Fact-checking data visualization: how to verify graphs, charts, and diagrams

In the previous part of this article, we discussed how manipulating infographics can be misleading. However, malicious intent is not always the reason for misinterpreting data. Let’s look at unintentional distortions and errors that can affect the perception of the information presented, as well as the main stages of verifying graphic information.

Accidental distortions and errors

1. Errors in design and visual coding. Even a graph that is accurate in terms of data can be difficult to understand due to inappropriate visualization, unaccounted-for color perception characteristics, and poor image quality.

Ineffective visual coding: One common mistake is using area or volume (e.g., circles in a diagram or pictograms) to convey linear values. Since the human eye is poor at estimating and comparing areas, this leads to significant distortion in perception. For example, an object that is only twice as large according to the data may appear four times larger visually.

Design overload: Infographics lose clarity due to excessive decorative elements that distract attention from the data itself and make it difficult to read. For example 3D effects, shadows, textures, bright backgrounds, and gradients.

2. Mistakes during the data collection phase

Unbiased sample for research. Research results will be unreliable if the sample is compiled incorrectly — for example, if it includes only volunteers, is limited to one social group, or is too small to be statistically significant. For example, if 10 people participate in a study of the effectiveness of a cream and one of them has a positive result, this is considered to be 10%.

Incorrectly worded questions. Questions that lead the respondent to a specific answer. Example: “Do you agree that the government should combat the appalling level of inflation?” (The word ‘appalling’ predetermines the answer). Opposite: “How would you rate the current level of inflation?”

3. Errors in the measurement and selection of indicators

Incorrectly selected indicators. Example: Measuring a doctor’s performance based on the number of patients seen rather than the quality of treatment and recovery. This may motivate them to work faster, but not better.
Incorrect units of measurement or aggregation. Example: Comparing countries by total CO₂ emissions rather than per capita emissions. China will be the leader, but for an individual US citizen, the figure will be many times higher.

Infographics quality check

Check the quality of infographics by answering the following questions:

Verification of source and context

— Is the original source of the data (research, institute, database) clearly indicated?

— Is the source authoritative and unbiased? Or is it, for example, a report by a commercial company advertising its product?

— Is it easy to find and verify the original dataset by link or title?

— Where is the infographic published (scientific journal, news site, and advertising brochure)? This determines the strictness of the verification.

— What is its purpose — to inform, persuade, or sell? The presence of commercial or political interests requires extra caution.

— Is the author/designer credited? This adds accountability.

— Is the date of creation indicated? The information may have been current at the time of creation but may be outdated by the time of publication.

Data integrity check

— Does the headline or main thesis exaggerate what the data shows? (For example, “sharp increase” actually amounts to 2%).

— Do the captions under the graphs correspond to the data itself?

— Compare the figures given in the infographic with those in the original document. Are there any discrepancies in absolute values or percentages?

— What data is the infographic based on? If it is a survey, how large and representative was the sample? (For example, “80% satisfied” out of 10 respondents is not an indicator).

— What period does the data cover? Are data from different periods being compared (e.g., monthly sales with annual sales)?

— Are the selected indicators appropriate for the specific measurement?

— Can other researchers verify your conclusions based on the information presented?

Visual design verification

— Does the Y-axis start at zero? If not, does cropping the scale unjustifiably exaggerate the difference between values?

— Are the proportions between the X and Y axes distorted to make the trend steeper or flatter?

— If several graphs are provided for comparison, do they have the same scale?

— Are all elements of the graph explained in the legend? Is the legend clear and unambiguous?

— Is the selected graph type appropriate for this data? Check that a logarithmic scale is not being used without warning. It changes the perception of growth rate. Also, remember that the area of a circle increases quadratically with the radius, which deceives the eye (an object that is twice as large according to the data will appear four times larger).

Data visualization is not just for show; it is a tool for decision-making. Therefore, the main feature of conscientious visualization is its transparency and accurate representation of the information being displayed. Remember: high-quality infographics are not afraid of critical scrutiny.