Median, Quartiles and Percentiles from a Curve
Reading the median, the lower and upper quartiles and any percentile from a cumulative frequency curve; the interquartile range; and what the steepness of the curve says about the spread of the data.
What You Should Already Know
- From Secondary 2: the median is the middle value of the data in order; for values in a list it is at position . For 7 marks in order, 68, 70, 75, 79, 82, 88, 91, the median is the 4th value, 79.
- From lesson 3.1: a cumulative frequency curve is plotted at the upper class boundaries, from (lowest class boundary, 0). To find how many values are less than , draw a line up from to the curve and across. The 80 masses of 3.1 (frequencies 3, 8, 16, 21, 19, 13 in classes of 5 kg from 70 kg to 100 kg) are used again here.
The Median and the Quartiles
In lesson 3.1 you read the curve from a value to a number of values: up from the value, across to the cumulative frequency. Now read it the other way: from a number of values to a value, across from the cumulative frequency to the curve, then down to the value axis.
The median has half of the values below it. On a curve of values, it is the value at a cumulative frequency of . In the same way the curve is cut into quarters by two more values, the quartiles.
The median and the quartiles from a curve of n values
The lower quartile is the value at a cumulative frequency of : about a quarter of the values are below it.
The median is the value at a cumulative frequency of : about half of the values are below it.
The upper quartile is the value at a cumulative frequency of : about three quarters of the values are below it.
To read each one: draw a line across from the cumulative frequency to the curve, then down to the value axis, and read the value there. It is an estimate: write "about" or "≈".
Why n/2 and not (n + 1)/2
For a list of values, the median is at position , because with values in order, the middle one has as many values before it as after it. A curve does not show the single values: it takes them to be spread smoothly, and the value with half of them below it is at a cumulative frequency of . For 80 values, 40 and 40.5 are a quarter of a small square apart, far less than a reading can show. On a curve use , and .
Example 1: The median of the 80 masses
Use the cumulative frequency curve of the masses of the 80 basketball players to estimate the median mass.
Solution
, so the median is at a cumulative frequency of .
Draw a line across from 40 to the curve, then down to the mass axis. One small square across is 1 kg.
The line meets the mass axis at about 88 kg. Check with the table: 27 players are under 85 kg and 48 under 90 kg, so the 40th lies between 85 kg and 90 kg, as the reading does.
The median mass is about 88 kg.
Example 2: The quartiles and the interquartile range
Use the curve of the 80 masses to estimate (a) the lower quartile, (b) the upper quartile, (c) the interquartile range.
Solution
(a) is at a cumulative frequency of . (b) is at .
Draw lines across from 20 and 60 to the curve and down. The median line from 40 is drawn too, so the four quarters can be seen.
(a) kg. (b) kg.
(c) The interquartile range is kg.
So about 20 players weigh less than 83 kg, about 20 from 83 kg to 88 kg, about 20 from 88 kg to 93 kg and about 20 more than 93 kg. The middle half of the players, about 40 of them, have masses within a span of about 10 kg.
(a) About 83 kg (b) about 93 kg (c) about 10 kg
The Interquartile Range
Interquartile range
Interquartile range upper quartile lower quartile
It is the spread of the middle half of the data: about half of the values lie between and .
A small interquartile range means that the middle half of the values are close together; a large one means that they are spread out.
In Secondary 2 you measured spread by the range, the largest value minus the smallest. A cumulative frequency curve of grouped data does not show the largest and smallest values: the 80 masses lie somewhere from 70 kg to 100 kg, but perhaps nobody weighs exactly 70 kg, and nobody can weigh 100 kg, since the last class stops below it. The interquartile range can still be read from the curve, and it does not depend on one very large or very small value. Lesson 3.3 finds the quartiles and the interquartile range of a list.
Percentiles
The quartiles cut the data into four parts. Percentiles cut it into a hundred. They are often used to say where one value stands in a large group: a child's height on a growth chart, or a mark in a national test.
Percentiles
The th percentile is the value below which about of the data lie.
On a curve of values, read it at a cumulative frequency of .
is the 25th percentile, the median the 50th and the 75th.
If your mark is at the 75th percentile, about 75% of the group scored less than you.
Example 3: The 90th and the 10th percentiles
Use the curve of the 80 masses to estimate (a) the 90th percentile, (b) the 10th percentile. (c) What do the two answers tell a coach?
Solution
(a) . (b) .
One small square is 1 kg. Both lines meet the mass axis about halfway across a small square, so read them to the half kilogram. (a) Across from 72 and down: about 96.5 kg. (b) Across from 8 and down: about 78.5 kg.
