The interquartile range (IQR) is not affected by outliers. Because it measures only the spread of the middle 50% of a dataset, extreme values at either end have no influence on the calculation. This is one of the main reasons statisticians use the IQR instead of the range or standard deviation when working with data that contains unusual values.
Why the IQR Ignores Extreme Values
The IQR is calculated by finding the difference between the 75th percentile (Q3) and the 25th percentile (Q1) of a dataset. Those two boundaries mark off the central half of your data, so anything happening in the top 25% or bottom 25% simply doesn’t enter the equation. If the highest value in your dataset doubles, triples, or jumps to a million, Q1 and Q3 stay exactly where they were.
Compare this to the range, which uses only the single highest and single lowest values. A dataset of coursework scores from 27 to 48 has a range of 21. But if one student scores zero, the range balloons to 48, even though the rest of the data hasn’t changed at all. The Australian Bureau of Statistics describes the IQR as “a better measure of spread than the range as it is not affected by outliers” for exactly this reason.
Standard deviation has a similar vulnerability. Because it squares every distance from the mean, one extreme value can pull both the mean and the standard deviation substantially. The IQR sidesteps this entirely by never referencing the mean and never looking beyond the 25th and 75th percentiles.
How the IQR Is Calculated
To find the IQR, you sort your data from smallest to largest, then identify two values. Q1 is the point where 25% of the data falls below, and Q3 is the point where 75% falls below. The IQR is simply Q3 minus Q1.
For a quick example: if Q1 is 8 and Q3 is 12, the IQR is 4. Now imagine you add an extreme value of 500 to the high end. Q1 and Q3 barely shift (if they shift at all, depending on dataset size), and the IQR stays close to 4. The range, meanwhile, would explode.
The IQR as an Outlier Detection Tool
The IQR doesn’t just resist outliers. It’s also the foundation of the most common method for identifying them. The technique, sometimes called Tukey’s fences or the 1.5×IQR rule, works by building a “fence” around the middle of the data. You 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 outside those fences is flagged as an outlier.
- Low outliers: values below Q1 − (1.5 × IQR)
- High outliers: values above Q3 + (1.5 × IQR)
This is the method behind the dots you see beyond the whiskers on a box plot. The box itself represents the IQR, the line inside it marks the median, and the whiskers extend to the most extreme non-outlier values. Points plotted individually beyond the whiskers are the outliers that fell outside the 1.5×IQR fence. It’s a clean visual system precisely because the IQR at its core is stable regardless of how extreme those outer dots are.
When the IQR Could Shift
There is one edge case worth understanding. A single outlier won’t budge the IQR, but if a very large proportion of your data changes, the percentiles themselves can shift. If you replaced 30% of your dataset with extreme values, Q1 or Q3 could move because the positions of the 25th and 75th percentiles within the sorted data have changed. This isn’t really an “outlier” scenario anymore, though. At that point, the data distribution itself has fundamentally changed, not just its tails.
In technical terms, the IQR has a high breakdown point: you’d need to corrupt a substantial fraction of the dataset before it responds. For practical purposes, with the kinds of outliers you encounter in real data (a few unusual observations in an otherwise consistent dataset), the IQR remains rock-solid.
When to Use the IQR Over Other Measures
The IQR is the better choice for describing spread whenever your data is skewed or contains extreme values. Income data is a classic example: a handful of very high earners can inflate the range and standard deviation, but the IQR captures what’s typical for the middle majority. Medical data, housing prices, and response time measurements all tend to have long tails that make the IQR more informative than alternatives.
If your data is roughly symmetrical with no extreme values, the standard deviation gives you more information because it uses every data point. But the moment you suspect outliers or skew, the IQR is the more reliable and honest summary of how spread out your data really is.