Choosing the right graph starts with one question: what are you trying to show? A bar chart compares categories, a line chart shows change over time, a scatter plot reveals the relationship between two numbers, and a pie chart shows parts of a whole. Match the graph to the question and the rest usually falls into place.
Most bad charts fail for the same reason. They were picked for how they look, not for what the data actually needs to say. The guide below walks through the main chart types, when each one works, and the common mistakes that make a chart misleading even when the numbers are correct.
How do I choose the right graph for my data?
Start with the shape of your data, not the chart menu. Two things decide almost everything: whether your variables are categories or numbers, and whether time is involved.
Categorical data puts things into groups. Months, brands, age brackets, regions. Numerical data is measured on a scale. Temperature, weight, income, blood pressure. Time-series data is numerical data collected in order over time.
Once you know which type you have, the choice narrows fast. Use this as your starting map:
- Comparing categories: bar chart
- Change over time: line chart
- Relationship between two numbers: scatter plot
- Parts of a whole: pie chart, used sparingly
- Distribution of one variable: histogram or box plot
That map covers most everyday needs. The sections below explain why each chart works and where it breaks down.
Should I use a bar chart or a line chart?
Use a bar chart when you are comparing separate groups. Use a line chart when the same thing is being tracked across a continuous sequence, usually time.
The key difference is connection. Bars represent distinct items that do not flow into each other. Sales by region. Votes by candidate. A line implies continuity, which is why it suits data where one point naturally leads to the next, like daily temperature or monthly revenue.
A common mistake is putting categorical data on a line chart. If the x-axis lists unrelated items, a line draws connections that do not exist. A line from “North” to “South” to “East” suggests a trend that means nothing.
Bar charts have their own traps. Starting the y-axis above zero exaggerates differences between bars. For most bar charts, the axis should start at zero. Line charts are more forgiving here, because the line shows rate of change rather than the size of each value.
When does a scatter plot make sense?
A scatter plot shows whether two numerical variables are related. Each dot represents one observation, placed by its value on two axes.
If you want to know whether taller people tend to weigh more, or whether study time relates to test scores, a scatter plot puts the pattern on display. A cluster rising from left to right suggests a positive relationship. A cluster falling suggests a negative one. A shapeless cloud suggests no clear relationship.
One clarification matters here. A scatter plot can show that two variables move together. It cannot show that one causes the other. That distinction is easy to forget when a chart looks convincing.
Scatter plots also reveal things summary statistics hide. A single outlier far from the rest of the dots can change an average without being obvious in a table. Seeing it plotted makes the odd value impossible to miss.
Is a pie chart ever the right choice?
A pie chart works when you are showing parts of a single whole and you have very few slices. Three or four is comfortable. Six is pushing it.
The problem is human perception. People are good at comparing lengths and positions. They are poor at comparing angles and areas. Research on how people read graphs has consistently found that bar charts are judged more accurately than pie charts.
When slices are similar in size, a pie chart makes them look nearly identical. A bar chart of the same data makes the differences clear.
Pie charts also fail when the slices do not add up to a meaningful whole. If you are comparing independent categories rather than parts of one total, a bar chart is the honest choice. Save the pie for a genuine breakdown, like budget allocation or survey responses that sum to 100%.
What graph should I use to show a distribution?
When you want to show how values are spread out, not just their average, use a histogram or a box plot.
A histogram groups numerical data into ranges and shows how many values fall in each. It reveals the shape of the data. Whether it clusters around a middle value, spreads evenly, or has two peaks. Two peaks often signal that you are looking at two different groups mixed together.
A box plot compresses a distribution into a compact summary. It shows the median, the spread of the middle half of the data, and any outliers. Box plots are useful for comparing several distributions side by side. They are less useful for a general audience, because many readers do not know how to read them.
This is where averages mislead. Two datasets can share the same average and look completely different once plotted. A histogram exposes that difference. A single number cannot.
How can a graph mislead even when the data is correct?
A chart can be technically accurate and still give the wrong impression. The data does not have to be wrong for the message to be.
Watch for these patterns:
- Truncated axes. A bar chart that starts at 90 instead of zero makes small differences look huge.
- Inconsistent scales. Mixing units or changing the axis between panels invites false comparisons.
- Too much decoration. 3D effects, shadows, and extra colors distort how values are read.
- Missing context. A number with no baseline or comparison point tells the reader very little.
- Cherry-picked ranges. Choosing a start and end date that happens to show a trend can hide the fuller picture.
The 3D pie chart deserves its own mention. Tilting a pie makes the front slices look larger than the back ones. That is a distortion built into the format, not a mistake in the data.
What should I check before finalizing a graph?
Run through a short checklist before you publish or present any chart.
First, confirm the chart answers the question you started with. If you wanted to show a trend over time, does the chart make the trend obvious within a few seconds? If not, simplify.
Second, check the axis. Does the y-axis start where it should? Are the units labeled? Is the scale consistent?
Third, label clearly. Every axis needs a title and units. The chart needs a title that states what it shows, not just what it is about. “Monthly Sales Rose 12% From January to June” tells the reader more than “Sales.”
Fourth, reduce clutter. Remove gridlines that add nothing, colors that carry no meaning, and labels the reader does not need. A clean chart is easier to trust.
Fifth, ask someone unfamiliar with the data to read it. If they cannot tell you what it shows in a few seconds, the chart needs work. This simple test catches problems that are invisible to the person who made the chart.
One habit worth building: write the sentence you want the reader to take away before you build the chart. Then build the chart that makes that sentence obvious. The chart serves the message, not the other way around.
Frequently Asked Questions
What is the best graph for comparing categories?
A bar chart is the best choice for comparing separate categories. It lets readers judge differences by length, which people read more accurately than angles or areas.
When should I use a line chart instead of a bar chart?
Use a line chart when your data is continuous, especially over time. Bars suit distinct groups, while a line shows how one value changes across a sequence.
Can a pie chart have more than five slices?
It can, but it becomes hard to read. Small or similar-sized slices are difficult to compare by eye, so a bar chart usually communicates the same data more clearly.
Does a scatter plot prove that one thing causes another?
No. A scatter plot shows whether two variables are related, not whether one causes the other. Establishing cause requires controlled studies, not a chart.

