Percentiles and quartiles are core descriptive statistics that turn a long list of numbers into an understandable picture of position, spread, and comparison. In data analysis, descriptive statistics summarize what a dataset looks like without making claims about why patterns exist or what will happen next. They include measures of center such as the mean and median, measures of spread such as the range and standard deviation, and measures of relative standing such as percentiles, quartiles, and ranks. I use these measures constantly when reviewing dashboards, checking survey distributions, and validating whether a reported average hides important variation.
A percentile tells you the percentage of observations at or below a specific value. If a student scores in the 90th percentile on a test, that student performed as well as or better than 90 percent of test takers. Quartiles are a special set of percentiles that divide ordered data into four equal parts. The first quartile, or Q1, marks the 25th percentile. The second quartile, Q2, is the median, or 50th percentile. The third quartile, Q3, marks the 75th percentile. Together, these values help analysts understand where data clusters and where it stretches.
This matters because raw numbers rarely speak for themselves. A salary of $75,000 means one thing in a national income dataset and something very different inside a local nonprofit payroll. A hospital wait time of 40 minutes may seem acceptable until you learn it sits in the 95th percentile for delays. Percentiles and quartiles provide that context. They support fairer comparisons across groups, improve reporting, and make distributions easier to communicate to nontechnical audiences. As a hub within descriptive statistics, this guide explains not only how percentiles and quartiles work, but also how they connect to median, interquartile range, box plots, outliers, skewness, and data interpretation.
Descriptive statistics are often the first serious quality check in any analysis workflow. Before building forecasts or testing hypotheses, I sort values, inspect missing data, review summary measures, and compare percentile cutoffs. That step catches data entry errors, reveals whether a few extreme values dominate the average, and shows whether different segments behave differently. Understanding percentiles and quartiles is therefore not just an academic exercise. It is a practical skill for analysts, managers, researchers, students, and anyone who needs to explain what data is saying clearly and accurately.
What percentiles and quartiles measure
Percentiles and quartiles measure relative position within an ordered dataset. To calculate them, arrange values from smallest to largest and locate the points that split the distribution by percentage. Unlike the mean, which summarizes a typical value by averaging all observations, percentiles focus on rank. That makes them especially useful when data is skewed, bounded, or filled with outliers. House prices, emergency room wait times, web page load speeds, and annual bonuses all commonly show this pattern.
Suppose a retailer tracks order values for one thousand purchases. The median order might be $48, while the 90th percentile is $130. That tells you half of orders are $48 or less, but the top tenth are much larger. If Q1 is $26 and Q3 is $78, the middle 50 percent of orders fall between $26 and $78. In plain terms, quartiles show the central spread of the data, while higher percentiles spotlight the upper tail. Teams use these measures to set thresholds, define service levels, and benchmark customer segments.
Because percentiles are positional, they are less sensitive than the mean to a handful of extreme numbers. If one luxury purchase of $5,000 appears in the retailer example, the average order value rises noticeably, but the median and quartiles may barely move. That stability is why operational teams often report median delivery time, 95th percentile latency, or quartile-based salary bands instead of relying on averages alone.
How to calculate percentiles and quartiles correctly
The basic process is straightforward: sort the data, determine the rank position, and interpolate when needed. The complexity comes from the fact that software packages do not always use the same formula. Excel, Google Sheets, R, Python NumPy, and SQL implementations may apply slightly different percentile methods, especially for small samples. In production reporting, consistency matters more than choosing a supposedly perfect method. Document the rule you use and keep it stable across reports.
For quartiles in a small ordered dataset such as 4, 7, 9, 10, 12, 15, 18, 21, the median is the average of the fourth and fifth values, which is 11. The lower half is 4, 7, 9, 10, so Q1 is the average of 7 and 9, which is 8. The upper half is 12, 15, 18, 21, so Q3 is the average of 15 and 18, which is 16.5. The interquartile range, or IQR, equals Q3 minus Q1, so the IQR here is 8.5. That single value summarizes the spread of the middle half of the dataset.
In larger datasets, interpolation is common. If the 90th percentile rank falls between two ordered observations, many tools estimate a value between them rather than selecting one exact observation. This is not an error. It is a standard statistical convention that produces smoother cut points. The important practice is to know your tool’s method. In Excel, PERCENTILE.INC and PERCENTILE.EXC behave differently. In Python, pandas and NumPy now expose percentile method options explicitly. Analysts should check defaults before comparing outputs across platforms.
