Different Types of Graphs: When to Use Each One and How to Read It

Short answer

• Different types of graphs answer different questions. A bar graph compares categories, a line graph shows change over time or in order, a pie chart shows parts of one whole, a histogram shows how values spread out, and a scatter plot shows the relationship between two numbers.
• Start with your question, not with the chart style. If you cannot state the question in one sentence, the graph will probably confuse the reader even if it looks neat.
• Match the graph to the data type. Categories with no natural order, such as regions or product types, suit a bar graph. Ordered values, such as months or years, suit a line graph. A single total split into shares suits a pie chart.
• Check the axes before you trust the message. A bar graph whose value axis does not start at zero can make a small difference look large. A line graph with uneven gaps between dates can make a slow change look sudden.
• A graph is not always the best choice. When exact numbers matter more than the pattern, a small table is often clearer than a chart.
• In mathematics, the word graph has a second meaning. It can also mean the picture of a function, such as a straight line or a curve on a coordinate grid, or a network of points joined by lines. This guide focuses on data graphs and charts, and explains that second meaning briefly near the end so the two uses are not confused.

How to Choose a Graph in Four Questions

First, what is the question? Are you comparing groups, showing change over time, showing shares of a whole, showing a distribution, or showing a relationship? Each job points to different types of charts.

Second, what kind of data do you have? Categories are labels such as city names; measured numbers, such as temperature, behave differently. Third, how many values are you showing? A pie chart with three slices works well, while fifteen thin slices do not, and ten overlapping lines usually need splitting or a table. Fourth, who will read it? A simple bar graph often beats a clever chart when the answer is needed in seconds.

Question you want to answer	Good first choice	What to read from it
Which category is largest, and by how much?	Bar graph	Bar lengths against a shared scale
How has a value changed over time?	Line graph	Direction, slope, and turning points
What share of a whole does each part take?	Pie chart or stacked bar	Slice or segment size as a share of the total
How are values spread out?	Histogram or box plot	Center, spread, and unusual values
Do two numbers move together?	Scatter plot	Pattern, direction, and strength of the pattern
How do parts add up to a total over time?	Stacked area chart	Total height plus the changing mix inside it

Bar Graphs

A bar graph uses rectangles to compare values across categories. In a vertical bar graph, often called a column chart, categories sit along the horizontal axis and bar heights show the values. In a horizontal bar graph, categories sit along the vertical axis and bar lengths show the values. Horizontal bars suit long labels, such as full country names, because the labels have room to read.

Bar graphs suit comparisons: sales by region, survey answers by age group, or scores by subject. The reader compares bar lengths against a shared baseline, so the value axis should normally start at zero. If it starts higher, visible bar lengths no longer match real differences, and a small gap can look large.

Grouped bar graphs place bars side by side for each category, so subgroups can be compared. Stacked bar graphs place segments on top of each other in a single bar, showing total and mix. Stacked bars work when the total matters and parts are few; for exact comparison of middle segments, grouped bars or a table are clearer, since stacked segments share no common baseline.

Read a bar graph in this order: title, category labels, value axis and unit, then bar lengths. Ask whether values are counts, percentages, averages, or totals, and whether a missing bar means zero or no data. Sort by size when categories have no natural order; keep natural order, such as months, when order carries meaning.

Line Graphs

A line graph places points on a grid and joins them with lines, usually to show change over time or across an ordered sequence. The horizontal axis often shows time, and the vertical axis shows the measured value. Rising means increase, falling means decrease, flat means little change, and a sharp bend marks a turning point.

Line graphs suit trends: temperature through a day, visits by week, a share price by month, or scores across school terms. Two or three lines together can compare trends, such as rainfall in two cities across the same months. Keep the number of lines small, or the graph shows clutter instead of a trend.

Spacing of points matters. If measurements were taken at irregular intervals, horizontal spacing should reflect the real gaps, or the graph should say spacing is even for display only. Equal spacing for unequal gaps can make a slow change look sudden. Missing values need care too: joining a line across a long gap hides the fact that no measurement exists. A break, a note, or a dotted segment is more honest than a smooth line that invents continuity.

Read a line graph by checking the time range first, then the overall direction, then peaks, dips, and flat stretches. Ask whether the vertical axis starts at zero. A line graph need not start at zero as a bar graph should, but a narrow range can still exaggerate small changes. Also ask what the points represent: single measurements, averages, or totals for each period. A stepped line graph, which stays flat then jumps at each new value, suits values that change in steps, such as a price that holds steady then changes on a set date.

