The box plot calculator above turns a pasted list of numbers into an actual drawn box-and-whisker diagram: a box spanning the first to third quartile, a line at the median, whiskers reaching the furthest values still inside the fences, and individual dots for every point beyond them. It computes the five-number summary at the same time, and reports it under all three of the quartile conventions that statistical software uses, because those conventions genuinely disagree.
Arb Digital publishes this alongside a set of statistics tools that each return a single number. This one exists because the numbers alone hide shape. A dataset can have a perfectly ordinary mean and standard deviation while being visibly lopsided, and a box plot exposes that in a second. Our five-number summary calculator gives you the same five values as text; this page is the one that draws them, adds fences, and marks outliers, which is the part a summary table cannot do.
What This Box Plot Calculator Does
It sorts your data, finds the minimum, first quartile, median, third quartile, and maximum, and then derives the two fences that define an outlier. The plot is drawn as a scaled diagram with a horizontal axis, so the width of the box and the length of each whisker are proportional to the real values rather than schematic. Points outside the fences are drawn separately and listed by value, and the whiskers are shortened to the last point still inside the fence, which is the correct Tukey construction and the part most hand-drawn box plots get wrong.
Alongside the plot you get the interquartile range, both fence positions, and a count of flagged outliers. Switch the quartile method and the whole diagram redraws, which makes it easy to see whether a borderline outlier is a real feature of the data or an artefact of the convention your software happens to use.
How to Use It
- Paste your data. Commas, spaces, tabs, or one value per line all work, so a column copied straight out of a spreadsheet needs no cleaning.
- Pick a quartile method. If you are checking against a textbook, exclusive is the usual answer. If you are checking against a spreadsheet, choose linear interpolation.
- Set the fence multiplier. Leave it at 1.5 unless you are specifically hunting extreme outliers, where 3 is the standard alternative.
- Read the plot. The box holds the middle half of the data; the median line inside it shows where the centre sits within that half.
- Check the outlier list. Every flagged value is printed with the plot, so you can go back to the source and decide whether it is an error or a genuine extreme.
The Formula and How It's Calculated
The median is the middle value of the sorted data, or the average of the two middle values when the count is even. The quartiles are the medians of the lower and upper halves. Under the exclusive method, when the count is odd the central value is left out of both halves; under the inclusive method it is placed in both. The interquartile range is IQR = Q3 − Q1, and the fences are Q1 − 1.5 × IQR and Q3 + 1.5 × IQR.
Work the default dataset by hand: 2, 4, 5, 7, 8, 9, 12, 15, 20, 45. Ten values, so the median is the average of the fifth and sixth, (8 + 9) ÷ 2 = 8.5. The lower half is 2, 4, 5, 7, 8, whose median is 5, so Q1 = 5. The upper half is 9, 12, 15, 20, 45, whose median is 15, so Q3 = 15. IQR = 15 − 5 = 10. The upper fence sits at 15 + 1.5 × 10 = 30, and the lower fence at 5 − 15 = −10. Only 45 lies outside, so it is drawn as a dot and the upper whisker stops at 20, the largest value still inside the fence. The NIST/SEMATECH e-Handbook section on box plots describes exactly this construction.
Why Three Quartile Methods Give Three Answers
Run the same ten numbers through the linear interpolation method and Q1 becomes 5.5 and Q3 becomes 14.25, against 5 and 15 from the exclusive method. The IQR drops from 10 to 8.75 and the upper fence moves from 30 to 27.375. On this dataset the conclusion does not change, because 45 is far outside either fence — but on data where a point sits near the boundary, the choice of method decides whether it is labelled an outlier at all.
This is not a bug in anyone's software. There is no single agreed definition of a sample quartile, and at least nine defensible definitions are in circulation. Spreadsheets default to linear interpolation. Many introductory textbooks teach the exclusive method. Some statistical packages default to a third variant. Penn State's STAT 200 course notes teach the split-at-the-median approach, which is why student answers so often differ from spreadsheet output. The practical rule is to state which method you used whenever a quartile is reported, and never to compare a quartile from one tool against a quartile from another without checking. Our quartile calculator and interquartile range calculator cover the same conventions in more depth if you need to reconcile two conflicting outputs.
Reading Skew Off the Plot
A box plot tells you about shape without any distributional assumption. If the median line sits closer to the bottom of the box and the upper whisker is longer, the data is right-skewed: most values cluster low with a tail stretching high. Incomes, page-load times, session durations, and order values almost always look like this. A left-skewed plot, with the median high in the box and a long lower whisker, is much rarer in practice and usually indicates a ceiling effect — a score capped at 100, or a metric that cannot exceed some hard limit.
