Unimodal vs bimodal distribution
By Jude Wallis · Published
A unimodal distribution has one prominent peak. A bimodal distribution has two prominent peaks with a dip between them. One mean or median summarizes a unimodal distribution reasonably well. Two peaks often signal two groups mixed together, and a single center can then land where few values are.
AP Statistics: Unit 1 (topics 1.6 Descriptions for One Quantitative Variable Distributions, 1.5 Graphical Representations for One Quantitative Variable). Unimodal and bimodal are shape descriptions from topic 1.6, Descriptions for One Quantitative Variable Distributions, in Unit 1 of the Fall 2026 AP Statistics course. The histograms, dotplots, and stemplots used to judge the number of peaks are topic 1.5, Graphical Representations for One Quantitative Variable.
Unimodal vs bimodal: the short answer
Both words describe the shape of a distribution by counting its prominent peaks. A unimodal distribution rises to one peak and falls away on both sides. A bimodal distribution has two peaks separated by a dip, so the values concentrate in two different regions.
The count matters because it tells you whether one measure of center can represent the data. With one peak, the mean or the median sits in or near the region where values are most common. With two peaks, the mean and the median often fall in the dip between them, where values are scarce. The first worked example has a mean commute time of about 20.9 minutes in a data set where no commute took between 14 and 27 minutes.
What a unimodal distribution looks like
A unimodal distribution has a single region where the values pile up. Its histogram climbs to one high area and then descends. The peak can be in the middle, giving a symmetric shape such as a bell, or off to one side, giving a skewed shape. Unimodal and symmetric are separate descriptions, and a full description of shape needs both a peak count and a statement about symmetry or skew.
Most of the distributions summarized with a single center and spread in this course are unimodal. The normal distribution is unimodal and symmetric. Distributions of incomes or waiting times are typically unimodal and skewed right.
Small bumps do not count as extra peaks. Real data are uneven, and a histogram bar that is one or two counts taller than its neighbor is ordinary sampling noise. A peak should be prominent: clearly higher than what surrounds it.
What a bimodal distribution looks like
A bimodal distribution has two prominent peaks with a clear dip between them. The peaks do not need to be the same height. A tall peak and a shorter, clearly separate one still make a bimodal shape.
Two peaks often mean the data combine two groups that differ on the variable. Heights of adults with men and women pooled together, or commute times for students who walk and students who ride a bus, tend to produce one peak per group. The useful response is to look for a variable that separates the groups and then describe each group on its own. This idea is developed in what is a bimodal distribution.
When the two peaks are well separated, the dip can become a gap, an interval with no values at all. A description of such data should mention the gap and the two clusters as well as the peak count.
The differences side by side
| Feature | Unimodal | Bimodal |
|---|---|---|
| Prominent peaks | One | Two, separated by a dip |
| Typical cause | One group measured on one variable | Often two groups mixed together |
| Single center (mean or median) | Usually a fair summary | Can fall between the peaks, where values are scarce |
| Symmetry | Can be symmetric or skewed | Can be symmetric or not |
| What to report | Shape, center, spread, unusual features | The two peaks or clusters and where each is located |
| Visible in a histogram or dotplot | Yes | Yes |
| Visible in a boxplot | No | No |
The last two rows matter in practice: a boxplot shows neither description, so the peak count has to come from a histogram, dotplot, or stemplot.
How to tell which one you have
Draw a histogram or dotplot and count the prominent peaks. Then test whether the count holds up. For a histogram, change the bin width and redraw. Very narrow bins break one smooth pile into several spikes, and very wide bins can merge two real clusters into a single bar. A second peak that appears at only one bin width is weak evidence. A second peak that persists across several reasonable bin widths is worth reporting.
Sample size matters as well. With a dozen observations, two small clumps can appear by chance. With several hundred, a clear dip between two peaks is much harder to produce by chance.