Pie Charts and Donut Charts

A pie chart divides a circle into slices. Each slice stands for one category, and its size stands for that category’s share of a single whole. The slices together make up the whole, so their shares should add to 100 percent. A donut chart is the same idea with the center cut out, leaving room for a total or a label in the middle.

Pie charts suit one simple message: how a total breaks down. Examples include a household budget split across categories, survey answers to a single question, or a day’s time divided between activities. They work best with a small number of categories, each large enough to see and label. When one slice dominates and others are tiny, a bar graph often shows the small categories more clearly, because bar lengths are easier to compare than slice angles.

Read a pie chart by finding what the whole circle represents before comparing slices. Check whether percentages are labeled on the slices or only in a legend. A legend forces the reader to match colors back and forth, which slows reading and invites error. Direct labels are usually better. Do not use a pie chart for values that do not add to a whole, such as counts from overlapping groups, or to compare change over time without stating the totals. A slice that grows as a share can still shrink in absolute terms if the total fell. Avoid three dimensional pie charts: tilting the circle distorts slices, so front slices look larger than back slices even when shares are equal.

Histograms

A histogram looks like a bar graph, but it answers a different question. Instead of comparing separate categories, it shows how measured values are distributed. Values are grouped into equal sized intervals called bins, and each bar shows how many values fall into that bin. The horizontal axis is a number scale, and the bars touch each other because bins sit next to each other. This differs from an ordinary bar graph, where gaps signal separate categories.

Histograms suit questions about spread: exam scores, delivery times, heights in a class, or daily temperatures over a year. The shape tells the reader where most values sit, whether values cluster around one peak or two, whether the shape leans to one side, and whether any values sit far from the rest.

Bin width changes what the reader sees. Very wide bins can hide real structure, such as two peaks merging into one hump. Very narrow bins can break the shape into noisy spikes. There is no single correct bin count for every data set, so choose bins that reveal the main shape without inventing detail, and say how bins were set when the choice matters. Read a histogram by looking at shape first, then center, spread, gaps, and outliers. Ask what unit and bin range the horizontal axis uses, and whether the vertical axis shows counts, shares, or percentages. A relative histogram shows proportions; a cumulative histogram adds bins so the final bar reaches the total. Label both clearly.

Scatter Plots

A scatter plot places one dot for each observation, using one number for the horizontal position and another for the vertical position. It suits the question of whether two numbers are related: height and weight, study time and test score, advertising spending and sales, or temperature and ice cream sales over a summer.

Look for a pattern in the cloud of dots. Dots rising left to right suggest larger values of one variable tend to go with larger values of the other. Dots falling left to right suggest the opposite. No clear pattern suggests little or no simple relationship. Dots may also curve, cluster in groups, or include a few far from the rest. A straight trend line can summarize overall direction, but it is a summary, not a claim that every point follows it exactly.

A scatter plot shows association, not proof of cause. Two variables can move together because one causes the other, because both respond to a third factor, or by coincidence in a small data set. Study time and test scores might move together because practice helps, but also because motivated students both study more and prepare in other ways. A scatter plot alone cannot settle which explanation is right. Outliers can also pull a trend line, so check whether the pattern survives without them, and state how many observations the plot contains, since a pattern from few points may vanish with more data.

Bubble charts extend a scatter plot with a third number, bubble size. Readers judge area less precisely than position, and large bubbles can hide smaller ones, so labels and a size key help.

Area Charts

An area chart is a line graph with the space below the line filled in. It suits the same trend questions as a line graph, but the filled area draws attention to the size of the value over time, not only its direction. A single area chart can show total energy use by month or total rainfall across a season.

A stacked area chart fills layers on top of each other, so total height shows the combined total and each layer shows one part, suiting a changing mix such as energy sources over years. Only the bottom layer and total share a stable baseline, so middle layers are harder to judge; small line graphs, one per part, are clearer when exact values matter. A 100 percent stacked area chart rescales each point to full height, showing only the mix, not the total, so labels must make that clear. Check that the area starts at zero.

Box Plots

A box plot, also called a box and whisker plot, summarizes a distribution in compact form. The box shows the middle half of the values. A line inside the box marks the median, the middle value when values are ordered. Lines or whiskers extending from the box reach toward lower and higher values, and dots beyond the whiskers mark values far from the rest.

Box plots suit comparisons of distributions: test scores across classes, delivery times across depots, or salaries across roles. A reader can compare medians, compare the spread of the middle half, and spot unusual groups, all in a small space. Several box plots side by side make this easy in a way that several histograms, each needing its own panel, do not.

