward both z-scores, comparison, and the final conclusion with the reason ("higher z-score means further above the mean").
2023 HSC-style4 marksHeights of Year 12 girls at a school are normally distributed with mean cm and standard deviation cm. Find the height that corresponds to (a) the 90th percentile, and (b) the 10th percentile. (Use for the 90th percentile.)Show worked answer →
(a) For the 90th percentile, . Use .
cm. Round to cm.
(b) By symmetry, the 10th percentile has .
cm. Round to cm.
Markers reward the formula rearrangement, the symmetric z-score for the lower tail, and answers to one decimal place.
Practice questions
Original practice questions graded from foundation to exam level, each with a full worked solution. Try them before revealing the solution.
foundation1 marksA runner's m time is seconds. The times for her squad are normally distributed with mean seconds and standard deviation seconds. Calculate her z-score.
Show worked solution →
Apply the z-score formula. Substitute , and into :
Keep the sign. Her time is below the mean, so the z-score is negative. (For a race, a faster time below the mean is a good result.)
Answer: .
foundation2 marksA student's NAPLAN-style numeracy score is . The scores are normally distributed with mean and standard deviation . (a) Calculate the z-score. (b) State how many standard deviations the score is above the mean.
Show worked solution →
Apply the z-score formula (a). Substitute , and into :
Interpret the size (b). A z-score of means the score sits standard deviations above the mean.
Answer: (a) ; (b) standard deviations above the mean.
foundation2 marksA class test is normally distributed with mean and standard deviation . A mark has a z-score of . Find the mark.
Show worked solution →
Choose the reverse formula. A z-score is given and the raw mark is wanted, so de-standardise with .
Substitute the values. Use , and :
Answer: the mark is .
foundation2 marksIQ scores are normally distributed with mean and standard deviation . (a) What z-score corresponds to the 50th percentile? (b) Use to find the IQ score at the 50th percentile.
Show worked solution →
Recall the 50th percentile z-score (a). The 50th percentile is the middle of a symmetric normal distribution, which sits exactly at the mean, so .
De-standardise (b). Substitute , and into :
Answer: (a) ; (b) the 50th percentile IQ score is (the mean).
core4 marksTwo swimmers each win their own event at a carnival, where a lower time is better. Mia swims the m event in seconds; that event has mean s and standard deviation s. Zoe swims the m event in seconds; that event has mean s and standard deviation s. Whose swim was better relative to her own event?
Show worked solution →
Standardise Mia's time. Use with her own event's mean and standard deviation:
Standardise Zoe's time. Use Zoe's event:
Compare, remembering lower is better. Both times are below their means (good for a race). Zoe's z-score of is further below the mean than Mia's , so Zoe is more standard deviations faster than her field.
Answer: Zoe's swim () was better relative to her event than Mia's ().
core3 marksMarks on a chemistry exam are normally distributed with mean and standard deviation . A scholarship is offered to students in the top . Using for the 75th percentile, find the lowest mark that earns the scholarship.
Show worked solution →
Identify the percentile (a). The top begins at the 75th percentile, which has .
De-standardise with . Substitute , and :
Interpret. A mark of marks the boundary, so a student needs about to be safely in the top .
Answer: the cutoff for the top is a mark of about .
core3 marksDaniel scores on a Paper 1 that is normally distributed with mean and standard deviation . His teacher wants the equivalent mark on Paper 2, which is normally distributed with mean and standard deviation . (a) Find Daniel's z-score on Paper 1. (b) Find the Paper 2 mark with the same z-score.
Show worked solution →
Standardise the Paper 1 mark (a). Use with , , :
De-standardise onto Paper 2 (b). The equivalent mark keeps the same z-score, so use with Paper 2's , and :
Answer: (a) ; (b) the equivalent Paper 2 mark is .
core3 marksThe resting heart rates of a group of athletes are normally distributed with mean beats per minute and standard deviation beats per minute. Using for the middle , find the two heart rates between which the middle of athletes lie.
Show worked solution →
Set up the symmetric pair. The middle runs from to , so apply at each end with and .
Lower bound ().
Upper bound ().
Answer: the middle of athletes have resting heart rates between and beats per minute.
exam4 marksIn her HSC trials, Priya scores in Mathematics, which is normally distributed with mean and standard deviation . She scores in English, which is normally distributed with mean and standard deviation . (a) Calculate her z-score in each subject, correct to two decimal places. (b) State the subject in which she performed better relative to her cohort, and justify your answer.
Show worked solution →
Standardise the Mathematics mark (a). Use with , , :
Standardise the English mark (a). Use , , :
Compare and justify (b). English has the higher z-score ( against ), so Priya sits more standard deviations above the mean in English. She performed better in English relative to her cohort, even though her raw Mathematics mark is higher.
