Error Bars Considered Harmful: Exploring Alternate Encodings for Mean and Error
Open accessAbstract
When making an inference or comparison with uncertain, noisy, or incomplete data, measurement error and confidence intervals can be as important for judgment as the actual mean values of different groups. These often misunderstood statistical quantities are frequently represented by bar charts with error bars. This paper investigates drawbacks with this standard encoding, and considers a set of alternatives designed to more effectively communicate the implications of mean and error data to a general audience, drawing from lessons learned from the use of visual statistics in the information visualization community. We present a series of crowd-sourced experiments that confirm that the encoding of mean and error significantly changes how viewers make decisions about uncertain data. Careful consideration of design tradeoffs in the visual presentation of data results in human reasoning that is more consistently aligned with statistical inferences. We suggest the use of gradient plots (which use transparency to encode uncertainty) and violin plots (which use width) as better alternatives for inferential tasks than bar charts with error bars.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Visualizations that place less emphasis on confidence range limits might better represent the continuous nature of measurement uncertainty, and may help reduce categorical reasoning about uncertainty in users, while increasing accuracy.
“In this respect, visualizations that place less emphasis on confidence range limits might better represent the continuous nature of measurement uncertainty, and may help reduce categorical reasoning about uncertainty in users, while increasing accuracy (Correll & Gleicher, 2014; Helske et al., 2021).”
From Visualizing Uncertainty to Promote Clinicians’ Understanding of Measurement Error (2023) · cites this paper as Correll & Gleicher, 2014
Error bars may overly emphasize the range within the bar.
“Error bars may also overly emphasize the range within the bar [24].”
From Uncertainty as a Form of Transparency: Measuring, Communicating, and Using Uncertainty (2021) · cites this paper as [24]
In bar graphs depicting average values, people judge data values that fall within the bar as being more likely to be part of the data set than points outside the bar, despite being equidistant from the mean.
“In bar graphs depicting average values, people judge data values that fall within the bar as being more likely to be part of the data set than points outside the bar, despite being equidistant from the mean [30, 6].”
From Illusion of Causality in Visualized Data (2020) · cites this paper as [6]
Bar encodings are visually vertically asymmetrical, with the area below the mean filled and the area above unfilled.
“Bar encodings, in contrast, are visually vertically asymmetrical, with the area below the mean filled and the area above unfilled.”
From Illusion of Causality in Visualized Data (2020) · cites this paper as [6]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
The citation lists haven’t been fetched yet
References 36 · Cited by 258
Research works printed but not linked: 1
- [8]Correll M and Gleicher M. Error bars considered harmful In IEEE Visualization Poster Proceedings. IEEE, 10 2013. PMC6214189 26356928
- Authors
- Correll, M., & Gleicher, M.
- Affiliations
- University of Wisconsin–Madison (US)
- Venue
- IEEE Transactions on Visualization and Computer Graphics
- Published
- 2014-12-31
- Volume
- 20
- Issue
- 12
- Pages
- 2142-2151
- ISSN
- 1077-2626 (ISSN-L)
- PMC
- PMC6214189
- OpenAlex
- W2069228960
- Cited by
- 258
- References
- 36
- Citations by year
- 2026: 10 · 2025: 17 · 2024: 25 · 2023: 38 · 2022: 32 · 2021: 27 · 2020: 27 · 2019: 17 · 2018: 23 · 2017: 17 · 2016: 16 · 2015: 8 · 2014: 1
- FWCI
- 12.52
- Type
- journal article
- Language
- en
- Open access
- green
- www.ncbi.nlm.nih.gov
- Locations
- IEEE Transactions on Visualization and Computer Graphics (publishedVersion)
- PubMed (publishedVersion)
- PubMed Central (acceptedVersion)
- Funders
- NSF (IIS-1162037)
- NIH (R01 AI077376)
- ERC Advanced Grant ¿Expressive¿
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Data Visualization and Analytics
- Social Sciences › Psychology › Experimental and Cognitive Psychology › Visual and Cognitive Learning Processes
- Physical Sciences › Mathematics › Statistics and Probability › Statistics Education and Methodologies
- Keywords
- error bars, bar charts, uncertainty visualization, confidence intervals, measurement error, violin plots, visual encoding, statistical inference, decision making under uncertainty, visual statistics, information visualization
- MeSH terms
- Adult, Computer Graphics, Female, Humans, Male, Task Performance and Analysis, Statistical Distributions, Empirical Research, Young Adult, Crowdsourcing
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