If you conclude the distribution is bimodal, say so, give the approximate location of each peak, and avoid presenting one mean and one standard deviation as if they described a typical value. If a variable that might explain the two groups is available, compare the groups separately, as in how to compare two distributions.
The classic mix-up and how to avoid it
The usual error comes from the word mode doing two jobs. For a list of data, a mode is a most frequent value, and a list can have two modes when two values tie. For a distribution's shape, a mode is a peak. These are different, and two tied values do not make a bimodal shape. The second worked example has two modes, 3 and 4, sitting next to each other at the top of a single peak. That distribution is unimodal.
The opposite error is requiring equal peaks. A distribution whose two peaks have different heights may have only one most frequent value, and it is still bimodal if the lower peak is prominent and separated from the higher one by a real dip.
A third error is describing peaks from a boxplot. A boxplot is drawn from five numbers and carries no information about the number of peaks, so a question about the number of peaks cannot be answered from a boxplot alone.
Two groups, two peaks, and a mean in the gap
Sixteen students report commute times in minutes. Eight walk: 8, 9, 10, 10, 11, 12, 12, 13. Eight ride a bus: 28, 29, 30, 31, 31, 32, 33, 35. Build a histogram of all 16 values with bins of width 5 starting at 5, find the mean and median of the combined data, and find each group's mean.
Count the bins, each including its left end. From 5 to 10: 8, 9, so 2. From 10 to 15: 10, 10, 11, 12, 12, 13, so 6. From 15 to 20: 0. From 20 to 25: 0. From 25 to 30: 28, 29, so 2. From 30 to 35: 30, 31, 31, 32, 33, so 5. From 35 to 40: 35, so 1.
Check the counts: .
Read the shape: one peak in the bin from 10 to 15, a second peak in the bin from 30 to 35, and two empty bins between them. The distribution is bimodal with a gap.
Combined mean: the walkers sum to and the bus riders sum to , so minutes.
Combined median: the 8th and 9th sorted values are 13 and 28, so the median is minutes.
Group means: walkers minutes, bus riders minutes.
The histogram counts are 2, 6, 0, 0, 2, 5, 1, a bimodal shape with a gap from 15 to 25. The combined mean is 20.875 minutes and the combined median is 20.5 minutes, and both fall in the gap: no student had a commute between 14 and 27 minutes. The group means, about 10.6 minutes for walkers and 31.1 minutes for bus riders, describe the data far better.
Two modes that make one peak
Twelve students report how many books they read over the summer: 1, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 6. Find the mode or modes of the list, then decide whether the distribution is unimodal or bimodal.
Tally the values: 1 occurs once, 2 twice, 3 three times, 4 three times, 5 twice, and 6 once.
Check the tally: .
The largest count is 3, and both 3 and 4 have it. The list has two modes, 3 and 4.
Now look at the shape. In a dotplot the column heights from 1 to 6 are 1, 2, 3, 3, 2, 1. The heights rise steadily to the middle and fall steadily after it.
The two tallest columns are adjacent, with no dip between them. They form the flat top of one peak.
The list has two modes, 3 and 4, and the distribution is unimodal (and symmetric). Bimodal refers to two separated peaks in the graph. Two neighboring values that tie for most frequent are a single peak.
Frequently asked questions
Do the two peaks of a bimodal distribution have to be the same height?
No. A bimodal distribution needs two prominent peaks with a clear dip between them. One peak can be taller than the other. Requiring equal heights would rule out almost all real data, where two peaks are rarely exactly level.
Can a distribution be both bimodal and symmetric?
Yes. If the two peaks are about the same size and the same distance from the center, the left and right halves mirror each other. The values 1, 1, 2, 9, 10, 10 are an example: symmetric about 5.5, with a cluster at each end.
What if a distribution has no clear peak?
If the frequencies are about the same across the whole range, the distribution is described as approximately uniform. If there are three or more prominent peaks, it is called multimodal. Both descriptions should be checked at more than one bin width.