Free Error Bar Chart Maker
Charts & GraphsMake a bar graph with error bars online: enter means with SD, SE or 95% CI, or paste raw measurements and let the tool work them out. Free PNG, no signup.
Free PNGNo signupNo watermark
Loading tool...
Was this tool helpful?
How to Use
- 1
Choose supplied errors or raw measurements
Use Means & errors when you already have a mean and an error amount for each group. Use Raw measurements to paste individual values and calculate the means and errors. The example below uses two small synthetic groups.
- 2
Enter data and check the feedback
Type or paste your groups, then check the accepted values and, in raw mode, each group’s n. The limit is 12 groups and 2,000 raw values in total. Review any skipped-word or truncation warning; a group with one observation has a mean but no error bar.
- 3
Choose what the errors represent
In Raw measurements, SD, SE and 95% CI calculate different errors. In Means & errors, the choice only sets the caption: it does not convert your supplied numbers. Custom is available for supplied errors; edit its caption to explain them.
- 4
Label and style the result
Replace the starter title, subtitle and axis units by clicking their text on the chart; entering data does not update them. Choose Bars, Points or Points + line, a theme and a canvas size. Keep Caption: what the bars mean enabled and inspect the labels and error caps.
- 5
Download and inspect the PNG
Download PNG gives a free standard PNG without signup or a watermark, at the selected canvas dimensions. SVG and HD PNG require Pro or an active export pass. Save requires an account. Open the downloaded file to check the result before using it.
Why Choose GraphMake?
Try a Two-Group Error-Bar Example
These are synthetic measurements in arbitrary units, not results from a study. Choose Raw measurements and enter the following two rows on separate lines:
Group A: 8, 10, 12
Group B: 17, 20, 23
Check that both groups show n = 3. Their means are 10 and 20. With SD selected, the errors are 2 and 3; with SE, approximately 1.1547 and 1.7321. Enable Error in the labels to display the errors alongside the means.
Choose 95% CI. With 2 degrees of freedom, t is approximately 4.3027, giving interval half-widths of about 4.9683 and 7.4524. The chart rounds the labels to 10.0 ± 5.0 and 20.0 ± 7.5. These are distances around each mean, not the lower and upper endpoints.
Click the starter text to set the title to Illustrative measurements, the subtitle to Synthetic samples; arbitrary units, the vertical axis to Measurement (units), and the horizontal axis to Group. Keep the caption enabled, select Slide and download the 1280 × 720 PNG. Inspect its title, units, n and error bars.
To see a boundary case, add Single: 7 on a new line. It has n = 1 and no error bar. An extra word in a numeric row is skipped with a warning; check the accepted observations rather than assuming every pasted item was used.
How Raw-Data Calculations Differ From Supplied Errors
Raw mode averages each group and uses the sample standard deviation, dividing the sum of squared deviations by n − 1 before taking the square root. SE is SD ÷ √n. The 95% interval is mean ± t × SE, using tabulated t critical values and interpolation for larger degrees of freedom. Values are approximate, not exact distribution quantiles.
Interval width depends on both sample size and observed variation. For the same SD, fewer observations produce a wider t interval. The t method assumes independent observations and is most defensible for small samples when the underlying population is approximately normal; this tool does not check those assumptions.
In Means & errors, you supply the arithmetic result. Selecting SD, SE, 95% CI or Custom changes its description, not its amount. This mode can display intervals calculated elsewhere, including asymmetric ones. Keep your calculation method and sample sizes with the chart when reporting the result.
Read the Chart Without Claiming a Statistical Test
SD describes the spread of observations; SE and a mean confidence interval describe precision. Under the method’s assumptions, 95% refers to the long-run coverage of intervals across repeated samples. It does not mean that 95% of observations fall inside the interval or assign a 95% probability to a fixed population mean.
This maker does not run a significance test. Do not use error-bar overlap alone to decide whether groups differ: the appropriate comparison depends on the study design, including whether observations are paired. Calculate and report a suitable comparison separately when that is your question.
Bars start at zero so their lengths represent the means. Use points when a tighter axis is appropriate, and join them only when group order has meaning. Name the groups, state the units and keep the error definition visible. For a simple comparison of counts without uncertainty estimates, use bar chart maker.
