Descriptive Statistics Data Visualization Frequency Distribution 22 min read September 13, 2026
BY: Statistics Fundamentals Team
Reviewed By: Minsa A (Senior Statistics Editor)

Histogram: Definition, Examples, and How to Read a Histogram

A histogram is a graph that shows how numerical observations are distributed across intervals. Paste your own numbers into the generator below, or scroll through the guide to understand bins, frequency, distribution shape, and common interpretation mistakes.

What You'll Learn
  • ✓ What a histogram is and how it differs from a bar graph
  • ✓ What bins are and how bin width affects the shape you see
  • ✓ How to calculate frequency, relative frequency, and percentage
  • ✓ How to read center, spread, skewness, and peaks from a histogram
  • ✓ Worked examples covering symmetric, right-skewed, and bimodal data
  • ✓ The automatic Histogram Calculator with adjustable bins

Interactive Histogram Generator

Histogram Calculator

Observations
Minimum
Maximum
Mean
Median
Std Dev
Bins
Bin Width
Bin Interval Frequency Relative Freq. Percentage

What Is a Histogram?

Definition
A histogram is a graph that displays the distribution of numerical data by grouping observations into consecutive intervals called bins and drawing a bar for each bin whose height represents the count (or another measure) of observations in that interval.
Relative Frequency = Frequency / n

The x-axis on a histogram carries numerical values, not categories. The bars represent adjacent intervals, so they touch for continuous data. That is the main visual difference from a bar graph, which displays categorical data with separated bars.

Histograms are one of the most common tools in descriptive statistics. They let you see the overall shape of a dataset at a glance: whether observations cluster in the center, whether there is a long tail on one side, and whether any gaps or unusual clusters exist.

🔎
Quick Answer

A histogram groups numerical observations into bins and draws a bar for each bin. Bar height = frequency. Bars touch because the intervals are continuous. That is the whole mechanism.

Parts of a Histogram

Each component of a histogram carries specific meaning. Knowing what each part represents makes reading and building one much more straightforward.

ComponentWhat It ShowsExample
X-axisThe numerical values or intervals of the variable being measuredExam scores from 0 to 100
Y-axisFrequency, relative frequency, or percentage for each binCount of students per score range
BarOne bin; height = count of observations in that intervalA bar from 70 to 80 with height 12 means 12 scores fell there
Bin widthThe span of each intervalA bin width of 10 means each bar covers a 10-unit range
Bin boundaryThe exact cut-point between adjacent bins[70, 80) means 70 is included, 80 is not
Touching barsVisual signal that the data is continuous, not categoricalNo gaps between bars for a height dataset

How to Make a Histogram

1

Collect the data

Gather your numerical observations. Each value will go into exactly one bin. There is no minimum sample size required by definition, but very small samples (fewer than about 10 values) rarely produce informative histograms.

2

Find the range

Range = Maximum − Minimum. This tells you the span of values your bins need to cover. For example, if scores run from 42 to 98, the range is 56.

3

Choose the number of bins

A rough starting point is Sturges' rule: k = 1 + log2(n). For 50 observations that gives about 7 bins. The section below explains other rules and why this choice matters.

4

Calculate bin width

Bin width = Range / Number of bins. Round up slightly so the bins cover all values. If range = 56 and k = 7, bin width = 56/7 = 8. You might round to 10 for clean labels.

5

Set the bin boundaries

Start the first bin at or below the minimum. Use a consistent convention: [lower, upper) means the lower boundary is included and the upper is not. Stick to that rule for all bins so no observation falls into two bins.

6

Count observations per bin

Go through each value and mark which bin it belongs to. The sum of all bin frequencies must equal n. If it does not, recheck your boundaries and counts.

7

Draw the bars

Draw a rectangle for each bin with width equal to the bin span and height equal to the frequency (or relative frequency). Bars should touch: no gaps between adjacent bins for continuous data.

8

Label the axes and add a title

The x-axis shows the variable name and the numerical scale. The y-axis shows "Frequency," "Relative Frequency," or "Percentage" depending on which you chose. A clear title names the variable and the dataset.

Histogram Bins and Bin Width

Bins are the intervals that divide the data's range into sections. Each observation falls into exactly one bin, and the bin's frequency is the count of observations it contains. Bin width is the span of each interval.