A symmetric plot has the median centred and whiskers of similar length. That is worth checking before you apply any method that assumes normality, because a badly skewed dataset will produce a mean that sits well away from the typical value. Confirm what the picture suggests numerically with the descriptive statistics calculator, which reports a skewness coefficient alongside the rest of the summary.
What an Outlier Flag Actually Means
The 1.5 × IQR rule is a screening device, not a verdict. It was chosen because for roughly normal data it flags only about 0.7% of observations, which is a useful rate for drawing attention to unusual points without burying you in them. It carries no claim that a flagged point is wrong. In a right-skewed dataset the rule will flag genuine, correct, expected high values simply because the distribution has a long tail — a customer who spends ten times the median is not a data error.
So treat a flag as a prompt to look at the record. Sometimes you find a decimal in the wrong place, a duplicate submission, or a test transaction, and removing it is right. Sometimes you find a real observation, in which case deleting it to make a chart tidier is data manipulation. A defensible middle path is to report the analysis both with and without the point and say what changed. The outlier calculator applies the same fence test along with the z-score approach if you want a second opinion on a borderline value.
Comparing Groups Side by Side
The reason box plots survive in professional reporting is comparison. Five box plots stacked against a shared axis let a reader compare five groups' centres, spreads, and skews in one glance, which no table of means achieves. If you are comparing conversion rates by traffic source, response times by server, or test scores by class, the box plot is usually the right first chart — before any significance test, because it shows immediately whether the groups even overlap.
Two cautions when comparing. First, box plots hide sample size completely: a box built from eight observations looks exactly as authoritative as one built from eight thousand, so always print the count next to each. Second, they hide multimodality. A dataset with two distinct clusters can produce a perfectly ordinary-looking box, because quartiles say nothing about what happens between them. When you suspect two populations are mixed together, plot a histogram as well — our frequency distribution calculator bins the same data and shows the shape a box plot smooths away.
Small Samples and Ties
Below about eight observations a box plot is close to meaningless, because each quartile is being estimated from two or three points and moves dramatically if any one of them changes. With five values the box is essentially three data points wide. It is not wrong to draw one, but do not read a shape into it, and never flag outliers from a tiny sample using the IQR rule — the IQR itself is too unstable to define a fence.
Heavy ties cause a different problem. If half your data is the same value, the box can collapse to zero width, with Q1, the median, and Q3 all equal. The plot degenerates to a line with whiskers. That is a correct rendering of the data and it is telling you something real: the variable is nearly constant with a few departures. Discrete rating scales with only five possible values produce this constantly, and for that kind of variable a frequency table is a better tool than a box plot.
Arb Digital builds analytics reporting that shows distribution and variance, not just an average with no context behind it.
Browse All Free Tools Talk To Our TeamCommon Mistakes to Avoid
- Drawing whiskers to the minimum and maximum — in a Tukey box plot the whiskers stop at the last value inside the fence, and anything further out is drawn as a separate point.
- Comparing quartiles from two different tools without checking that both used the same quartile definition.
- Deleting flagged outliers automatically — the fence rule identifies unusual values, it does not identify wrong ones.
- Reading a box plot from a handful of observations as if the shape were reliable.
- Assuming a symmetric box means normal data — a box plot cannot detect two clusters hiding inside the middle half.
Related Free Tools From Arb Digital
Get the same five values as plain text from the five-number summary calculator, isolate a single cut point with the percentile calculator, measure spread with the standard deviation calculator, or bin the data for a histogram view with the frequency distribution calculator. Everything else is on the free online tools hub.
Frequently Asked Questions
The middle half of the data as a box from the first to the third quartile, the median as a line inside it, whiskers to the most extreme values still inside the fences, and any points beyond the fences drawn individually as outliers.
Subtract 1.5 times the interquartile range from the first quartile for the lower fence, and add 1.5 times the interquartile range to the third quartile for the upper fence. Values outside that range are flagged.
There is no single agreed definition of a sample quartile. Spreadsheets typically interpolate between values, while many textbooks split the sorted data at the median and take the median of each half, and the two approaches disagree on the same numbers.
Only when there are no outliers. If any value falls outside a fence, the whisker stops at the last value inside the fence and the extreme values are drawn as separate points.
Yes. A median sitting low in the box with a long upper whisker indicates right skew, and the reverse pattern indicates left skew. A centred median with even whiskers suggests rough symmetry.
Roughly eight observations at a minimum, and more is much better. Below that, each quartile rests on so few points that the shape of the box is not stable enough to interpret.
Not automatically. Investigate each one first. Data entry errors and test records should go, but genuine extreme observations are part of the distribution and removing them changes the result you are reporting.
This tool is provided for educational and analytical use. Outlier flags are a statistical screening rule, not a judgement about whether any particular observation is correct.
