Bar graph vs line graph: when to use each

The choice comes up in lab reports, team slides and homework that just says "graph the data." Picking the wrong chart can hide the point of your data or suggest a trend that isn't there, so it's worth a minute to check before you build it in the bar graph maker or the line graph maker.

The rule: categories or continuous time

When to use a bar graph vs line graph comes down to one question: does the space between your data points mean anything?

A bar graph compares things that stand alone. Favorite lunch in a class survey is a good example, because pizza, salad and sandwiches are separate choices and nothing sits between pizza and salad. Test scores by class work the same way, with each class as its own group and its own bar.

A line graph connects points in a sequence, and the line itself makes a claim that values exist in between and moved from one point to the next. Test scores across a semester fit, since each week follows the one before and a score halfway between two weeks is still a real number even if nobody measured it. Temperature through a day, website visits across a year and plant height each week work for the same reason.

You can test your own data by reordering the labels in your head. If you can shuffle them and the chart still says the same thing, use bars. If shuffling breaks the meaning because the order is time, use a line. Time order is the usual signal but not the only one, since a few time points can still work well as bars.

The same monthly data drawn both ways

Here's one made-up dataset, monthly visits to a small website over a year.

We built both versions in Chartjuice from this exact table. The bar graph is twelve bars, one per month, each starting at zero. The line graph is one line through twelve points, and Chartjuice marks each month with a small hollow circle and also starts the vertical axis at zero. The numbers are identical, but each version makes a different question easy to answer.

What the bar graph makes easy

Bars are good at single values. You can see right away that November is the tallest bar and February the shortest, and because every bar starts at zero, the heights compare honestly. November's 5,900 visits look about twice February's 2,900, which is what they are.

What bars make harder is the shape of the year. To see that traffic slipped in June and July and then climbed through the fall, your eye has to hop from the top of one bar to the next and build the pattern in your head, which is doable with twelve bars but takes effort.

What the line graph makes easy

A line shows direction at a glance. In our version, the February dip, the first high in May, the summer slide, the climb from August to November and the drop in December all show up as slopes, and the turning points are simply where the line changes direction.

What a line makes harder is reading one exact month. The hollow circles help you find each month, but you're still estimating its value against the gridlines, so turn on Value labels in the Style tab if exact numbers matter. A line also implies values between the months, which is fine here because traffic kept moving through each month.

How to choose between them

For this data, think about the question your reader will ask. If it's how one month compared with another, bars answer it faster. If it's what happened over the year, the line answers it faster. A report on traffic trends usually wants the line, and a slide that calls out November's record month can use bars.

Few time points vs many

Time data doesn't always call for a line, because the number of points matters as much as the kind of data.

Four quarters of sales read fine as bars. Each quarter is a chunk of the year, the question is usually which quarter was biggest, and with only four values a line would be three short segments with little trend to trace. Bars let you compare the chunks directly and label each one with its value.

Now take 24 months of the same sales, or a daily series with a point for every day. As bars, the chart turns into a wall of thin columns with x-axis labels that collide. As a line, it stays readable, and you can see the climbs, the dips and any repeating pattern, such as a summer slump two years in a row.

There's no hard cutoff. Twelve months, like the traffic table above, works either way. As a rough guide, if you can still read every bar label without squinting, bars are an option, and if you find yourself thinning out labels to make the axis fit, switch to a line. Chart size moves the line too, since a wide chart in a report has room for more bars than a small image on a slide. When you're unsure, paste the same table into both makers and compare them side by side.

Comparing groups: double bar graph or multiple lines

Two groups raise a new question, which is whether your reader should compare the groups at each point or follow each group's trend. Here are two classes' averages on four quarter exams, again made up for illustration.

With only four points, a double bar graph is the clear choice. Each exam gets a pair of bars standing side by side, so you can see at a glance that Class B beat Class A by 8 points on the second exam. The double bar graph maker builds this from the same two columns of data.

