A quantile is a value that divides a dataset into equal-sized groups. If you’ve ever heard that a child is “in the 90th percentile for height,” you’ve already used a quantile: it means 90% of children the same age are shorter. Percentiles, quartiles, and deciles are all specific types of quantiles, each splitting data into a different number of groups.
How Quantiles Work
The core idea is simple. Line up all your data points from smallest to largest, then pick the values that cut that line into equally sized portions. Those cut points are the quantiles. A quantile always answers the question: “What value sits at this particular position in the data?”
More formally, a quantile is the value below which a given fraction of the data falls. The 25th percentile, for instance, is the value where 25% of observations are smaller and 75% are larger. This makes quantiles a way of describing position within a distribution rather than summarizing the whole thing with a single average.
Common Types of Quantiles
Different fields use different levels of granularity, but they all follow the same logic:
- Quartiles divide data into 4 equal groups. There are three cut points: Q1 (25th percentile), Q2 (the median, 50th percentile), and Q3 (75th percentile). Each group contains roughly 25% of the values.
- Quintiles divide data into 5 equal groups. You’ll see these often in economics when researchers split households into income fifths.
- Deciles divide data into 10 equal groups. The 6th decile, for example, is the same as the 60th percentile.
- Percentiles divide data into 100 groups, each containing about 1% of the values. The p-th percentile separates the bottom p% from the upper (100 − p)% of ordered data.
So percentiles are just very fine-grained quantiles, and quartiles are coarser ones. They’re all the same concept at different resolutions.
Why Quantiles Matter More Than Averages Sometimes
An average (mean) tells you where the center of your data is, but it can be pulled dramatically by extreme values. If nine people in a room earn $50,000 and one earns $10 million, the average income is over $1 million, which describes nobody in the room accurately. The median (the 50th percentile) stays at $50,000, which is far more representative.
This is the key advantage of quantiles: they’re resistant to outliers. They describe specific positions in the data rather than blending everything together. When data is heavily skewed, as income, home prices, and medical costs tend to be, quantiles give you a much clearer picture of what’s typical and what’s extreme.
Spotting Outliers With the IQR
One of the most common practical uses of quantiles is detecting outliers through the interquartile range (IQR). The IQR is simply the distance between Q1 and Q3, capturing the middle 50% of your data. The standard rule: any data point more than 1.5 times the IQR below Q1 or above Q3 counts as an outlier.
For example, if Q1 is 80 and Q3 is 90, the IQR is 10. Multiply by 1.5 to get 15. Anything below 65 (80 minus 15) or above 105 (90 plus 15) would be flagged as unusually far from the pack. This method works well precisely because the quartiles themselves aren’t distorted by the very outliers you’re trying to find.
Box Plots: Quantiles You Can See
If you’ve ever seen a box-and-whisker plot, you’ve seen quantiles visualized. A box plot is built from five values: the minimum, Q1, the median, Q3, and the maximum. The box spans from Q1 to Q3, so it contains roughly the middle 50% of the data. A line inside the box marks the median. The “whiskers” extend from each end of the box out to the smallest and largest data values (or to the outlier boundary, depending on the style).
This makes box plots an efficient way to compare distributions at a glance. A box that sits high, a long upper whisker, or a median pushed to one side all tell you something about the shape of the data without needing to look at every individual point.
Quantiles in Medicine
Pediatric growth charts are one of the most familiar real-world applications of quantiles. When a doctor plots a child’s weight on the 90th percentile for age, it means only 10 out of 100 children of the same age and sex in the reference population weigh more. Clinicians use specific percentile thresholds to screen for potential health concerns: a BMI-for-age below the 5th percentile may indicate underweight, between the 85th and 95th percentile flags overweight, and at or above the 95th percentile suggests obesity. Height-for-age below the 3rd percentile can signal stunting or a growth disorder.
Growth charts also track patterns over time. A child who unexpectedly crosses downward across two or more percentile lines may be experiencing growth failure, even if their current percentile isn’t alarming on its own. The trajectory matters as much as any single reading.
Quantile Regression in Research
Standard regression (the kind you learn in an introductory statistics course) models how explanatory variables affect the average outcome. But sometimes the average isn’t the interesting part. Maybe you want to know whether a drug works differently for patients with the most severe symptoms versus the mildest, or whether education affects income differently at the bottom versus the top of the wage distribution.
Quantile regression lets researchers model any percentile of the outcome, not just the mean. Instead of asking “how does X change the average Y?” it asks “how does X change the 10th percentile of Y, or the 90th?” This is especially useful with skewed data or when the relationship between variables isn’t uniform across the range. A factor might have a strong effect on low values and almost none on high values, and standard regression would completely miss that pattern.
Quantiles in Everyday Contexts
Beyond medicine and formal research, quantiles show up in places you might not expect. Standardized test scores are typically reported as percentiles. Income data is routinely broken into quintiles or deciles for policy analysis. Credit scoring models use percentile ranks. Air quality indices, student performance benchmarks, and even internet speed comparisons all rely on quantile-based thinking to communicate where a specific value falls relative to a larger population.
The underlying question is always the same: “Where does this value sit compared to everything else?” That’s what makes quantiles so broadly useful. They turn raw numbers into context.