A modified box plot is a version of the standard box plot that individually marks outliers instead of hiding them inside the whiskers. In a standard box plot, the whiskers stretch all the way to the minimum and maximum values in the dataset, so extreme data points blend in with everything else. A modified box plot caps the whiskers at a calculated boundary and plots any data points beyond that boundary as separate symbols, usually dots or asterisks.
How It Differs From a Standard Box Plot
The core distinction is simple. A standard box plot extends its whiskers to the absolute smallest and largest values in the dataset. You see the full range, but you can’t tell whether those extreme values are unusually far from the rest of the data. A modified box plot solves this by using a formula to decide where the whiskers stop. Any data points that fall outside those limits get their own individual markers, making them immediately visible as outliers.
Most modern statistical software, including R’s ggplot2 and Python’s matplotlib, generates modified box plots by default. So if you’ve seen a box plot with scattered dots above or below the whiskers, you were already looking at a modified version.
The Parts of a Modified Box Plot
A modified box plot has five main components plus the outlier markers:
- The box spans from the 25th percentile (Q1) to the 75th percentile (Q3). The height of the box represents the interquartile range, or IQR, which is the middle 50% of your data.
- The median line sits inside the box at the 50th percentile, dividing the data in half.
- The upper whisker extends from Q3 up to the largest data point that still falls within the calculated fence (more on that below).
- The lower whisker extends from Q1 down to the smallest data point that still falls within the fence.
- Outlier markers are individual points plotted beyond the whiskers, typically shown as dots, circles, or asterisks.
The whiskers do not necessarily reach a fixed percentile. They stop at the last actual data point before the fence, so their length depends on where your real values happen to fall.
How the Fences Are Calculated
The fence is the boundary that separates regular data points from outliers. The math is straightforward. First, calculate the IQR by subtracting Q1 from Q3. Then multiply that value by 1.5. The lower fence is Q1 minus 1.5 times the IQR, and the upper fence is Q3 plus 1.5 times the IQR.
Here’s a concrete example. Suppose Q1 is 80 and Q3 is 90. The IQR is 10, and 1.5 times that is 15. The lower fence lands at 65 (80 minus 15) and the upper fence at 105 (90 plus 15). Any data point below 65 or above 105 gets flagged as an outlier and plotted individually.
The 1.5 multiplier isn’t arbitrary. For data that follows a normal distribution, this rule captures roughly 99.3% of values inside the fences, meaning only genuinely unusual observations get flagged.
Mild vs. Extreme Outliers
Some modified box plots go a step further and distinguish between two levels of outliers. This uses a second set of fences, called outer fences, calculated with a multiplier of 3 instead of 1.5.
- Mild outliers fall between the inner fence (1.5 × IQR) and the outer fence (3 × IQR). These are unusual but not dramatically so.
- Extreme outliers fall beyond the outer fence. These are data points that are very far from the bulk of the distribution.
When this distinction is shown visually, mild outliers are often marked with open circles and extreme outliers with filled circles or asterisks. Not all software makes this distinction by default, but it’s worth knowing when you encounter it.
Reading Skewness From the Plot
Modified box plots give you a quick visual read on whether your data is symmetric or skewed. In a symmetric distribution, the median line sits roughly in the center of the box, and both whiskers are about the same length. The distance from the median to Q3 equals the distance from the median to Q1.
In a right-skewed distribution (common with income data, home prices, or reaction times), the upper half of the box is wider than the lower half, and the upper whisker stretches farther. You’ll often see more outliers above the upper whisker than below the lower one. Left-skewed data shows the opposite pattern: the lower whisker is longer and outliers cluster on the low end.
One important caveat: the standard 1.5 × IQR rule was designed for roughly symmetric data. When applied to heavily skewed distributions, it tends to flag too many points on the longer tail as outliers, even though those values may be perfectly normal for that type of data. If you’re working with highly skewed data and seeing a large number of flagged outliers on one side, that’s worth keeping in mind.
When Box Plots Work Best
Box plots need a minimum sample size of about 5 data points to be meaningful. Below that, it’s better to just plot the individual values. A Nature Methods recommendation suggests showing raw data points for any sample smaller than 5, since the quartile calculations become unreliable with so few observations.
Where box plots really shine is in comparing groups. Placing modified box plots side by side for three, five, or ten groups lets you quickly compare medians, spreads, and outlier patterns across categories. Histograms need at least 30 data points to look useful and take up far more space when you’re comparing multiple groups. Box plots give you a readable comparison with as few as 5 observations per group.
The outlier display in modified box plots is especially valuable in fields like environmental monitoring, clinical research, and quality control, where identifying unusual values matters as much as understanding the central tendency. A standard box plot would silently absorb those extreme readings into the whisker length, making them invisible. The modified version puts them front and center, so you can investigate whether they represent errors, rare events, or genuinely important signals in your data.