What You Can Create
Lab Reports
Summarize repeated measurements with a mean and clearly identified SD, SE or confidence interval. Check the sampling assumptions and your report’s requirements.
Papers and Posters
Try the black-and-white Journal theme for a figure, then check the downloaded file against your required dimensions, format and caption style.
Survey Scores
Plot group means with separately calculated errors, and state how each estimate and interval was obtained. The raw-data calculation does not account for survey weights or a complex sampling design.
Teaching Statistics
Use the two-group synthetic example to compare SD, SE and 95% CI on the same observations and explain why their lengths differ.
Frequently Asked Questions
What do error bars show?
SD describes variation among individual measurements. SE describes the estimated precision of the mean. A 95% confidence interval uses a method intended to cover the population mean in 95% of repeated samples when its assumptions hold. It is not a range containing 95% of individual observations. Keep the caption enabled so the exported chart identifies the chosen meaning; raw mode also includes group sizes.
Should I use SD, SE or a 95% confidence interval?
Use SD when you want to show variation in the observations, and SE or a mean confidence interval when you want to show precision. In raw mode the studio calculates your selection. In manual mode it only labels the errors you supplied, so you must enter the matching amounts yourself. Custom lets you describe another supplied interval. Check any reporting requirements for your assignment or publication.
Can it calculate error bars from my raw data?
Yes. Choose Raw measurements and enter each group as a labeled row or spreadsheet column. The studio calculates the mean, sample SD using n − 1, SE = SD ÷ √n, and a 95% mean interval with half-width t × SE. It uses tabulated and interpolated approximations to Student’s t critical values for n − 1 degrees of freedom. A single observation has no estimated SD, SE or CI bar; the feedback and enabled caption explain this.
Can I edit the error bars directly on the chart?
In Means & errors, drag a bar top or point to change its mean and an error cap to change its error distance. The table updates with the drag. Click a group name to rename it; select a bar and use the arrow keys to nudge it or Delete to remove it. The dashed + adds a group within the 12-group limit. In Raw measurements, edit the underlying observations instead: clicking a bar shows its calculated values.
How do I make asymmetric error bars?
In Means & errors, switch Errors to Asymmetric and supply positive distances below and above the mean. Dragging each cap adjusts that side. Pasted columns headed Lower and Upper are treated as interval endpoints and converted to distances. For a mean of 10 with endpoints 8 and 13, enter distances 2 and 3, not endpoints 8 and 13, in the manual distance fields.
Can I paste data from Excel or Google Sheets?
Yes. In Means & errors, paste group/mean, group/mean/error, or group/mean/minus/plus columns. Header names can identify the error type or interval endpoints, and a cell such as 12.4 ± 1.3 is supported. Raw measurements accepts labeled rows or columns; choose Rows or Columns if Auto interprets the layout incorrectly. Only the first 12 groups and up to 2,000 raw values across them are used. Check warnings and group counts, especially when words or extra values are skipped.
What happens with negative values?
A negative mean is drawn below zero. Error distances are magnitudes: a negative manual distance is used as positive, with feedback showing the normalization. Bars start at zero, while Points and Points + line can use a tighter axis. The scale accounts for error endpoints, but you should still inspect labels and caps in the downloaded file.
What size can I download, and is it free?
Standard PNG is free without signup or a watermark: Slide is 1280 × 720, Figure is 960 × 720 and Square is 1080 × 1080. The on-page preview may be scaled to fit your screen. SVG and HD PNG at 2×, 3× or 4× require Pro or an active export pass. Saving a design to reopen later requires an account.
Related Tools
Free Bar Chart & Bar Graph Maker
Create bar charts, bar graphs, and column charts online for free. Use horizontal or vertical bars, grouped or stacked layouts, custom colors, and PNG export - no signup, no watermark.
Try it freeFree Histogram Maker
Create histograms online for free. Visualize frequency distributions with touching bars that show how data is spread across ranges. Download as PNG — no signup.
Try it freeFree Stem and Leaf Plot Maker
Make a stem-and-leaf plot online: paste your numbers and get the plot with its key, decimals, split stems, back-to-back plots, median and mode. Free PNG, no signup.
Try it freeFree Scatter Plot & XY Chart Maker
Plot two-variable data as points on an XY grid. Build a scatter plot, scatter chart, or XY chart online — spot correlations, outliers, and clusters. Free, no signup.
Try it free