Bin Width Formula
Bin Width = Range / Number of Bins
Range = Maximum − Minimum k = number of bins

Automatic Binning Rules

No single rule produces the best histogram for all datasets. The goal is a bin count that shows the distribution's real shape without so many bins that the chart looks like noise.

RuleFormulaGood For
Sturges' rulek = 1 + log2(n)Approximately normal data; simple starting point
Square root rulek = √nVery quick estimate; tends toward more bins than Sturges
Scott's ruleh = 3.49 × s × n−1/3Approximately normal data; uses standard deviation
Freedman-Diaconish = 2 × IQR × n−1/3Skewed or heavy-tailed data; more robust to outliers

The calculator above uses Sturges' rule by default and Freedman-Diaconis as an alternative. If the IQR is zero (all values identical), it falls back to a single bin and reports the issue.

Effect of Bin Width on What You See

Bin width is one of the most important histogram decisions

The same dataset can look symmetric, bimodal, or nearly uniform depending purely on how many bins you choose. Always try a few different values before drawing conclusions about shape.

Too few bins merge distinct groups into a single blob. Too many bins turn a clean distribution into an irregular spike chart. The sweet spot is usually where you can see the overall shape clearly without the bars becoming so narrow that random gaps appear between adjacent counts.

Frequency, Relative Frequency, and Percentage

The y-axis of a histogram can show three related quantities. All three describe the same data; they just express it on different scales.

Frequency Measures
Relative Frequency = Frequency / n  |  Percentage = (Frequency / n) × 100
n = total number of observations Sum of all frequencies = n Sum of relative frequencies ≈ 1

Relative frequency histograms are especially useful when you want to compare two groups of different sizes, because the y-axis puts both groups on the same 0-to-1 scale regardless of how many observations each contains.

How to Read a Histogram

Reading a histogram means assessing four things: center, spread, shape, and any unusual features like gaps or isolated bars.

Center

The center is where observations are most concentrated. You can read a rough center visually by finding the approximate midpoint of the distribution. For a symmetric, unimodal histogram, the center is near the tallest bar. For a skewed distribution, the mean and median differ, with the mean pulled toward the longer tail. Keep in mind that a histogram does not directly show the exact mean or median; those require calculation from the raw data or from the frequency table.

Spread

Spread refers to how widely observations are distributed across the axis. A narrow histogram means values are concentrated; a wide one means they vary substantially. The full width of the histogram gives a rough sense of the interquartile range and overall standard deviation, though those metrics require direct calculation.

Shape

ShapeWhat You SeeCommon Context
SymmetricLeft and right sides roughly mirror each other; one central peakHeights, measurement errors
Right-skewedTail extends to the right; most values are on the leftIncome, waiting times
Left-skewedTail extends to the left; most values are on the rightExam scores when most students do well
UniformBars are roughly equal in height across the rangeDice rolls, random number generators
BimodalTwo distinct peaksMixed populations, shift work hours
MultimodalThree or more peaksMultiple subgroups in one dataset

Skewness

A right-skewed histogram has a longer tail pointing toward larger values. The mean is typically pulled toward that tail and sits above the median. A left-skewed histogram has a longer tail pointing toward smaller values, pulling the mean below the median. These are common patterns, not absolute rules; always check the actual statistics rather than relying entirely on a visual impression. You can learn more about skewness and kurtosis in our dedicated guide.

Peaks and Modes

The tallest bin is the modal bin. It is not necessarily the statistical mode of the raw data, especially when bin width is large relative to the data's variation. A bimodal histogram has two prominent peaks, which often suggests the data comes from two distinct groups or processes. Bimodality deserves investigation rather than assumption.

Gaps and Clusters

A gap is a bin with zero or very few observations. It can indicate that the variable has a natural break, that there are two separate populations, or simply that the sample is small and the gap is random. Clusters are regions where observations bunch together. Neither gaps nor clusters confirm anything on their own; they guide further questions about the data.

Outliers in a Histogram

A histogram can hint at possible outliers when a small bar appears isolated far from the main distribution. However, a histogram alone does not establish outliers according to any formal rule. For that you need the raw data and a criterion such as the IQR rule or a box plot. A distant bar might reflect a genuine unusual value, a data entry error, or simply a sparse region of the distribution.

