The Core Aim of Education | How to Read a Split Stem-and-Leaf Plot Without Losing Repeated Values

Two students can look at the same stem-and-leaf diagram, count a different number of scores and both think their method is obvious. The trouble often begins when each stem appears twice. Does the second row repeat the data, or does it divide the data into smaller groups?

A split stem-and-leaf plot repeats a stem label but divides its leaves into two non-overlapping bands, usually 0–4 and 5–9. Each leaf still represents one observation. Repeated leaves represent repeated observations, not typing errors. Read the key first, identify the low and high rows, then count every leaf exactly once.

Did you know that a stem-and-leaf plot can display the actual recorded values while also showing their distribution? A bar chart groups counts visually; a properly labelled stem-and-leaf plot lets you reconstruct each number. That makes it particularly useful when a student must find the median or check whether a result has been omitted.

Start with the reader’s diagnosis, not a memorised method

Place a diagram in front of a learner and ask: “What does 4 | 7 mean? Why might I write the stem 4 twice? If two leaves both say 4, how many measurements have I recorded?” These questions reveal three distinct gaps: understanding place value, understanding split rows and respecting frequency. A student who says “4 | 7 is 4.7” may simply be using the wrong key. A student who counts two 4s as one may misunderstand how a statistical display preserves duplicates.

The key is part of the mathematical data. It may say 4 | 7 represents 47 minutes, or it may say 4 | 7 represents 4.7 kilograms. The same written marks do not have one universal meaning. Check the unit, stem and leaf place value before calculating anything.

A complete worked example: journey times for a school event

Imagine an illustrative, invented set of 18 journey times, in whole minutes. The values are 31, 33, 33, 35, 37, 38, 40, 41, 44, 44, 45, 47, 49, 51, 52, 52, 56 and 59. No real students or tracked journeys are represented. The purpose is to learn the method with a dataset whose calculations can be checked by hand.

Step 1: Keep the units and the full original list

The smallest time is 31 minutes and the largest is 59 minutes. Their difference is 59 − 31 = 28 minutes, so the range is 28 minutes. There are 18 entries, not 15, because 33, 44 and 52 each occur twice. Do not remove their repetitions: those observations matter when counting the number of journeys and finding a median.

Step 2: Make a key and assign each leaf to its band

We use the tens digit as the stem, the ones digit as the leaf and a key of 4 | 4 = 44 minutes. Each stem gets a lower row for leaves 0, 1, 2, 3 and 4, and an upper row for leaves 5, 6, 7, 8 and 9. The rows must stay in that order. An entry of 40 belongs on the lower 4 row because the leaf 0 is a genuine digit.

Split diagram, with each row described explicitly:

Stem 3, low row (0–4): 1 3 3
Stem 3, high row (5–9): 5 7 8
Stem 4, low row (0–4): 0 1 4 4
Stem 4, high row (5–9): 5 7 9
Stem 5, low row (0–4): 1 2 2
Stem 5, high row (5–9): 6 9

The plot is intentionally presented as text with each band named. Readers who cannot distinguish a repeated stem visually still receive the band information. On paper, draw the stem column and divider, then write the leaves to its right in ascending order. Add a legend that explicitly says each stem is split into low and high rows; a plain “4 | 4 = 44” key alone does not explain which duplicate stem is which.

Step 3: Rebuild the original observations from the graph

Read from top to bottom. The first row gives 31, 33 and 33. The next gives 35, 37 and 38. The lower 4 row produces 40, 41, 44 and 44; the upper 4 row gives 45, 47 and 49. The 5 rows yield 51, 52, 52, 56 and 59. Every reconstructed number matches the original list. That reverse-check is powerful because a good diagram must be decodable, not merely tidy.

Step 4: Count frequency and calculate the median correctly

Count the leaves by row: 3 + 3 + 4 + 3 + 3 + 2 = 18 observations. Because 18 is even, the median is the mean of the 9th and 10th observations in ascending order. Counting carefully, both the 9th and 10th values are 44 minutes, so the median is (44 + 44) ÷ 2 = 44 minutes. The range remains 28 minutes. Three values share the highest frequency of two: 33, 44 and 52 minutes. The dataset therefore has three modes; it does not have a single unique mode.

Why the same stem appears twice

In a normal one-row-per-stem display, all values in the 40s occupy one line. If many scores lie in that band, the line becomes crowded and hides patterns. Splitting each stem creates a finer display without changing the dataset. We gain a closer view of whether observations cluster in the early or late part of a ten-unit interval.

