Histogram shapes / Bimodal
Bimodal Histogram (Two Peaks)
A bimodal histogram has two separate peaks, usually a sign that two different groups are mixed in one dataset. See an example and how to split the groups apart.
What a bimodal histogram looks like
A bimodal histogram has two distinct peaks with a dip between them. Instead of one hump, you see two, like a camel's back. It almost always means two different groups have been mixed into a single dataset.
The example above has one cluster low and another high, with a valley in the middle where few values land.
Two peaks usually means two groups
The most useful thing a bimodal shape tells you is to split the data. Common causes are two shifts, two machines, two age groups, before and after a change, or two species measured together. Each group has its own center, and the histogram is showing both at once.
When you see two peaks, ask what could divide the data in two, then chart each group on its own. Two clean single-peaked histograms are far easier to reason about than one bimodal blur.
Mean and median can mislead
For bimodal data the mean and median often fall in the dip between the peaks, a value that few or none of your actual data points sit near. That is why the average of a bimodal set can be misleading, and why the histogram matters more than the summary numbers.
Paste your data into the histogram maker. If two humps show up, you have found a hidden split worth investigating.
Frequently asked questions
- What does a bimodal histogram tell you?
- That two different groups are probably mixed together in one dataset. Each peak is the center of one group. The fix is usually to separate the groups and chart each one on its own.
- Is the mean useful for bimodal data?
- Often not. The mean of a bimodal set tends to land in the empty valley between the two peaks, a value few real data points are near, so it describes the data poorly. Split the groups instead.