How percentiles fit within descriptive statistics
Descriptive statistics work best as a set, not as isolated numbers. Percentiles and quartiles answer where a value sits, but they become much more informative when read alongside the mean, median, mode, range, variance, standard deviation, and distribution shape. A common mistake is to report a median without any spread measure. Another is to report a standard deviation on highly skewed data without any percentile context. Good interpretation combines location and variability.
Consider employee commute times. If the mean is 42 minutes, the median is 31, Q1 is 18, and Q3 is 54, the story is clear: most employees commute far less than the mean suggests, and a long right tail is pulling the average upward. Add the 95th percentile at 96 minutes and leadership can see that a smaller group experiences severe travel burden. This kind of summary supports decisions about hybrid schedules, office relocation, or transit subsidies more effectively than one average alone.
Percentiles also connect directly to common visuals. A box plot displays Q1, the median, Q3, and potential outliers beyond whiskers often defined as 1.5 times the IQR. A cumulative distribution chart shows the share of observations below each value. Histograms reveal why percentile spacing changes across a distribution. When data is densely packed, many percentiles sit close together; when a tail stretches out, percentile gaps widen. These are essential concepts in descriptive statistics because they translate formula results into interpretable patterns.
When to use percentiles instead of averages
Use percentiles when the distribution is skewed, when outliers are operationally important, or when stakeholders care about thresholds rather than central tendency alone. In digital performance monitoring, median page load time may represent a typical user, but the 95th percentile captures frustrating experiences affecting customer retention. In healthcare operations, median wait time is useful, yet the 90th percentile often better reflects system strain because it highlights patients waiting unusually long. In compensation analysis, quartiles describe pay bands more fairly than averages when executive salaries distort the mean.
In my own reporting work, I default to percentile summaries whenever audiences need to compare groups with uneven distributions. Customer support resolution times, claim processing durations, and order fulfillment cycles are classic examples. Two teams can share the same average while delivering very different customer experiences. One may resolve nearly all cases consistently, while the other resolves many quickly but leaves a minority unresolved for weeks. Quartiles and upper percentiles expose that difference immediately.
| Scenario | Average alone misses | Useful percentile or quartile view |
|---|---|---|
| Website speed | Slowest users hidden by many fast visits | Median and 95th percentile load time |
| Income data | High earners pull mean upward | Median income and quartile bands |
| Hospital waits | Extremely delayed cases diluted | 50th, 75th, and 90th percentiles |
| Student scores | Average obscures rank position | Percentile rank by grade or school |
| Delivery times | Typical speed hides service failures | Median, Q3, and 95th percentile |
Averages still matter. The mean is valuable for budgeting, forecasting, and modeling because it uses every observation and supports many statistical methods. But if your goal is interpretation, fairness, or service-level monitoring, percentiles and quartiles often provide the more honest summary. The best practice is not choosing one measure forever. It is matching the measure to the question.
Interquartile range, outliers, and box plot interpretation
The interquartile range is one of the most practical spread measures in descriptive statistics. It equals Q3 minus Q1 and captures the middle 50 percent of observations. Because it ignores the most extreme values, it is robust in messy real-world data. Analysts use the IQR to compare consistency across groups and to flag outliers using Tukey’s rule. Values below Q1 minus 1.5 times the IQR or above Q3 plus 1.5 times the IQR are often labeled potential outliers.
That rule is a diagnostic, not a verdict. An outlier can be an error, a rare but valid event, or the very signal you care about. In fraud detection, outliers may identify suspicious claims. In manufacturing, they may indicate a failing machine. In executive pay data, they may be entirely real yet unsuitable for describing the typical employee. I never remove outliers automatically. I inspect source records, understand the data-generating process, and decide whether the analysis should describe the population as observed or a cleaner operational baseline.
Box plots help communicate this quickly. The box spans Q1 to Q3, the line inside marks the median, whiskers extend to non-outlier extremes by the chosen rule, and isolated points show outliers. When the median sits low in the box and the upper whisker is long, the distribution is right-skewed. When the box is narrow, the central data is tightly clustered. Managers grasp these signals faster from a box plot than from a paragraph of explanation, which is why quartiles are central to dashboard design and exploratory analysis.