Worked Histogram Examples

Example 1: Basic Histogram from Raw Data

Worked Example 1 — Simple Histogram

Data: 2, 3, 4, 5, 5, 6, 7, 8, 8, 9  (n = 10)

1

Range: 9 − 2 = 7

2

Bins (Sturges): k = 1 + log2(10) ≈ 1 + 3.32 ≈ 4 bins. Bin width = 7/4 = 1.75; round to 2 for clean labels.

3

Bin boundaries:

BinValues IncludedFrequencyRel. Freq.%
[2, 4)2, 320.2020%
[4, 6)4, 5, 530.3030%
[6, 8)6, 720.2020%
[8, 10]8, 8, 930.3030%
Total101.00100%

Interpretation: the distribution is roughly bimodal, with concentrations in the [4, 6) and [8, 10] bins. The data is not strongly skewed in either direction. With n = 10, the histogram provides a rough picture; a larger sample would give a clearer shape.

Example 2: Exam Score Histogram (Right-Tailed)

Worked Example 2 — Exam Scores (Illustrative Data)

25 illustrative exam scores: 42, 48, 51, 55, 57, 58, 60, 62, 63, 65, 67, 68, 70, 72, 74, 75, 78, 80, 82, 85, 88, 91, 94, 97, 99

1

n = 25, Min = 42, Max = 99, Range = 57

2

Bins (Sturges): k = 1 + log2(25) ≈ 6 bins. Bin width ≈ 57/6 = 9.5; round to 10 for clean intervals.

BinFrequencyRel. Freq.%
[40, 50)20.088%
[50, 60)40.1616%
[60, 70)60.2424%
[70, 80)50.2020%
[80, 90)40.1616%
[90, 100]40.1616%
Total251.00100%

Interpretation: the histogram is roughly bell-shaped with a peak in the 60s. There is a slight upper tail (four scores in the 90-100 range). This is fairly typical of exam score distributions where most students cluster in the middle range. The data are illustrative and do not represent any real class.

Example 3: Right-Skewed Data (Waiting Times)

Worked Example 3 — Right-Skewed Histogram

Illustrative waiting times (minutes): 1, 1, 2, 2, 2, 3, 3, 3, 3, 4, 4, 5, 5, 6, 7, 9, 12, 15, 22, 35

1

n = 20, Min = 1, Max = 35, Range = 34

2

Bins (Sturges): k ≈ 5. Bin width = 34/5 ≈ 7; use width 7.

BinFrequency%
[1, 8)1680%
[8, 15)210%
[15, 22)15%
[22, 29)00%
[29, 36]15%
Total20100%

Interpretation: most waiting times are short (under 8 minutes), and the distribution has a long right tail. This pattern is characteristic of right skew. The mean will be pulled above the median because the few large values stretch the average upward. Such a shape is common for service waiting times, where most customers are served quickly but an occasional long wait occurs.

Example 4: Bimodal Histogram

Worked Example 4 — Bimodal Distribution

Illustrative heights (cm) from a mixed group: 152, 154, 155, 156, 158, 159, 160, 161, 162, 163, 172, 173, 174, 175, 176, 177, 178, 179, 180, 182

1

n = 20, Min = 152, Max = 182, Range = 30

2

Bins: 6 bins with width 5. Boundaries: 152, 157, 162, 167, 172, 177, 182.

BinFrequency
[152, 157)3
[157, 162)6
[162, 167)1
[167, 172)0
[172, 177)5
[177, 182]5

Interpretation: two clusters appear: one around 157-162 cm and one around 172-182 cm, with a gap between them. This bimodal pattern often indicates the data comes from two distinct groups, for example, a mixed sample of men and women measured separately and then combined. The histogram prompts the question "are there two subgroups here?" rather than answering it; checking the raw data or grouping variable would be the next step.

Histogram vs Bar Graph

This comparison trips up a lot of students. The two charts look similar but represent fundamentally different types of data.

Feature Histogram Bar Graph
Data typeNumerical (quantitative)Categorical
BarsTouch (continuous intervals)Separated (distinct categories)
X-axisNumeric scale with bin boundariesCategory labels in any order
Bar widthRepresents the interval spanArbitrary; carries no data meaning
Bar orderFixed by numerical orderCan be reordered without distorting meaning
PurposeShow the distribution of one numeric variableCompare quantities across categories

The most reliable test: if you could reorder the bars and the chart would still make sense, it should be a bar graph. If reordering would destroy the meaning, it belongs as a histogram.

