Can a range be negative?

For valid ordered real-valued observations, maximum minus minimum is nonnegative. A negative result suggests reversed subtraction or another error.

Does a wider range always mean a larger standard deviation?

No. The range uses only two extremes, while standard deviation depends on all observations and their distances from the mean.

Is Python's range the same statistic?

No. Python's range type represents an integer sequence specified by start, stop and step; it is not the maximum-minus-minimum summary of a dataset.

AI Glossary / Data science and analytics / Beginner

Range

Definition

In descriptive statistics, the range is the largest observed value minus the smallest observed value. It is a simple measure of spread that describes the distance between a dataset's two extremes, in the same units as the original measurements.

What it means

If recorded waiting times extend from 3 to 18 minutes, the observed interval is 3-18 minutes and the numerical range is 15 minutes. These are related descriptions, but the interval and its width are not the same thing.

The range is easy to calculate and can reveal an unexpectedly wide spread. However, it uses only the minimum and maximum. It does not show where the other observations lie, how common the extremes are or whether there are several distinct groups.

This entry focuses on the statistical meaning used in data summaries. Elsewhere, “range” can mean a set of possible outputs, a permitted interval or a programming construct. Those meanings should be distinguished from the observed maximum-minus-minimum statistic.

How it works

Identify the valid numerical observations for a clearly defined population, sample or reporting period. Check that values use compatible units and that special missing-value codes are not being treated as measurements.

Find the minimum and maximum, then subtract the minimum from the maximum. A nonempty dataset containing only one value, or several identical values, has range zero. An empty dataset has no observed minimum or maximum, so its range is not ordinarily defined without a special software convention.

The range is sensitive to extreme observations. Adding records to the same dataset cannot reduce its range, although a new reporting window or a different sample may have a smaller one. Larger samples often have more opportunity to include extreme values.

Pair the range with other summaries, such as the median, interquartile range or standard deviation, and inspect the distribution when the pattern of variation matters.

Three practical examples

A service manager checks the minimum and maximum completion times to identify the full observed span before investigating unusually long cases.
A data-quality analyst compares recorded values with plausible measurement limits, treating an unexpected range as a reason to inspect the source.
An AI evaluator reports the range of observed task scores alongside their distribution, avoiding the impression that two extremes describe typical performance.

Think of it this way

Think of measuring the distance between the first and last books on a shelf. That tells you the occupied span, but not how tightly the books are grouped between the endpoints.

The analogy has limits because a statistical range concerns numerical values rather than physical placement. Its useful point is that the two endpoints alone cannot reveal the pattern inside the span.

Worked example

Consider two invented five-value datasets. A is 0, 5, 5, 5, 10. B is 0, 0, 0, 10, 10. Both have minimum 0, maximum 10 and range 10, yet their internal distributions are quite different.

Now consider waiting times 4, 5, 6, 7 and 8 minutes. Their range is 8 − 4 = 4 minutes. Adding one genuine 40-minute case makes the range 40 − 4 = 36 minutes, even though the original five cases are unchanged.

The new range correctly records the wider observed span. It does not establish that most cases became slower or that the 40-minute value is erroneous. Those are separate questions.

If the same waiting times are converted to seconds, the original range becomes 240 seconds. The physical spread is unchanged; only the measurement unit and numerical value change.

Illustrative case study

Illustrative fictional case: Maple Operations compares the range of case duration for two teams. One team has ten completed cases in the period; the other has a thousand. A draft report calls the second team less consistent solely because its range is wider.

The analyst explains that sample size and case mix differ. The team examines the distributions, central summaries and long cases, and checks whether the measurement rules are comparable. It retains the ranges as descriptive facts without using them alone to rank performance.

The fictional lesson is that a correct statistic can support an unjustified comparison when context is omitted. The case does not claim that sample size explains every difference or that extreme cases should be ignored.

Common misconceptions and mistakes

Do not confuse the range with the maximum value, the number of observations or the interquartile range. Do not assume two datasets with the same range have similar distributions.

Avoid interpreting an observed range as a guaranteed boundary for future values. Check missing-value codes and unit errors before drawing conclusions. Do not delete a genuine extreme merely to reduce the statistic. A range of zero also does not prove perfect measurement: limited precision or repeated default values can hide variation.

Business use

Use the range as a quick description of observed extremes, especially when a dashboard also shows the endpoints and number of observations. It can help direct investigation but rarely answers a performance question on its own.

Keep the period, units, inclusion rules and sample size visible. For AI-generated summaries, distinguish “the observed values spanned” from “all future values will remain within.” When comparing processes, consider their distributions and case mix rather than using the range as an automatic quality score.

Check your understanding

A report says recorded values run from -3 to 7. What is their numerical range, and does it describe how values are distributed inside those limits?

Answer: The range is 7 − (-3) = 10. It gives the observed span but does not describe the frequency or arrangement of values within it.

Related terms

Standard deviation
Outlier
Descriptive (or summary) statistics
Minimum
maximum
spread
interquartile range
descriptive statistics
measurement unit

Learn next

Read interquartile range and standard deviation. Construct two small datasets with the same minimum and maximum but different internal patterns. Compare their summaries and explain which additional information is needed for a useful interpretation.

Standard deviation

Sources

Australian Bureau of Statistics, statistical range definition: https://www.abs.gov.au/statistics/understanding-statistics/statistical-terms-and-concepts/statistical-terms-and-concepts-glossary

NIST, measures of scale and sensitivity to extremes: https://itl.nist.gov/div898/handbook/eda/section3/eda356.htm

Python documentation, distinct programming meaning of range: https://docs.python.org/3/library/stdtypes.html#ranges

The datasets and operations case are original fictional teaching examples.

Previous: Randomness test Next: Range coding