Is it the same as a histogram?

No. Both show distributional shape, but a stem and leaf plot retains individual displayed values.

Can a value appear twice?

Yes; identical observations need repeated leaves.

Does every dataset suit this plot?

No. Large samples or awkward scales may need another display.

AI Glossary / Statistics and probability / Beginner

Stem and leaf plot

Definition

A stem and leaf plot is a display of numerical observations that separates each value into a leading part, the stem, and a trailing part, the leaf. Leaves are listed beside their stems so the overall distribution is visible while individual values remain recoverable at the displayed precision. It is useful for exploring small or moderate datasets, especially when readers need to see both the pattern and the underlying numbers.

What it means

For whole numbers such as 24, a convenient convention uses 2 as the stem and 4 as the leaf. The row 2 | 4 7 7 then represents 24, 27, and 27. A key is essential because the same symbols could represent 2.4, 2.7, and 2.7 under another scale.

Longer rows indicate more observations in that interval, giving a histogram-like view without replacing all observations by a frequency count. Repeated leaves must remain visible: repetition is information about frequency, not a formatting error. The display is descriptive and does not by itself establish a probability model or explain why a pattern occurred.

How it works

Choose a stem and leaf scale suited to the range and precision of the data. Sort the observations, place stems in numerical order, and list the corresponding leaves in order within each row. Include empty stems when they help show gaps in the distribution.

Add a key that states the represented value and measurement unit. If values have been rounded or truncated before plotting, explain that choice; recovery is then limited to the displayed precision. Check that the number of leaves equals the number of observations represented.

If rows are too crowded, split each stem into two rows, commonly for leaves zero through four and five through nine. Label or explain that convention. If there are too many stems or observations, another display may communicate the data more clearly.

Use the finished plot to inspect concentration, spread, gaps, and unusually distant values. Follow up apparent anomalies against the original records. A long or isolated leaf row can prompt investigation, but it does not diagnose recording errors, outliers, or process changes on its own.

Three practical examples

1. A trainer displays scores from a small workshop so participants can see the distribution and identify the middle observations directly.

2. An operations analyst reviews a short set of service durations and notices a gap before several long cases that deserve investigation.

3. An AI team inspects rounded response times from a small pilot, retaining the individual times while checking whether a few slow requests dominate the experience.

Think of it this way

Think of sorting house numbers into streets labeled by their leading digits. The street labels organize the collection, while each house number's final digit remains visible. The organization makes the crowding easy to see without replacing every house by an anonymous tally mark.

Worked example

Take this original dataset of ten service durations, in whole minutes: 12, 14, 14, 18, 21, 23, 27, 31, 33, and 39. Use tens as stems and units as leaves. The display has these rows:

1 | 2 4 4 8

2 | 1 3 7

3 | 1 3 9

The key is 2 | 3 = 23 minutes. There are four leaves in the first row, three in the second, and three in the third, totaling ten observations. The repeated 4 in the first row correctly preserves two separate 14-minute cases.

The fifth and sixth ordered values are 21 and 23, so the median is (21 + 23) / 2 = 22 minutes. The range is 39 − 12 = 27 minutes. Four of the ten durations are below 20 minutes, giving a descriptive proportion of 40%.

If each stem were split, the first stem would have a lower row containing 2, 4, 4 and an upper row containing 8. The observations would be unchanged; only the display would become more detailed. None of these calculations establishes how future service times will behave.

Illustrative case study

In a fictional business, Cedar Helpdesk compares the response times from an initial AI-assisted pilot. A single mean obscures the spread, so the analyst adds a stem and leaf plot for the small sample. Staff can recover individual rounded times and identify several requests that took longer than expected. They inspect those cases and discover a shared document-format issue. The plot helps direct investigation, but the team records the finding as exploratory and tests the suspected cause separately before changing the service process.

Common misconceptions and mistakes

The stem is not automatically the tens digit: its meaning depends on the key. Decimal values, large numbers, and negative numbers require conventions that prevent ambiguity.

Removing repeated leaves changes the data. Sorting the leaves improves readability but must preserve every observation being represented.

A stem and leaf plot does not necessarily preserve all original precision if values were rounded first. It also becomes unwieldy for very large samples. Choosing a histogram, empirical distribution plot, or another display can be more informative than forcing thousands of leaves into a crowded layout.

Business use

This plot is useful in workshops, small audits, pilot reviews, and teaching because readers can check the arithmetic against visible observations. It makes descriptive summaries more transparent without requiring an elaborate dashboard.

For business communication, explain the unit, sample scope, and key beside the display. Avoid exposing sensitive individual measurements where disclosure matters. Use the plot as an exploratory aid and combine it with a suitable comparison or sampling framework when a decision requires conclusions beyond the observed cases.

Check your understanding

With the key 4 | 2 = 4.2 seconds, what does 4 | 2 2 8 represent?

Answer: Three observations: 4.2, 4.2, and 4.8 seconds. It does not represent a single number 4.228.

Related terms

Learn next

Compare stem and leaf plots with histograms and dot plots. Then examine how display choices and rounding affect the patterns readers notice.

Sources

NIST, Dataplot Reference Manual, Stem and Leaf Plot: https://www.itl.nist.gov/div898/software/dataplot/refman1/ch2/stemleaf.pdf . This primary manual describes the distributional purpose and suitability for smaller datasets. The displayed dataset, calculations, and fictional case are original teaching examples.

Previous: Stein discrepancy Next: Stemming