The weakness is what a box plot hides. Two distributions can share the same box shape while differing in important ways, such as one having two peaks and the other one. A box plot also hides sample size: a box from ten values and a box from ten thousand values can look identical. State the sample size, and pair the plot with a histogram when the shape itself matters. Read it by finding the median first, then the box edges marking the middle half, then whiskers and outlying dots. Ask what whisker rule was used, since different tools use different rules.

Heatmaps

A heatmap uses color to show values in a grid. Rows and columns stand for categories, such as days of the week and hours of the day, and each cell’s color stands for its value. Darker or warmer colors usually mean larger values, but the exact mapping depends on the color scale, so the scale must be shown and labeled.

Heatmaps suit patterns across two categories at once: visits by day and hour, sales by product and region, or temperature by month and city. A reader can spot busy periods, quiet periods, and repeating patterns, such as weekday peaks and weekend dips, faster from color than from a large table.

Color choices matter. A scale jumping between strong colors can imply a sharp break where data change slightly, while a pale scale can hide differences. A scale varying in lightness, plus labels on key cells, helps readers with color vision differences. Never rely on color alone: title, labels, and a few values should carry the finding even in black and white. Read it by checking the scale first, then scanning for blocks of similar color, then cells breaking the pattern. Ask whether values are counts, averages, or percentages.

Radar Charts, Pictographs, and Combination Charts

A radar chart, sometimes called a spider chart, places axes radiating from a center, one per variable, and joins the values into a shape. It suits profile comparisons, such as two products across price and speed, but its shape depends on axis order and its area is hard to judge, so use it only with few variables on a shared scale. A pictograph uses repeated icons, such as one person icon per hundred people, for simple messages; state what each icon represents and avoid partial icons that imply false precision. A combination chart overlays bars with a line for two series with different scales, but dual axes can create a misleading apparent relationship, so label both axes clearly and check whether two separate charts would be more honest.

How to Read a Graph

Read the title first and ask what question the graph claims to answer. Read axis labels and units next, including whether values are counts, percentages, averages, or rates. Read the legend, and check whether any category is grouped as other, which can hide detail. Read footnotes, source line, and date range. A graph without a source, a date, or a clear definition of its values should be treated with caution, even if it looks professional.

Then look at the pattern: largest value, smallest value, overall direction, gaps, breaks, and unusual points. Do labels match the data, or does the title promise more than the graph shows? Finally, ask what is missing. Averages can hide variation, totals can hide population size, and one group alone can hide the comparison group that would change the conclusion.

Take a worked example. A bar graph titled “Sales by Region” shows the North tallest. Before concluding the North is strongest, check whether bars show total sales or sales per store, and whether the axis starts at zero. If the North has twice as many stores, total sales alone do not prove stronger performance per store.

Common Misreadings and How to Avoid Them

Truncated axes are the most common trap in bar graphs. Cutting the axis so it starts near the smallest value stretches visible differences. Start at zero for bar graphs, or label a truncated axis clearly and explain why. In line graphs a narrow range is less deceptive, but still needs a clear label so small changes are not mistaken for large ones.

Cherry picked time ranges are another trap. A line graph starting just after a dip shows only a rise, while a longer view might show a cycle. Check start and end dates, and ask whether a longer or shorter range would change the story. The same applies to categories: dropping an inconvenient category without saying so presents a partial comparison as complete.

Correlation is not causation, and scatter plots are where this warning matters most. A strong pattern justifies a closer look, not a causal claim. Name alternative explanations, such as a third factor or a shared trend over time, before concluding. Percentages without totals also mislead: a share can rise while the absolute number falls, and a rate can look high in a small group because a few cases move the percentage a lot. Put the total near any headline percentage.

Three dimensional effects, decorative icons, and overloaded colors all make graphs harder to read. Depth distorts heights and slices, icons scaled in both height and width exaggerate large values, since area grows as the square of the scaling, and too many colors force legend decoding instead of pattern reading. Missing data should never be smoothed over silently: a line joined across a gap, a dropped category, or a zero used where data were not collected all create false precision. A short note, such as data not collected for this period, costs little space and prevents a wrong conclusion.

Data Charts and Mathematical Graphs: Clearing Up the Confusion

The phrase different types of graphs causes confusion because graph has two common meanings. In data writing, a graph is a chart: a bar graph, a line graph, or a pie chart. In mathematics, a graph can also mean the picture of a function on a coordinate grid. A straight line shows a linear function, a parabola shows a quadratic function, and a wave shape shows a trigonometric function such as sine or cosine. In a function graph, axes show input and output values of the function, not survey counts.