We also drew the same table as two lines. The lines cross three times, once between each pair of exams, which shows how the lead kept changing hands better than the bars do. For that version we set the vertical axis to run from 60 to 90 under Vertical axis, which is fine for lines but would exaggerate the gaps on a bar graph. With only four exams the bars are still easier to read, and the lines only pull ahead when you stretch the comparison to ten monthly quizzes, where grouped bars become twenty bars in crowded pairs. The gap between the lines is the comparison and the slopes are the trends.

Stacked bars answer a different question. They fit when the parts add up to a meaningful total, such as the number of students earning an A, B or C each month, where the full bar is the whole class. They're a poor fit for the table above, because adding two class averages gives a number that means nothing. When your segments do add up to something real, the stacked bar chart maker handles it.

• Few points and you want to compare groups at each one: grouped bars.
• Parts that add up to a total: stacked bars.
• Many points and you want to compare trends: one line per group.

Common mistakes to avoid

Most chart problems come from the chart type saying something the data doesn't, and four mistakes cover most of them.

Drawing a line between unrelated categories

A line tells the reader that values exist in between. Connect pizza, salad and sandwiches and the slope between pizza and salad means nothing, because nothing sits between them. If the labels can be shuffled without changing the meaning, use bars.

Using bars for a long daily series

A year of daily values as bars is a wall of thin columns with colliding labels. Switch to a line, or group the days into weeks or months and keep the bars, so the reader can find the pattern without squinting.

Starting a bar graph's axis above zero

A bar's height is its value, so the bar has to start at zero to be honest. Cut the axis well above zero and a small difference between two bars looks several times larger than it is. If the differences are too small to see from zero, a line graph is the better choice, since a line is read by slope and position rather than bar length. In the line graph maker you can set the range under Vertical axis with Min and Max, as we did for the two-class lines, and it's worth labeling the axis clearly when it doesn't start at zero.

Putting too many lines on one chart

Two or three lines are readable, but eight lines cross each other, the colors blur together and the reader has to trace each one with a finger. Cut the chart down to the lines that carry your point, or split it into two or three smaller charts.

Before you share any chart, look at it the way a stranger would and ask what the shape alone suggests. If it suggests something the data doesn't support, change the chart type or the axis.

Is a line plot the same as a line graph?

No. If your math homework says "make a line plot," it isn't asking for a line graph. In US classrooms, a line plot is a number line with an X or a dot stacked above a value each time that value shows up in the data.

Say a class records how many books each student read last month. On a line plot, you draw a number line covering the answers and put one X above a number for each student who gave it, so the tallest stack is the most common answer. Nothing is connected and there's no time axis. A line graph tracks change over time with a connected line, while a line plot counts how often each value appears. For homework that shows Xs over a number line, use the line plot maker, or the dot plot maker if you want dots instead.

Make your bar graph or line graph

If the side-by-side test favored bars, open the bar graph maker, and if it favored a line, open the line graph maker. Both take the same data, so switching is easy.

1. Click Paste data, paste your table from Excel or Google Sheets with the column names in the first row, and click Use this data. You can also click Upload file and choose a CSV, TSV, XLSX, XLS or ODS file, or type straight into the table under the chart.
2. Under Chart columns, check that the months or categories are the Category and that each number column you want is ticked under Values. In the line graph maker, each ticked column becomes its own line.
3. Open the Style tab to add a title and axis titles. In the line graph maker, Line options holds Smoothness, Line width, Points and Steps, and Multiple series switches between Side by side, Stacked and 100%.
4. Click Download and pick a PNG at 1x, 2x or 4x, or an SVG, or use Copy image to paste it into a slide.

Files are read in your browser and never uploaded, and a chart can hold up to 5,000 rows and 50 columns. Making and downloading charts is free with no signup, and the charts are yours to use anywhere, including commercial work. A free account saves up to 3 charts. For two groups, the double bar graph maker and the stacked bar chart maker cover the layouts above, and types of graphs covers every other chart choice.