Continuous numerical data is the best type to display in a histogram. This means data you measure rather than count: things like height, weight, temperature, income, time, and test scores. Histograms work by grouping these measured values into ranges (called bins) and showing how many observations fall into each range, which reveals the overall shape and spread of your data.
Why Continuous Data Fits Histograms
Continuous data can take on any value within a range, including fractions and decimals. That creates a nearly infinite number of possible values, which makes it impractical to plot each one individually. A histogram solves this by lumping nearby values into bins. For example, if you’re looking at the heights of 500 people, you wouldn’t plot a separate bar for 5’7.23″ and another for 5’7.24″. Instead, a histogram groups everyone between 5’7″ and 5’8″ into one bin and shows you how many people fall there.
This is fundamentally different from a bar chart, which displays categorical data like favorite colors, countries, or product types. Bar charts have a finite set of distinct categories with gaps between the bars. Histogram bars touch each other because the bins represent a continuous number line with no gaps between ranges.
Common Examples of Histogram Data
Almost any measured quantity with enough variation makes a good histogram candidate. Some of the most common examples include:
- Exam scores across a class of students
- Income distribution across a population
- Ages of participants in a study
- Reaction times in a psychology experiment
- Daily temperatures recorded over a year
- Wait times at a hospital or call center
In each case, you’re dealing with measured values that span a range, and the interesting question is how those values are distributed. Are most people clustered in the middle? Is there a long tail of extreme values? A histogram answers these questions visually in a way that a simple average never could.
What Histograms Reveal About Your Data
The real power of a histogram is showing you the shape of your data’s distribution. That shape tells you important things about what’s typical and what’s unusual.
A symmetrical, bell-shaped histogram indicates a normal distribution. In this pattern, the mean, median, and mode are all the same value, sitting right at the center. About 68% of values fall within one standard deviation of the center, 95% within two, and 99.7% within three. If you’ve ever seen grades curved on a bell curve, this is what that looks like.
A skewed histogram has a longer tail on one side. Income data is a classic example of a right-skewed (positively skewed) distribution: most people cluster at lower to middle incomes, but a smaller number of very high earners stretch the tail far to the right. When data is skewed like this, the median is a better measure of what’s “typical” than the mean, because the mean gets pulled toward the extreme values.
You might also see a bimodal histogram with two distinct peaks, which often signals that your data contains two separate groups. Test scores might look bimodal if half the class understood the material well and half didn’t, creating clusters at both ends of the scale.
How Bins Change What You See
The width and number of bins you choose can dramatically change how a histogram looks. Too few bins and you lose detail, collapsing meaningful variation into overly broad groups. Too many bins and the histogram becomes noisy, with spiky bars that obscure the underlying pattern.
A common rule of thumb is that somewhere between 5 and 20 bins works for most datasets. Many software tools default to 10. For more precise control, several mathematical formulas exist. Sturges’ rule calculates bin count based on sample size and works well for roughly normal data. The Freedman-Diaconis rule uses the spread of your data (specifically the interquartile range) and tends to perform better when data is skewed or has outliers. Scott’s rule optimizes for normally distributed data by factoring in the standard deviation. You don’t need to memorize these formulas, but knowing they exist helps if you’re using software that lets you choose a binning method.
Frequency, Percentage, and Density Histograms
The y-axis of a histogram can show different things depending on your goal. A frequency histogram is the most straightforward: the height of each bar represents the raw count of observations in that bin. If 47 students scored between 70 and 80 on an exam, that bar reaches 47.
A percentage histogram replaces raw counts with the proportion of total observations. This is useful when you want to compare datasets of different sizes, since the percentages are standardized. Instead of knowing that 47 students scored in that range, you’d see that 23.5% of the class did.
A density histogram scales the y-axis so the total area of all bars equals one. This version is less intuitive for general audiences but is standard in statistics and probability. It’s especially useful when you want to overlay a smooth density curve on top of the histogram to approximate the underlying distribution. Density plots also have a practical advantage when you need to compare multiple distributions on the same chart, where overlapping histograms would become difficult to read.
When a Histogram Isn’t the Right Choice
Histograms are not a good fit for categorical data (like survey responses of “yes,” “no,” and “maybe”), data with very few observations, or time-series data where you care about the order of values. For categories, use a bar chart. For trends over time, a line graph is almost always clearer. And if your dataset has fewer than about 20 to 30 data points, individual dot plots or stem-and-leaf plots will show more detail than bins can.
Discrete count data occupies a gray area. If you’re counting something that takes on many possible integer values (like the number of daily website visitors, which might range from 200 to 5,000), a histogram works fine because the data behaves similarly to continuous data at that scale. But if your count data only takes on a handful of values (say, the number of children in a household, ranging from 0 to 5), a bar chart for each value is more informative than forcing those values into bins.