Statistics › Section B · Descriptive statistics › Topic B.4

Box plots

Five numbers, drawn to scale, can show a whole distribution's centre, spread, skew and outliers in one compact picture - and line several groups up for comparison at a glance.

The five-number summary, drawn

Percentiles locate values one at a time; the box plot locates a whole distribution at once. It is simply the five-number summary - minimum, Q1, median, Q3, maximum - drawn to scale on a number line, so one compact picture shows centre, spread, skew and outliers together. Because the picture is so compact, several box plots stack neatly on the same scale, which makes them the standard tool for comparing groups - test scores by class, wait times by clinic, harvests by field.

The five-number summary

Five values summarise a distribution's position: the minimum (smallest value), the first quartile Q1 (25th percentile), the median (50th), the third quartile Q3 (75th) and the maximum (largest). Each of the four segments between consecutive marks - min to Q1, Q1 to median, median to Q3, Q3 to max - holds roughly a quarter of the data. The five numbers are the entire recipe for a box plot, and they also give two spread measures as simple differences: the IQR from the previous topic (Q3 minus Q1) and the range (max minus min).

Drawing a box plot

Construction takes four marks on a number line. 1. Draw a box from Q1 to Q3: its width is the IQR, the middle half of the data. 2. Draw a line inside the box at the median. 3. Extend whiskers from each box end to the most extreme values - or, in the outlier convention, only to the most extreme values inside the 1.5 × IQR fences. 4. Plot any values beyond the fences as separate dots. Under the plain convention the whisker tips are the minimum and maximum; under the outlier convention they are the extreme non-outliers, and the dots carry the rest.

Reading a box plot

Each quarter of the plot - whisker, box-half, box-half, whisker - holds about 25 percent of the data, so the shares are equal and the lengths compare spread. A long lower whisker means the lowest quarter stretches far; a long upper whisker means the top quarter does. The median line's position inside the box signals direction: pushed toward the low end, it means the upper half of the middle 50 percent is wider, a hint of a high tail; pushed toward the high end, it suggests a low tail. Tight sections mean crowded values; long sections mean scattered ones.

Worked example: one data set, one box plot

Delivery times of 15 orders, sorted (minutes): 5, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 20, 35. 1. Five-number summary, using the position rule \(i = \frac{k}{100}(n+1)\) with \(n + 1 = 16\): min 5. Q1: \(i = 0.25 \times 16 = 4\), the 4th value, 9. Median: \(i = 8\), the 8th value, 13. Q3: \(i = 12\), the 12th value, 17. Max 35. 2. \(\text{IQR} = 17 - 9 = 8\); fences at \(9 - 12 = -3\) and \(17 + 12 = 29\). 3. The value 35 exceeds 29: draw it as a dot and run the upper whisker only to 20. 4. Check: the whisker stops at the largest non-outlier, and the dot carries the one unusual order.

Worked example: comparing two classes

Two classes take the same test. Class A: min 42, Q1 55, median 70, Q3 82, max 95. Class B: min 58, Q1 66, median 71, Q3 76, max 84. 1. Draw both on the same scale, boxes above one shared axis. 2. Centres: medians 70 and 71 - nearly identical. 3. Spreads: A's IQR is \(82 - 55 = 27\); B's is \(76 - 66 = 10\). 4. Reading: the classes performed similarly in the middle, but A's middle half ranges widely while B's is tightly packed; A's whiskers also reach further, down to 42 and up to 95. 5. Check: same centre, very different consistency - a comparison that a single class average would have hidden.

Odd shapes are honest

Real data produce lopsided boxes. If the median equals Q3, the median line sits on the box's upper edge - not a mistake, it means roughly a quarter of the values equal that number. If the minimum equals Q1, the lower whisker vanishes. Tied values bunch the data, and the quarter-share reading becomes approximate: a short-looking segment may hold several identical values and still count a quarter of the data. Small data sets give chunky plots, since the box is drawn from just a few order statistics. None of this is an error; the picture reports exactly what the five numbers say.

What the box plot hides

The compactness has a price: individual values disappear. Two quite different data sets can share one five-number summary, and no box plot can tell you which one you have. A dot plot or stemplot keeps every value but becomes cluttered past a few dozen observations; a box plot survives any size but shows only five numbers plus outlier dots. Choose by size and purpose: for small data where exact values matter, use a dot plot or stemplot; for comparing groups of any size, use box plots on a shared scale.

Common mistakes to avoid

First, drawing the box from the minimum to the maximum instead of Q1 to Q3 - the box is the middle half. Second, putting the mean at the box's centre line: the line is the median. Third, stretching whiskers out to the fences themselves - fences are thresholds for outlier dots, not whisker ends. Fourth, reading a longer box half as "more data": each quarter holds roughly 25 percent, so length means spread, not count. Fifth, comparing box plots drawn on different scales - line them up on one axis, or the comparison is meaningless.

Key takeaways

The five-number summary - minimum, Q1, median, Q3, maximum - is the box plot's entire recipe. The box spans Q1 to Q3 (the IQR), the inner line is the median, and whiskers run to the extremes, or to the extreme non-outliers once values beyond the 1.5 × IQR fences are drawn as dots. Each of the four sections holds about a quarter of the data, so lengths compare spread, the median's position hints at skew, and several plots on one shared scale compare groups at a glance. The price is that individual values vanish, so small data sets may deserve a dot plot instead.

Flashcards

Flashcard

Five-number summary

Minimum, Q1, median, Q3, maximum

Everything a basic box plot needs.

Test your understanding

Question 1 of 5

In a box plot, the box spans:

Question 2 of 5

The line inside the box of a box plot marks:

Question 3 of 5

Roughly what percent of the data falls between Q1 and the median?

Question 4 of 5

On a box plot that draws outlier dots, the whiskers extend to:

Question 5 of 5

Class A's box plot shows an IQR of 27; Class B's shows an IQR of 10. Which class's middle half is more tightly packed?

← Previous topicNext topic →