A histogram shows how a list of numbers is spread out. The numbers are sorted into equal ranges called bins, and each bar’s height is how many values landed in that bin.
Made-up example data. The 70–80 bin holds 11 of the 30 scores: the most common range.
Where’s the peak? The tallest bar is the most common range of values.
How wide is it? Bars spread across a wide range mean the values vary a lot; a narrow cluster means they’re similar.
Is it lopsided? A long tail on one side is called skew, named for the side the tail is on.
Any gaps or strays? A bar sitting off on its own may be an outlier, or a mistake in the data.
Each bar includes its left edge and not its right. In the graph above, a score of exactly 80 is counted in 80–90, not 70–80.
Common shapes
Symmetric, one peak (bell-shaped)
Most values sit near the middle and thin out evenly on both sides. Heights and many test scores look like this.
Skewed right
A long tail toward the high values: most are low, a few are much higher. Incomes, house prices and waiting times often look like this. The mean is pulled above the median.
Skewed left
A long tail toward the low values: most are high, a few are much lower. Scores on an easy test often look like this. The mean is pulled below the median.
Two peaks (bimodal)
Two clusters, which often means two different groups are mixed together, such as the heights of children and adults.
Flat (uniform)
Every range is about equally common, like the results of rolling a fair die many times.
The five shape examples are 400 simulated values each, drawn so the shapes are clear.
Choosing the bins
The same data can look quite different with different bin widths. With too few bins, the shape collapses into two or three blocks; with too many, most bars hold one or two values and the outline turns ragged. Somewhere between 5 and 20 bars usually works. A common starting point is Sturges’ rule (1926): about log₂(n) + 1 bins for n values, which is what the histogram maker uses before rounding the width to a tidy number like 2, 5 or 10. If the shape changes a lot when you try a different width, there isn’t enough data to trust it.
Histogram or bar graph?
A histogram starts from a list of measurements; a bar graph starts from categories with a number each. The bars of a histogram touch and can’t be reordered; the bars of a bar graph have gaps and can be sorted. The full comparison →
The table behind a histogram is a frequency table: each bin, its count, its share of all values, and the running totals.