Answer: (a) Maths , English ; (b) English, because its higher z-score puts her further above her cohort's mean.
exam5 marksAnnual salaries in finance are normally distributed with mean $ and standard deviation $. Salaries in technology are normally distributed with mean $ and standard deviation $. Sam earns $ in finance; Lee earns $ in technology. (a) Find each z-score, to two decimal places. (b) State who earns more relative to their own industry. (c) Given , find the percentage of technology workers who earn more than Lee.
Show worked solution →
Standardise Sam's salary (a). Use for finance:
Standardise Lee's salary (a). Use technology:
- Compare (b)
- Lee's z-score () is higher than Sam's (), so Lee earns more relative to his own industry despite the smaller dollar salary.
- Use the table for the tail (c)
- The area below Lee is , so the proportion above is .
- Answer
- (a) Sam , Lee ; (b) Lee; (c) about of technology workers earn more than Lee.
exam6 marksThe heights of adult men in a town are normally distributed with mean cm and standard deviation cm. (a) Calculate the z-score of a man who is cm tall. (b) Using for the 95th percentile, find the height that a man must exceed to be in the tallest . (c) Given , find the probability that a randomly chosen man is taller than cm.
Show worked solution →
Standardise the height (a). Use with , , :
Find the top cutoff (b). The tallest start at the 95th percentile, . De-standardise with :
Find the right-tail probability (c). First standardise cm: . The area below is , so the area above is .
Answer: (a) ; (b) a man must exceed about cm; (c) , about .
exam5 marksThe life of a brand of phone battery is normally distributed with mean hours and standard deviation hours. Use the values and , and recall that . (a) Find the z-score of a battery lasting hours and of one lasting hours, each to two decimal places. (b) Find the probability that a battery lasts between and hours. (c) Hence find the probability that a battery lasts either under hours or over hours. (d) In a delivery of batteries, about how many would you expect to last between and hours?
Show worked solution →
Standardise both endpoints (a). Apply with and at each value:
Convert the lower tail with the symmetry rule (b). The interval becomes . The area below the upper z is . For the lower z, use .
Subtract to get the between-area (b). The probability between the two values is the difference of the two areas:
Take the complement for the two outer tails (c). Lasting under or over hours is everything outside the interval, so subtract from :
Scale to the delivery (d). Multiply the between-probability by the number of batteries: , so about batteries.
Answer: (a) , ; (b) (about ); (c) (about ); (d) about batteries.
exam6 marksA machine stamps out metal washers whose thickness (in microns) is normally distributed with unknown mean and unknown standard deviation . Quality records show that the 92nd percentile thickness is microns (for which ) and the 31st percentile thickness is microns (for which ). (a) By writing at each percentile, set up two equations and find . (b) Hence find . (c) Using for the middle , find the two thicknesses between which the middle of washers lie.
Show worked solution →
Write an equation at each percentile (a). Substitute each known thickness and z-score into :
Eliminate by subtracting (a). Subtracting the second equation from the first cancels :
Back-substitute for the mean (b). Put into the first equation:
A quick check with the second equation agrees: .
Build the symmetric middle- interval (c). The central runs from to , so apply at each end with and :
Answer: (a) microns; (b) microns; (c) the middle of washers have thicknesses between and microns.
exam6 marksIn a course, a student scores on the Major Project, which is normally distributed with mean and standard deviation . The same student scores on the Written Exam, which is normally distributed with mean and standard deviation . (a) Calculate the student's z-score on each task, to two decimal places. (b) State which task was the stronger result relative to the cohort, and justify your answer. (c) Given , find the probability that a randomly chosen student scores higher than on the Written Exam, and hence, in a cohort of students, about how many score above this student on that task. (d) The student claims the equal-looking marks mean the two performances were equally good. Explain why this reasoning is flawed.
Show worked solution →
Standardise the Major Project mark (a). Apply with , , :
Standardise the Written Exam mark (a). Apply the formula with , , :
- Compare the relative standings (b)
- The Major Project has the higher z-score ( against ), so the student sits further above the mean, in standard-deviation units, on the Major Project. That was the stronger result relative to the cohort.
- Find the right-tail probability (c)
- Scoring above on the Written Exam means . The area below is , so the area above is . In a cohort of , the expected number above is , so about students.
- Explain the flaw (d)
- Raw marks from two different tasks are not comparable, because the tasks have different means and standard deviations. Equal-looking raw marks can sit at different distances above their own means once standardised, so the z-scores ( and ) show the performances were not equally good.
- Answer
- (a) Project , Exam ; (b) the Major Project, because its higher z-score places the student further above the cohort mean; (c) (about ), so roughly of the students; (d) raw marks from different distributions are not directly comparable, and the z-scores show the Major Project was the better result.
