## Explaining Bimodal Histograms

written by: Michele McDonough โข edited by: Ronda Bowen โข updated: 1/2/2011

This guide takes a look at what a bimodal histogram is and gives examples, including an instance when a histogram might appear to be bimodal but really isn't.

• slide 1 of 5

### What is a Bimodal Histogram?

Basically, a bimodal histogram is just a histogram with two obvious relative modes, or data peaks. For example, take a look at the histogram shown to the right (you can click any image in this article for a larger view). This graph is showing the average number of customers that a particular restaurant has during each hour it is open.

Note that there are two relative peaks in the data โ one at 12:00 PM and one at 7:00 PM. When you think about it, this makes perfect sense, because noon corresponds to the average person's lunch time and 7:00 PM is a common dinner hour for many people. So, like many restaurants, the one depicted in this histogram can expect a lot more customers around noon and 7:00 PM than at other times of the day. This makes the data bimodal since there are two separate periods during the day that correspond to peak serving times.

Note: The graphs used as examples in this article were created in Excel. For more details on how to do this yourself, see Construct a Histogram in Microsoft Excel.

• slide 2 of 5

### Recognizing a Bimodal Histogram

Since bimodal histograms are really just histograms with two data peaks, they should be pretty easy to recognize, right? It is important to take a little care here and make sure that you understand the data before deciding if it is truly bimodal. For instance, look at the image of the histogram below.

• slide 3 of 5

This histogram depicts the average number of complaints received every two hours during a 24-hour period. At first glance, it appears to be bimodal with two relative peaks, one on the far left of the graph and one on the far right. But, is it really bimodal?

The problem with this depiction is that the data being represented covers a cyclical 24-hour period. We chose to represent the data in the histogram above by starting at midnight and continuing on to 10:00 PM. We could have just as easily chosen another hour during the day as the starting interval of the graph. For instance, the histogram below shows the exact same data, but this time the starting interval is noon and the final interval is 10:00 AM.

• slide 4 of 5

When viewing this histogram, the data looks quite different โ in fact, this second histogram almost seems to have a roughly normal distribution (or slightly skewed distribution) with a single peak at midnight (12:00 AM). In this case, the data in the original histogram really isn't bimodal. It only appeared that way because of how we chose to represent the information. This is a situation that can crop up frequently when dealing with cyclical data, so it's important not to make an interpretation of the data based on just a quick glance.

• slide 5 of 5