(c) About 10% of the players (8 of them) weigh less than 78.5 kg, and about 10% weigh more than 96.5 kg. The other 80% have masses between about 78.5 kg and 96.5 kg.
(a) About 96.5 kg (b) about 78.5 kg (c) About 80% of the players weigh between 78.5 kg and 96.5 kg.
Example 4: The percentile of a value
Jun Hao is one of the 80 players. His mass is 92 kg. At about which percentile is his mass?
Solution
This is the reading of lesson 3.1: up from 92 kg to the curve and across.
About 56 players weigh less than 92 kg. As a percentage of the 80 players:
His mass is at about the 70th percentile: about 70% of the players weigh less than he does.
Example 5: An award for the top 10%
The curve shows the marks of 120 pupils in a mathematics test. (a) Estimate the median and the interquartile range. (b) The top 10% of the pupils receive a certificate. Estimate the lowest mark that earns one.
Solution
(a) : at 30, the median at 60, at 90. On the marks axis a small square is 5 marks; read to the nearest mark.
Median marks. Interquartile range marks.
(b) The top 10% are above the 90th percentile, at a cumulative frequency of
About 108 pupils scored less than 83 marks, so the top 12 pupils scored about 83 or more.
(a) Median about 57 marks; interquartile range about 28 marks. (b) About 83 marks.
What the Shape of the Curve Says
Where the curve is steep, the cumulative frequency rises quickly: many values lie in a short stretch of the value axis. Where it is gentle, few values lie in a long stretch.
If the middle part of a curve is steep, the middle half of the data is bunched together: and are close, and the interquartile range is small. If the curve rises gently all the way, the data are spread out and the interquartile range is large.
Example 6: Two classes on one grid
The curves show the marks of the 40 pupils of class A and the 40 pupils of class B in the same test. (a) Estimate the median and the interquartile range of each class. (b) Describe how the curves differ, and what that says about the marks.
Solution
(a) for each class: at 10, the median at 20, at 30.
Both curves pass through (50, 20), so both medians are about 50 marks.
A small square is 5 marks. Class A's two lines meet the marks axis halfway across a small square: , , interquartile range marks. Class B: , , interquartile range marks.
(b) Class A's curve is steep in the middle: 24 of its 40 pupils scored from 40 to under 60. Its marks are bunched about the median. Class B's curve rises gently all the way: its marks are spread across the whole range.
(a) Class A: median about 50, interquartile range about 15 marks. Class B: median about 50, interquartile range about 42 marks. (b) The two classes did equally well on average, but class A's marks are much closer together (its curve is steeper in the middle and its interquartile range smaller).
Common Mistakes
- Giving the cumulative frequency as the answer ("the median is 40"). 40 is where you start; the median is the value read on the other axis.
- Reading the median at on a curve. On a curve use , and ; is for a list.
- Finding a percentile at a cumulative frequency of instead of . For the 90th percentile of 80 values, read at 72, not at 90.
- Giving the interquartile range as two numbers ("83 to 93"). It is one number, .
- For "the top 10%", reading at a cumulative frequency of 10% of . The top 10% are above the 90th percentile.
- Writing a reading as exact. Every value read from the curve is an estimate.
Practice Questions
Try each question first, then open the solution to check your working. Readings from a curve are estimates: give them to the nearest whole number (or the nearest half, where a line meets the axis halfway across a small square of 1 unit), with "about" or "≈". Your readings may differ from ours by about half a small square; that is fine.
Question 1
The curve shows the heights of 50 seedlings. Estimate (a) the median, (b) the lower and upper quartiles, (c) the interquartile range.
ShowHide solution
: at , the median at 25, at .
A small square is 1 cm. The line meets the height axis halfway between 7 and 8.
(a) Median cm. (b) cm, cm. (c) About cm.
(a) About 11 cm (b) about 7.5 cm and 15 cm (c) about 7.5 cm
Question 2
The curve shows the times of 200 runners in a 10 km run. (a) Estimate the median time. (b) Find the lower quartile, and estimate the upper quartile and the interquartile range. (c) The fastest 10% of the runners receive a medal. Estimate the slowest time that earns a medal.
ShowHide solution
(a) and (b) : at 50, the median at 100, at 150. (c) The fastest 10% took the shortest times: the 10th percentile, at .
(a) Median minutes.
(b) 50 is the cumulative frequency at the plotted point (60, 50), so minutes from the table. minutes, and the interquartile range minutes.
(c) About 53 minutes. Here the 10% wanted are the fastest, the smallest times, so the reading is the 10th percentile, not the 90th.
