Yes, the interquartile range (IQR) is a resistant statistic. It resists the influence of outliers because it only measures the spread of the middle 50% of your data, ignoring extreme values at both ends. This makes it a more reliable measure of spread than the range or standard deviation when your dataset contains unusually high or low values.
What “Resistant” Means in Statistics
A statistic is called resistant when a few extreme values in the dataset don’t dramatically change its result. The median is the classic example: if you have five test scores of 70, 80, 85, 90, and 95, the median is 85. Change that last score to 950, and the median stays at 85. The mean, by contrast, jumps from 84 to 255.
The IQR works the same way. It depends on two values, Q1 and Q3, that sit in the interior of the dataset. Q1 is the point where 25% of values fall below it, and Q3 is the point where 75% fall below it. The IQR is simply Q3 minus Q1. Because both of these markers are anchored to positions in the middle of the data rather than at the extremes, adding or changing an outlier typically has no effect on them.
Why IQR Resists Outliers
The IQR is resistant for the same reason the median is: it relies on the rank (position) of data points, not their actual magnitude. If you have 100 data points sorted in order, Q1 is determined by the values around position 25 and Q3 by the values around position 75. Making the largest value ten times bigger doesn’t shift those positions at all.
Compare this to the standard deviation, which calculates the average distance of every data point from the mean. A single extreme value inflates the mean and then contributes a large squared distance on top of that, pushing the standard deviation up significantly. The range is even more vulnerable since it’s defined entirely by the minimum and maximum values.
How to Calculate the IQR
The process is straightforward once you sort your data from smallest to largest:
- Find the median to split the dataset into a lower half and an upper half.
- Find Q1 by taking the median of the lower half.
- Find Q3 by taking the median of the upper half.
- Subtract: IQR = Q3 minus Q1.
For an odd number of data points, the overall median is excluded from both halves before finding Q1 and Q3. For an even number, you split the dataset evenly and may need to average two adjacent values when a half has an even count. As a quick example: given 11 sorted values where the 3rd value is 15 and the 9th value is 43, Q1 is 15, Q3 is 43, and the IQR is 28.
How the IQR Detects Outliers
The IQR’s resistance to outliers actually makes it useful for identifying them. The most common method, often called Tukey’s fences, works like this: multiply the IQR by 1.5, then subtract that number from Q1 to get the lower fence and add it to Q3 to get the upper fence. Any data point that falls outside those fences is flagged as an outlier.
This approach is what generates the “whiskers” in a box plot, and it’s the default outlier detection method in many statistical software packages. Because the IQR itself isn’t distorted by the outliers it’s trying to find, the fences stay in a reasonable position. If you used the standard deviation instead, the extreme values would inflate the measure and potentially mask their own existence as outliers.
IQR vs. Standard Deviation
The tradeoff for resistance is efficiency. When your data follows a normal (bell-shaped) distribution with no outliers, the standard deviation uses information from every data point and gives you a more precise estimate of spread. In a perfectly normal distribution, the IQR is approximately 1.349 times the standard deviation, but it captures less information because it discards the outer 50% of the data entirely.
This is why the IQR and standard deviation serve different roles. For clean, symmetric data, the standard deviation is the better summary. For skewed distributions, datasets with outliers, or situations where you simply don’t know whether extreme values are trustworthy, the IQR gives you a more honest picture of how spread out the bulk of your data really is. In research papers, you’ll often see skewed variables reported as “median (IQR)” rather than “mean (standard deviation)” for exactly this reason.
When Resistance Matters Most
Resistance becomes important whenever your data is likely to contain values that don’t represent the typical pattern. Income data is the classic case: a few extremely high earners can inflate the range and standard deviation, making it look like incomes are more spread out than most people experience. The IQR cuts through that distortion by focusing on the middle half of earners.
The same logic applies to medical data (a few extreme lab results from measurement errors), response time studies (a few participants who got distracted), or any dataset where you expect a handful of values to be far from the rest. In all of these cases, the IQR gives you a measure of spread that reflects the core of the data rather than its most unusual members.