Histogram vs Box Plot

A histogram and a box plot describe the same data but emphasize different things. They are commonly used together.

🔑
Use both together when possible

A histogram shows distribution shape in detail. A box plot shows the five-number summary compactly and makes outliers easy to spot. Together they give a fuller picture than either alone.

ComparisonHistogramBox Plot
What it showsFull distribution shapeQuartiles, median, and potential outliers
Modes visible?YesNot directly
Skewness visible?YesYes, but less precisely
Outliers visible?Sometimes, as isolated barsExplicitly plotted as dots
Best forUnderstanding overall shapeComparing distributions across groups

See our complete guide to data visualization for more comparison tools.

Real-World Applications

🏫

Education

Exam score distributions, grade distributions, time-on-task measurements for students.

📈

Business

Sales transaction amounts, customer purchase values, order processing times.

🏥

Healthcare

Blood pressure readings, cholesterol levels, patient age distributions in clinical trials.

🏭

Manufacturing

Component measurements, defect counts, process variation around a target value.

🏞

Finance

Daily returns, transaction amounts, loan repayment periods.

🌧

Research

Survey response distributions, experimental measurements, time-series values.

Advantages and Limitations

Advantages

  • Shows distribution shape at a glance: symmetric, skewed, bimodal, or uniform
  • Handles large datasets well, where dot plots become cluttered
  • Reveals where observations concentrate and where they are sparse
  • Can suggest whether a normal distribution might be plausible
  • Frequency tables that underlie histograms are easy to audit and reproduce

Limitations

  • The appearance depends on bin choice; different bins can imply different shapes
  • Individual observations are not visible in the chart
  • Very small samples rarely produce informative histograms
  • Two histograms with different bin boundaries cannot be compared directly
  • A histogram does not prove that data follow a normal or any other named distribution
  • It shows only one variable; it cannot show the relationship between two variables

Common Histogram Mistakes

MistakeWhat Goes WrongWhat to Do Instead
Using a histogram for categories The bars have no numeric meaning, so the chart is misleading Use a bar graph for categorical data
Leaving gaps between continuous bins Implies there are no values between the bars, which is false for continuous data Make bars touch for continuous numerical data
Calling the tallest bar the mean The modal bin and the arithmetic mean are different quantities Calculate the mean from raw data or the frequency table
Assuming the histogram proves normality A bell-shaped histogram is consistent with normality but does not confirm it Use a formal normality test such as the Shapiro-Wilk test or QQ plot
Using different bins when comparing two datasets Different boundaries make visual comparison invalid Synchronize bin boundaries before putting histograms side by side
Automatically labeling a distant bar as an outlier A small isolated bar might not meet any formal outlier criterion Apply the IQR rule or box plot before deciding
Truncating the y-axis Makes small differences look large; distorts visual comparison Start the y-axis at zero for frequency histograms

Practice Problems

Practice 1: Construct a frequency table for the data set: 3, 5, 7, 7, 9, 11, 13, 15, 15, 17. Use 4 bins starting at 3 with width 4.
Bins: [3, 7) freq 2, [7, 11) freq 4, [11, 15) freq 2, [15, 19) freq 2. Total = 10. Relative frequencies: 0.20, 0.40, 0.20, 0.20. The distribution is fairly uniform with a slight concentration in the middle.
Practice 2: A histogram shows bars at [0, 10), [10, 20), [20, 30), [30, 40) with heights 2, 8, 6, 4 respectively (n = 20). What is the relative frequency of the [10, 20) bin?
Relative frequency = 8/20 = 0.40, or 40%. The [10, 20) bin contains the most observations. The histogram is approximately right-skewed with the bulk of values in the lower two bins.
Practice 3: The same data is plotted with 3 bins and then with 15 bins. The 3-bin version looks roughly symmetric. The 15-bin version shows noisy irregular spikes. What does this tell you?
Three bins mask all internal structure by merging everything into three large blocks. Fifteen bins are probably more than the sample size can support; individual random variation between adjacent bins dominates over the true distribution shape. A moderate number of bins, somewhere in between, will typically reveal the actual distribution most clearly.
Practice 4: A histogram of incomes has most bars near the left and a thin tail extending far to the right. Is the distribution left-skewed or right-skewed? Where would you expect the mean relative to the median?
Right-skewed: the tail points toward higher values on the right side of the x-axis. The mean is typically pulled toward the tail and would sit above (to the right of) the median. This pattern is very common in income and wealth distributions, where a small number of high earners pull the average above the midpoint value.