(a) About 67 minutes (b) 60 minutes; about 75 minutes; about 15 minutes (c) about 53 minutes
Question 3
Use the curve of the 120 marks in Example 5. (a) Estimate the 80th percentile. (b) Mei Ling scored 75 marks. At about which percentile is her mark?
ShowHide solution
(a) . (b) Up from 75 to the curve and across.
(a) The line across from 96 meets the curve above about 75 marks.
(b) This is the same line read the other way: about 96 pupils scored less than 75, and
(a) About 75 marks (b) about the 80th percentile
Question 4
Use the curve of the 80 masses (Example 1). 60% of the players have a mass of less than kg. Find .
ShowHide solution
is the 60th percentile: . Draw a line across from 48 to the curve and down.
The cumulative frequency 48 belongs to the plotted point (90, 48): the table says that 48 players weigh less than 90 kg. So , read exactly.
Question 5
Wei Jie found the median time of the 200 runners in Question 2 like this: "Half of 200 is 100, so the median is 100 minutes." Explain his mistake and give the right answer.
ShowHide solution
100 is the cumulative frequency at which the median is read: the number of runners faster than it. It is not a time, and 100 minutes is not even on the time axis, which ends at 90 minutes.
He should draw a line across from 100 to the curve and down to the time axis: about 67 minutes (Question 2).
100 is the position on the cumulative frequency axis, not the median. The median is about 67 minutes.
Question 6
Use the curve of the 60 shoppers below. (a) Estimate the median time and the interquartile range. (b) Estimate the 80th percentile. (c) About how many shoppers spent a time between the lower and upper quartiles?
ShowHide solution
(a) : at 15, the median at 30, at 45. A small square across is 2 minutes.
(a) Median minutes. Interquartile range minutes. The line meets the axis a little to the right of 38, so and an interquartile range of 18 minutes are fair readings too.
(b) : across from 48 to the curve and down.
About 41 minutes.
(c) About half of them: shoppers.
(a) Median about 29 minutes; interquartile range about 17 minutes (b) about 41 minutes (c) about 30
Question 7
Can the range of the 80 masses be found from the curve? What can be said about it?
ShowHide solution
No. The curve starts at 70 kg and ends at 100 kg, the lowest and highest class boundaries, but the lightest player need not weigh exactly 70 kg, and the heaviest weighs less than 100 kg. The smallest and largest values are not shown.
All that can be said is that the range is less than kg. The interquartile range, about 10 kg, can be read.
No: the range is less than 30 kg, but the curve does not show the smallest and largest masses.
Question 8
The curves show the lengths of 100 fish caught in each of two ponds, P and Q. (a) Estimate the median and the interquartile range of each pond. (b) Which pond's fish are more alike in length? Give a reason from the curves.
ShowHide solution
(a) : at 25, the median at 50, at 75. A small square across is 1 cm.
Pond P: , median , , interquartile range cm.
Pond Q: (the plotted point (20, 25)), median (the two curves cross at a cumulative frequency of 50), , interquartile range cm.
(b) Pond P. Its curve is much steeper in the middle (70 of its fish are from 20 cm to under 30 cm), and its interquartile range, about 6 cm, is about half of pond Q's, 12.5 cm. The medians are about the same.
(a) P: median about 26.5 cm, interquartile range about 6 cm; Q: median about 26.5 cm, interquartile range about 12.5 cm. (b) Pond P: its middle half of lengths are closer together.
Question 9
For the 80 masses, Priya says: "The 25th percentile is the mass at a cumulative frequency of 25." Is she right? Explain.
ShowHide solution
No. The 25th percentile is read at , not at 25. The two numbers agree only when there are 100 values.
Read at 20, the 25th percentile is the lower quartile, about 83 kg (Example 2).
No: it is read at 20, giving about 83 kg, the lower quartile.
Question 10
Two cumulative frequency curves for the same number of values have the same median. Curve X is steep in the middle and curve Y rises gently. Without any numbers, which has the larger interquartile range? Explain.
ShowHide solution
The quartiles are read across from and , the same heights for both curves.
Between those heights curve X climbs steeply, so it rises from to over a short stretch of the value axis: its quartiles are close together. Curve Y needs a long stretch to rise the same amount.
Curve Y: a gentle curve spreads the middle half of the data over a wider interval.
Key Points
- On a curve of values: at , the median at , at ; read across to the curve and down.
- The th percentile is read at ; about of the values are below it.
- , the median and are the 25th, 50th and 75th percentiles.
- Interquartile range : the spread of the middle half of the data.
- A steep middle means the data are bunched (small interquartile range); a gentle curve means they are spread out (large interquartile range).
- Every value read from a curve is an estimate, except at a plotted point.