Introduction
The interquartile range (IQR) is a simple, robust measure of spread that describes how data are distributed around the middle of a dataset. Unlike the full range, which depends on the single smallest and largest values, the IQR focuses on the middle 50% of observations — giving a clearer picture of typical variability without being dominated by extreme values or outliers.
- What the interquartile range is — plain explanation
- Why the IQR resists influence from extremes
- How to compute quartiles and the IQR
- Interpretation: what the IQR tells you
- When the IQR is more appropriate than other measures
- Practical examples and comparisons
- Applications across fields
- Best practices when using the IQR
- Conclusion
What the interquartile range is — plain explanation
The interquartile range is the difference between the third quartile (Q3) and the first quartile (Q1) of a dataset. Quartiles divide ordered data into four equal parts: Q1 is the value below which 25% of the data fall, the median (Q2) lies at 50%, and Q3 sits at 75%.
So IQR = Q3 − Q1. It measures the spread of the central half of the data. Because it ignores the lowest 25% and highest 25%, the IQR is not strongly affected by unusually small or large observations.
Why the IQR resists influence from extremes
Measures of spread differ in how they respond to extreme values. The range equals maximum minus minimum, therefore a single outlier can make the range arbitrarily large. The sample standard deviation also gives greater weight to points far from the mean because deviations are squared before averaging.
The IQR, in contrast, is based only on order positions (percentiles) rather than distances from a central value. To change the IQR you must move or add observations that cross the 25th or 75th percentile lines. A single extreme value outside the central 50% does not move those percentiles, so it does not change the IQR.
Simple example to illustrate
Suppose you have test scores for 11 students sorted lowest to highest: 45, 48, 51, 55, 60, 65, 70, 75, 80, 88, 95.
• Median (Q2) is 65.
• Q1 is the median of the lower half (45, 48, 51, 55, 60) which is 51.
• Q3 is the median of the upper half (70, 75, 80, 88, 95) which is 80.
• IQR = Q3 − Q1 = 80 − 51 = 29.
Now replace the largest score 95 with an extreme value 200. The sorted list becomes 45, 48, 51, 55, 60, 65, 70, 75, 80, 88, 200. Q1, median and Q3 remain 51, 65 and 80 respectively, so IQR remains 29. The range rose from 50 (95−45) to 155 (200−45) and the standard deviation increased, but the IQR was unchanged.
How to compute quartiles and the IQR
There are a few legitimate methods for finding quartiles; they differ slightly when datasets have even numbers of observations or when using interpolation. Two common approaches are the “inclusive” (also called median-based) method and the “exclusive” method used by statistical software. Despite small differences, both yield the same idea: find values delimiting the middle 50%.
Steps using a straightforward approach:
1. Sort the data in ascending order.
2. Find Q2 (the median). For odd counts, it is the middle value; for even counts, it is the average of the two middle values.
3. Identify the lower half (values below the median) and upper half (values above the median). Some methods include the median in both halves when the count is odd; others exclude it. Consistency is key.
4. Find Q1 as the median of the lower half and Q3 as the median of the upper half.
5. Compute IQR = Q3 − Q1.
Notes on software differences
Different statistical packages use different definitions. For example, older versions of Excel, R’s default ‘quantile’ settings, and other tools may use slightly different interpolation rules. These differences rarely matter for large datasets and do not change the conceptual usefulness of the IQR; just be consistent and document the method if reproducibility is needed.
Interpretation: what the IQR tells you
The IQR summarizes the typical spread of the middle 50% of data. A smaller IQR means the central values are close together; a larger IQR means they are more spread out.
Because it focuses on the middle, the IQR is a good choice when you care about the “typical” dispersion and want to downweight rare extremes. For skewed distributions or datasets with outliers, the IQR often gives a more meaningful sense of typical variability than the standard deviation.
Relationship to box plots and outlier detection
Box plots visually show the median, Q1, Q3 and IQR. The box spans Q1 to Q3, the line inside marks the median, and “whiskers” extend to observations within a commonly used boundary. A standard rule of thumb sets whiskers to the most extreme observations within 1.5 × IQR of the quartiles; points beyond that are often plotted individually as potential outliers.
This rule is not a strict test for outliers but a convention useful for identifying values that are unusually distant from the central 50%.
When the IQR is more appropriate than other measures
Use the IQR when you need a robust measure of spread that is not overly sensitive to outliers or when the data are skewed. Typical situations include income distributions, response times, biological measurements with measurement errors, and any case where single extreme observations might misrepresent the variability of the majority of cases.
For symmetric, well-behaved data without outliers (for example, approximate normal distributions), the standard deviation and IQR both provide useful information, and their values can be related: for a normal distribution, the IQR is approximately 1.349 × the standard deviation. This relationship is useful when comparing results expressed in different measures, but it depends on the assumption of normality.
Limitations of the IQR
The IQR ignores 50% of the data, so it does not reflect what happens in the tails. If tail behavior matters — for example, in risk analysis where rare extreme events are crucial — the IQR should be complemented by other measures such as range, percentiles (90th, 95th), or measures specifically designed for tails.
Also, because it is based on ranks, the IQR does not account for how far points within the central 50% are from one another beyond their quartile boundaries. Two datasets can have identical quartiles yet differ in distribution inside those quartile ranges.
Practical examples and comparisons
Example 1 — skewed salaries: A company has many employees earning similar salaries but a few executives with very high pay. The full range and mean-based measures would be pulled up by the executives. The IQR shows how much variation most employees experience without being distorted by the executive salaries.
Example 2 — measurement error: Suppose repeated laboratory measurements include occasional instrument glitches producing wildly incorrect values. The IQR summarizes the spread of the reliable common measurements while ignoring sporadic errors that would otherwise inflate standard deviation.
Comparing IQR and standard deviation: If you have two samples with the same mean but one has a much larger standard deviation while their IQRs are similar, this suggests the larger spread arises from extreme values rather than broader dispersion among most observations.
Applications across fields
Statistics and data analysis: IQR is widely used to summarize distributions, construct box plots, and flag potential outliers in exploratory data analysis.
Quality control: IQR helps describe variability in repeated measurements when occasional faults produce extreme readings.
Health and social sciences: Researchers report medians and IQRs when data are skewed (for example, length of hospital stay or income) because medians and IQRs are robust to outliers.
Education and testing: When scores are skewed or contain a few anomalous results, reporting the median and IQR gives a clearer picture of typical performance than the mean and standard deviation.
Best practices when using the IQR
Report the IQR together with a measure of central tendency (usually the median) so readers understand both center and spread of the middle 50%.
When using IQR to identify outliers, state the rule you used (for example, the 1.5 × IQR convention) and remember this is a heuristic, not a definitive test of error or falsity.
Complement the IQR with tail-focused measures if extremes matter for your question. Percentiles (10th, 90th), trimmed means, or explicit modeling of tails provide additional insight when needed.
Be consistent in how you compute quartiles if you need to compare results across datasets or reproduce an analysis.
Conclusion
The interquartile range captures the spread of the central 50% of data, giving a robust measure of typical variability that is not driven by extreme values. It complements other descriptive statistics: use it with the median for skewed data, with box plots for visualization, and alongside tail measures when extremes matter. Because it depends on rank positions rather than numeric distances, the IQR resists distortion from outliers and offers a practical, easy-to-interpret measure of spread for many real-world datasets.