The extra line does not double the sample size. It is not a second group of learners unless the source explicitly says so. The two lines are mutually exclusive leaf bands covering one stem. Confusing presentation rows with population groups is a common, understandable misreading—and one reason to insist on a labelled legend.

Four errors that change the answer

Ignoring the zero leaf: 4 | 0 means 40 minutes under our key, not a blank or a missing score.
Deleting repeated leaves: two 4 leaves on the lower 4 row mean two values of 44, not one value typed twice by mistake.
Putting 5 on the low row: the chosen bands are 0–4 and 5–9; the 5 begins the high row.
Using row numbers as observations: two lines labelled 4 do not themselves count as two measured values; only their leaves do.

A fifth, subtler mistake is confusing split stems with a back-to-back stem-and-leaf plot. A back-to-back display shows two separate sets of leaves on opposite sides of a central stem, often comparing two groups. A split stem changes the intervals within one dataset. The layout and question both matter.

Guided practice: decode before calculating

Try this small example: key 2 | 3 = 23 pages. Stem 2 low leaves 1 2 2; stem 2 high leaves 5 9; stem 3 low leaves 0 3; stem 3 high leaves 6. Before reading on, say how many books were recorded and which page count was repeated.

The reconstructed list is 21, 22, 22, 25, 29, 30, 33 and 36 pages: eight observations, with 22 appearing twice. The range is 36 − 21 = 15 pages. The median is the average of the 4th and 5th values, (25 + 29) ÷ 2 = 27 pages. Notice how the key defines pages, not minutes, and how the repeated 22 changes the order.

Independent practice and an answer-check habit

Now use a new dataset without copying the finished plot: 12, 15, 18, 18, 20, 21, 24, 27, 31 and 34. Construct split stems 1, 2 and 3, each with low and high rows. Mark the stems and leaves, write the key, count all observations, find the range and decide whether there is a mode. Do not look at the next paragraph until you have reconstructed every entry from your finished plot.

Self-check: there are ten observations. The low 1 row contains 2; the high 1 row contains 5 8 8; the low 2 row contains 0 1 4; the high 2 row contains 7; the low 3 row contains 1 4; the high 3 row is empty. The range is 34 − 12 = 22; 18 is the single mode. The median is (20 + 21) ÷ 2 = 20.5. Keeping the empty final high row makes the split convention visually consistent, although a well-labelled display may omit a trailing empty line.

What visible progress looks like

After one explanation, the learner may decode a labelled plot when the original list is alongside it. Next, remove the list and see whether they can reconstruct the values. Later, switch the unit from minutes to centimetres or decimal kilograms while keeping the same underlying method. Independence means they check the key, bands and count before being prompted.

Parents can ask the child to explain one leaf aloud instead of marking a whole diagram right or wrong. Teachers can give a deliberate duplicate and a zero leaf as diagnostic items. Tutors can use a quick reverse-encoding task: supply the plot, then ask students to list the observations and check whether the count matches. This is more revealing than another page of identical median questions.

Limitations: when a different graph may serve better

A stem-and-leaf plot is easiest for relatively small datasets with sensibly chosen leaf units. For a very large dataset or values spanning too many stems, it becomes unwieldy. A histogram may show broad frequency patterns more clearly, while a box plot summarises spread without retaining individual observations. A split stemplot also needs a clear convention for negative values and decimals; do not reuse this integer key without adapting it.

Some textbook methods for quartiles differ in how they divide an ordered list, especially when the sample size is odd. The median and range in this worked example are unambiguous. If you later calculate quartiles, state and follow your course’s method consistently rather than claiming there is only one convention.

Frequently asked questions

Does every stem need two non-empty rows?

No. A low or high row can be empty if the dataset has no values in that band. The important point is to keep the splitting convention clear. An empty band does not mean an unrecorded observation.

Should I round data before making the plot?

Only if the data-handling task explicitly requires rounding or sets a leaf unit. Otherwise rounding can merge distinct measurements and conceal information. Write the leaf unit and key so the reader knows what precision the plot represents.

Can a repeated leaf be an error?

It could be a transcription error in the original data, but the diagram itself cannot establish that. Verify against the source list before changing a measurement. If two learners genuinely recorded the same value, both observations belong in the graph.

Is a split stem-and-leaf plot better than a box plot?

Neither is universally better. A stem-and-leaf plot lets you recover individual observations and duplicates; a box plot quickly communicates median and quartile structure. Choose based on whether the reader needs exact values or a compressed comparison.