Reading Stemplots for Shape and Unusual Values

Reading the Pattern in Ordered Leaves

A stemplot preserves the individual data values while arranging them in order. That makes it possible to see both the overall pattern and the observations behind it. Reading one well means looking beyond the shape of the rows: use the key to recover the actual values, then support descriptions of concentration, skewness, clusters, gaps, and unusual observations with those values.

A stem represents a range or group of values, and each leaf represents one observation within that group. In a common format, the stem gives the leading digit or digits and the leaf gives the final digit. The key specifies the scale: for example, \(2\mid 7=27\) means that stem 2 and leaf 7 together represent 27. Read the key before describing the data, especially when values are decimals or stems have been split.

Begin by checking the key and the ordered leaves. Then look at how many observations fall in each range, where the data are most concentrated, and how the distribution extends away from its main cluster. Rows with many leaves show more observations in their ranges than rows with few leaves. Empty stems can indicate an interval with no observations, but the actual gap is best described by citing the nearest observed values on either side.

Concentration and Overall Shape

A distribution’s shape is its overall pattern across the values. A stemplot can show a single main concentration (often called unimodal) or more than one concentration (often called multimodal). If the leaves thin out on both sides of the main concentration in a broadly similar way, the distribution may be described as roughly symmetric. These are descriptions, not demands that every value mirror another exactly.

For a skewed distribution, one side extends farther from the main concentration than the other. The direction of skewness names the direction of that longer tail: a longer tail toward higher values is right-skewed, and one toward lower values is left-skewed. The peak or main cluster is not itself the tail. A few distant values may contribute to an apparent tail, so describe what is visible and cite the values rather than relying on a label alone.

Clusters, Gaps, and Sparse Ranges

A cluster is a group of observations concentrated in a particular part of the scale. A gap is a stretch of values without observations, separating parts of the data or leaving an isolated value. In a stemplot, an entirely empty stem is a useful clue, but a gap can also occur within a stem. Because a stem covers a range, stating the nearest observed values gives a clearer account than saying only that “there is a gap.”

Sparse rows are not necessarily gaps. A stem with one or two leaves still contains observations. Nor does a plot show why a gap or cluster occurred. It records the data values; explanations require information about how the data were collected and what they represent.

Unusual Observations and Long Tails

An observation is visually unusual when it lies apart from the main body of leaves, perhaps across a sizable gap or at a far end of the scale. Describe its value and its relation to the rest of the distribution. For example, say that one observation is much higher than the main group, and state the values that make that separation visible. Avoid calling every endpoint unusual: a largest or smallest value is not necessarily far from the others.

An unusual observation is not automatically a mistake, an outlier under a formal rule, or evidence that the value should be removed. A stemplot alone cannot establish whether the measurement is correct or whether there is a substantive reason for it. Those judgments require checking the data and understanding their context.

A Justification: Why Leaf Counts Show Concentration

Each leaf represents one observation, and the stem places that observation in a particular range. Therefore, counting the leaves on a stem counts the observations in that range. If one stem has more leaves than a neighboring stem, more observations fall in its range than in that neighboring range. This is why a group of longer rows can indicate a concentration in that part of the scale.

The conclusion has a useful limit: row lengths compare counts across the ranges defined by the stems, not the precise spacing of all observations. The key and the way stems are arranged determine the range represented by each row. In a plot with split stems, for instance, a row may represent only part of a ten-unit interval. Interpret the visible pattern at the scale the plot actually displays.

A Reliable Reading Process

Use this sequence when describing a stemplot. Each step links a claim to evidence in the plot.

Common Pitfalls and Why Descriptions Matter

A common mistake is to call a distribution right-skewed because its main cluster sits on the right side of the plot. That confuses the peak with the tail. The correction is to locate the main concentration first, then see which direction the thinner, more extended part of the data reaches. If the leaves trail toward higher values, describe a right tail, even if the main group is nearer the lower end.

Another mistake is to report a feature without evidence: “the data have a gap” or “there is an outlier.” Instead, identify the values around the gap or the observation that stands apart. Also avoid treating an empty stem as proof that every nearby value is absent; state the actual interval represented by the empty stem and the nearest recorded values.

These careful descriptions matter because a graph is evidence, not a substitute for words. Naming the variable, its units, the range of a feature, and the observations that support it makes the interpretation understandable to someone who has not seen the plot. It also prevents a visual impression from becoming a stronger claim than the data justify.