Mathematics also uses graph for a network: points, called vertices, joined by lines, called edges. A road map, a social network diagram, and a computer network diagram are graphs in this sense. Points and connections matter, not bar heights or slice sizes. These meanings share a name but not a method. This guide treats data charts as the main topic, because that is what most searches for different types of graphs ask about, and notes the mathematical meanings here so readers who meet them elsewhere can place them correctly.

When a Table Beats a Graph, and Making Graphs Accessible

A graph shows patterns, but a table shows exact values, so a table wins when the reader needs a specific number or must check a calculation, and when the data set is small. For accessibility, use direct labels rather than color only legends, add patterns or shapes as well as color, and write alt text stating the main finding, not only the chart type. Titles and labels must stay readable at published size.

Choosing the Right Graph: A Short Checklist

Before publishing a graph, run through this checklist:

1. State the question in one sentence. If the graph does not answer it directly, choose another type or rewrite the question.
2. Name the data type for each variable: category, ordered value, or measured number.
3. Count the categories, lines, or slices. If the count is high, split the graph, simplify it, or use a table.
4. Check the axes: labels, units, start values, and spacing of time points.
5. Check the whole: for a pie chart or stacked chart, confirm parts add to a stated total.
6. Label values directly where the reader needs them, and keep the legend short.
7. Add the source, the date range, and a note about missing data.
8. Read the graph as a stranger would, and ask whether the first impression matches the data.

If any step fails, fix the graph first.

FAQ

What is the simplest way to choose between a bar graph and a line graph?

Ask whether the horizontal axis shows categories or an ordered sequence. Categories with no natural order, such as regions, suit a bar graph. Values following time, where change is the story, suit a line graph. When months are compared as separate categories rather than as a trend, a bar graph can still be right.

What is the difference between a bar graph and a histogram?

A bar graph compares separate categories, and its bars are normally separated by gaps. A histogram shows measured values grouped into bins on a number scale, and its bars touch because bins sit next to each other. In a bar graph the horizontal axis lists categories. In a histogram it is a number scale, and bar height shows how many values fall into each bin.

When should I use a pie chart instead of a bar graph?

Use a pie chart when the message is the share of a single whole and parts are few. Use a bar graph when exact comparison matters, parts are many, or parts must be compared across groups or time. Bar lengths are judged more precisely than slice angles.

Can a scatter plot prove that one thing causes another?

No. A scatter plot shows whether two numbers tend to move together, and how strong that tendency looks. It does not show why. A causal claim needs further evidence, such as a controlled comparison or a mechanism explaining the link, and should rule out plausible alternatives first.

Why does a bar graph axis usually start at zero?

Because the reader judges a bar by its length. If the axis starts above zero, visible lengths no longer match real values, and small differences look large. Line graphs differ, since the reader judges slope and position rather than filled length, but a very narrow range still needs a clear label.

How many slices should a pie chart have?

Keep the number small enough that every slice can be labeled and compared without effort. A pie chart with a handful of clear slices works well, while many thin slices should usually become a bar graph or a table. Grouping the smallest slices as other can help, but only if the grouped share is labeled and the grouping is stated.

What is a box plot used for?

A box plot summarizes a distribution: median, middle half, and extremes. It is most useful for comparing distributions across groups, such as scores across classes. It hides detailed shape, so pair it with a histogram when shape matters, and state the observation count behind each box.

Should I use a 3D graph?

In most cases, no. Three dimensional effects distort heights, slice sizes, and positions. A flat graph with clear labels is easier to read. Reserve three dimensional displays for data that genuinely have three dimensions.

Sources & references

• University of Sheffield, MEE Group, "Types of graph (and when to use them)": bar charts for comparing results under different conditions, pie charts for showing how a total breaks down, and scatter plots for showing relationships between variables. https://mee.group.shef.ac.uk/Report_writing_website/4.8.html
• National Institute of Standards and Technology, Dataplot Reference Manual, "HISTOGRAM": a histogram divides a variable into equal sized intervals or bins, counts the occurrences in each bin, and plots frequencies or relative frequencies on the vertical axis against the variable on the horizontal axis. https://itl.nist.gov/div898/software/dataplot/refman1/auxillar/histogra.htm
• NIST/SEMATECH e-Handbook of Statistical Methods, histogram chapter, as cited in the NIST Dataplot reference above: https://www.itl.nist.gov/div898/handbook/