Frequently Asked Questions

What is a histogram in statistics?
A histogram is a graph that displays the distribution of numerical data. It groups observations into consecutive intervals (bins) and draws a rectangle for each bin whose height equals the frequency, relative frequency, or percentage of observations falling in that interval. Histograms are used to understand the shape, center, and spread of a numerical variable.
What is the difference between a histogram and a bar graph?
A histogram uses numerical intervals on the x-axis and bars touch because the data is continuous. A bar graph uses categories on the x-axis and bars are typically separated. You cannot reorder the bars of a histogram without changing its meaning; you can reorder the bars of a bar graph freely.
What is a bin in a histogram?
A bin (also called a class interval) is a consecutive range of values that groups observations together. Each observation belongs to exactly one bin. For example, the bin [20, 30) includes all values from 20 up to but not including 30. The number of bins and the bin width control how finely the distribution is divided.
Do histogram bars always touch?
For continuous numerical data, yes. Touching bars signal that the intervals are adjacent with no gap, which is correct for measurements that can take any value. For discrete data (like counts) some practitioners add small gaps, though conventions vary. What matters is that the visual correctly represents the interval structure of the bins.
Can a histogram have negative values on the x-axis?
Yes. A histogram can display any numerical data, including negative numbers. The x-axis simply spans from the minimum to the maximum of the data, whether those values are negative, zero, or positive. Temperature measurements, financial returns, and standardized scores frequently include negative values.
What is a relative frequency histogram?
A relative frequency histogram uses relative frequency (frequency divided by n) on the y-axis instead of raw counts. Each bar shows the proportion of observations in that bin, and all bars together sum to 1. This is especially useful when comparing two datasets of different sizes, since both histograms can be placed on the same 0-to-1 scale.
How do you choose the number of bins?
Several rules exist: Sturges' rule (k = 1 + log2(n)) works well for roughly normal data. The Freedman-Diaconis rule uses the IQR and is more robust for skewed or heavy-tailed data. The square root rule (k = sqrt(n)) provides another quick estimate. No rule is universally best; try two or three bin counts and pick the one that best reveals the distribution's real shape without excessive noise.
Does a histogram prove that data are normally distributed?
No. A roughly bell-shaped histogram is consistent with normality but does not confirm it. The visual impression depends on bin width, and small samples can look normal by chance. For a formal assessment, use a normality test (Shapiro-Wilk, Anderson-Darling) or a QQ plot. See our guide to normality tests for details.
What is the Freedman-Diaconis rule?
The Freedman-Diaconis rule calculates bin width as h = 2 x IQR / n^(1/3), where IQR is the interquartile range and n is the sample size. It tends to use fewer, wider bins than Sturges' rule for heavily skewed data because it bases the calculation on the IQR rather than the overall range, making it less sensitive to extreme values.
When should I use a histogram instead of a box plot?
Use a histogram when you want to see the full shape of the distribution, especially to detect modes or unusual patterns like gaps and clusters. Use a box plot when you want a compact summary (quartiles, median, IQR) or when comparing several groups side by side. The two charts complement each other well and are often used together in exploratory analysis.

Summary

A histogram groups numerical observations into bins and draws a bar for each one. Bar height equals frequency. The x-axis is a numeric scale with bin boundaries, and bars touch for continuous data. That structure makes histograms distinct from bar graphs, which show categorical comparisons.

Reading a histogram means looking at center (where observations concentrate), spread (how wide the distribution is), and shape (symmetric, right-skewed, left-skewed, bimodal, or uniform). Bin width changes what you see, so trying a few values before drawing conclusions is good practice.

The Histogram Calculator above accepts any numerical data you paste in, applies automatic or manual binning, and shows the frequency table alongside the chart. For related tools, see the Descriptive Statistics section, the Mean, Median, and Standard Deviation guides, or the Descriptive Statistics Calculator for a full summary of any dataset.