Common mistakes and interpretation pitfalls
The most common mistake is confusing percentile with percent. A score of 80 percent on an exam is a proportion of questions answered correctly. A score in the 80th percentile means the student scored at or above 80 percent of test takers. Those are different ideas. Another mistake is assuming percentile differences are equal in raw units. The gap between the 50th and 60th percentiles may be tiny, while the gap between the 90th and 100th percentiles may be enormous in a skewed dataset.
Analysts also misread quartiles when sample sizes are small. With very few observations, quartile estimates can shift a lot with the addition of one value. That does not make them useless, but it does require caution. Small samples call for transparent reporting, sometimes with raw values shown alongside summaries. Another pitfall is comparing percentiles across populations with different definitions. A blood pressure percentile for children depends on age and sex standards; a percentile rank from one standardized test is not automatically comparable to another.
Finally, remember that descriptive statistics describe observed data; they do not establish causes. If one store’s 75th percentile sales are higher than another’s, the reason might be pricing, foot traffic, staffing, local demand, or data timing. Percentiles are excellent for surfacing patterns and directing questions. They are not a substitute for deeper analysis.
Tools, formulas, and practical applications
Most analysts calculate percentiles and quartiles in Excel, Google Sheets, SQL, Python, R, Tableau, or Power BI. In spreadsheets, functions such as QUARTILE.INC, QUARTILE.EXC, PERCENTILE.INC, and MEDIAN are widely used. In SQL, functions like PERCENTILE_CONT and NTILE support percentile and quartile grouping, though exact syntax varies by database. In Python, pandas describe(), quantile(), and plotting libraries make exploratory work efficient. In R, quantile() and boxplot() remain standard tools for statistical reporting.
Practical applications span nearly every industry. Schools use percentile ranks to compare student performance within large cohorts. Human resources teams build salary structures around quartiles so pay decisions align with market distributions. Operations teams monitor the 95th percentile of response times to set service-level objectives. Public health researchers examine income or exposure quartiles to study inequality. Finance teams review portfolio return percentiles to compare performance under different market conditions. In each case, the value lies in contextualizing an observation within the whole distribution.
As a descriptive statistics hub, this topic naturally links outward to deeper articles on median, mean versus median, range, variance, standard deviation, skewness, box plots, histograms, outlier detection, z-scores, and data visualization. Learning those topics alongside percentiles and quartiles gives you a complete foundation for interpreting distributions responsibly.
Percentiles and quartiles explained simply come down to this: they show where values fall and how data is spread, especially when averages alone are misleading. Q1, the median, and Q3 divide ordered data into meaningful sections, while percentiles extend that logic across the full distribution. Used together with measures like the mean, standard deviation, and interquartile range, they create a much more accurate descriptive summary than any single statistic can provide.
The main benefit is better interpretation. Whether you are reviewing salaries, test scores, wait times, customer behavior, or operational performance, percentiles reveal typical outcomes, upper-tail risk, and group differences clearly. They also improve communication. Nontechnical audiences understand statements like “90 percent of deliveries arrived within two days” faster than abstract formulas. That clarity helps teams make better decisions because the data story is grounded in relative position, not just one overall average.
If you are building your knowledge of descriptive statistics, start by calculating quartiles, the median, and the interquartile range on a real dataset you know well. Then compare those results with the mean and standard deviation. That simple exercise will show you how much distribution shape matters. From there, explore the related topics in this sub-pillar and use percentiles and quartiles as your foundation for stronger data analysis and interpretation.
Frequently Asked Questions
What are percentiles and quartiles, and how are they different?
Percentiles and quartiles are both measures of relative position, which means they tell you where a value falls within a dataset compared with all the other values. A percentile divides data into 100 equal parts. For example, if a test score is at the 85th percentile, that means the score is higher than or equal to about 85% of the scores in the dataset and lower than about 15%. Quartiles work the same way in principle, but they divide the data into four equal parts instead of 100. The first quartile, often written as Q1, marks the 25th percentile. The second quartile, Q2, is the 50th percentile and is also the median. The third quartile, Q3, marks the 75th percentile.
The key difference is level of detail. Percentiles give a very fine-grained view of position, while quartiles provide a broader summary of how the data is distributed. Quartiles are especially useful when you want a quick sense of the spread of the middle half of the data, while percentiles are helpful when you want to compare specific observations more precisely. Both are part of descriptive statistics because they summarize the structure of a dataset without explaining causes or predicting future outcomes.
How do you calculate quartiles and percentiles in a dataset?
The first step in calculating either quartiles or percentiles is to sort the data from smallest to largest. Once the values are in order, you identify the position associated with the percentile or quartile you want. For quartiles, Q2 is the median, which is the middle value if there is an odd number of observations or the average of the two middle values if there is an even number. Q1 is the median of the lower half of the data, and Q3 is the median of the upper half. Depending on the textbook, software package, or statistical convention, the exact method for finding Q1 and Q3 can vary slightly, especially in small datasets, but the overall goal is always to locate the 25th, 50th, and 75th percentiles.
Percentiles follow the same logic but with more possible cut points. A common approach is to compute a rank position using a formula such as P/100 multiplied by the number of observations, sometimes adjusted by adding or subtracting 1 depending on the method being used. If the resulting position falls between two data values, interpolation may be used to estimate the percentile. This is why different calculators or spreadsheet functions can sometimes return slightly different answers for the same percentile. In practice, what matters most is being consistent about the method you use and understanding that all accepted methods are trying to estimate the same general idea: the value below which a certain percentage of the data falls.
Why are quartiles and percentiles useful in descriptive statistics?
Quartiles and percentiles are useful because they turn raw data into something interpretable. A long list of numbers can be difficult to evaluate at a glance, but relative-position measures make it much easier to understand who or what is near the top, near the middle, or near the bottom of the distribution. They are especially valuable when the mean alone does not tell the full story. For example, two datasets can have the same average but very different spreads and patterns. Quartiles and percentiles reveal how values are distributed across the range, which helps you see whether the data is tightly clustered, widely spread, or skewed.
They are also highly practical for comparison. In education, percentiles help compare student performance across a large group. In healthcare, growth percentiles show how a child compares with peers of the same age. In business, quartiles can help segment customers, products, or regions into groups based on performance. The interquartile range, which is Q3 minus Q1, is one of the most important spread measures because it describes the range of the middle 50% of the data and is less affected by extreme values than the full range. That makes quartiles and percentiles especially powerful when datasets contain outliers or are not symmetrically distributed.
What does it mean when a value is in the 25th, 50th, or 90th percentile?
When a value is in the 25th percentile, it means that about 25% of the observations are at or below that value, and about 75% are above it. A value in the 50th percentile sits at the middle of the distribution and corresponds to the median, meaning half the observations are below it and half are above it. A value in the 90th percentile is relatively high, indicating that about 90% of observations are at or below that point and only about 10% are above it. These interpretations help place individual observations in context rather than looking at them in isolation.
It is important not to confuse percentile rank with percent correct or percentage of a maximum score. A student who scored in the 90th percentile on an exam did not necessarily get 90% of the questions right. Instead, the student performed better than roughly 90% of the reference group. This distinction matters because percentiles describe relative standing within a distribution, not absolute performance. That is why percentiles are so valuable in benchmarking situations where the goal is to compare one observation with others in the same population or sample.
How do quartiles relate to box plots, outliers, and data spread?
Quartiles are central to understanding box plots because a box plot is built directly from the median, Q1, and Q3. The box itself stretches from Q1 to Q3, representing the middle 50% of the data. A line inside the box marks the median, or Q2. This visual summary allows you to quickly assess central tendency, variability, and possible skewness. If the median is centered within the box and the whiskers are about equal in length, the data may be fairly symmetric. If the median is closer to one side or one whisker is much longer, the distribution may be skewed.
Quartiles also help identify outliers through the interquartile range, or IQR. The IQR is calculated as Q3 minus Q1. A common rule is that any value below Q1 minus 1.5 times the IQR or above Q3 plus 1.5 times the IQR may be considered an outlier. This approach is widely used because it is based on the middle portion of the data rather than the full range, making it more resistant to distortion from unusually high or low values. In short, quartiles do more than mark data positions. They provide a framework for understanding spread, spotting unusual observations, and creating meaningful visual summaries of a dataset.
