Visualizing Data for Impact: Analyzing Misleading Visualizations

Data visualizations should be formatted with an effective chart type. The chart type should create clarity, not confusion. The visual format (bar, pie, line, scatter plot, etc.) should help convey the data insight.

In this example, a grouped bar chart effectively compares customer satisfaction ratings across categories. With this visual format, the ratings for “In-Flight Comfort” and “Booking Process” are next to each other, so we can quickly spot the differences. We can spot that most of the Booking Process ratings are 4s and 5s. By contrast, most of the In-Flight comfort ratings are 1s and 2s.

Data omission often results in misleading charts. Omitting data means leaving out key information, cherry-picking data points, or using a truncated scale. A chart must include enough data so viewers can draw an accurate conclusion.

For example, Best Deal Airlines made a line chart to demonstrate the change in flight crew salaries. It appears that salaries have steadily increased from 2015 to 2023.

However, essential data has been omitted! Flight crews consist of pilots and flight attendants. Unlike pilot salaries, flight attendant salaries have been stagnant for three years. Best Deal Airlines omitted flight attendant salaries to fabricate a more favorable picture of their payment practices.

Truncated graphs are generally misleading. Data points should not be left out to fit a desired narrative or present a skewed picture. As a general rule, axes should start at zero and numbers should not be skipped.

For example, Best Deal Airlines made a chart to capture ticket sale data. The line chart shows a dramatic increase in ticket sales, with 2024 showing a record high for the company.

Looking closely, we’ll notice that the graph is truncated: the first year on the x-axis is 2020. Due to travel restrictions during the Covid-19 pandemic, 2020 had extremely low sales. Thus, the years after 2020 look quite favorable by comparison. Truncating the scale exaggerates the success of sales in recent years.

Scale manipulation is a common tactic for creating misleading data visualizations. A graph will be inaccurate if the scale is inconsistent, truncated, or unsuitable for the context.

For example, Best Deal Airlines is publishing a report about changes in airplane seat width. The graph shows a subtle downward slope, implying only a slight decrease over time.

You may have noticed something odd with the scale: seat width is measured in feet. For this context, the scale should use smaller units like inches or centimeters. Relatively small changes to seat size make a big difference to the passenger. The seats have shrunk nearly six inches, which is a significant loss of space.

Appropriate color palettes should be applied to data visualizations. Color should be used consistently across multiple visualizations and meet accessibility standards for colorblindness. Color choices should align with the target audience’s color associations without reinforcing biases or judgments.

The example heatmap ignores a common convention for applying gradients. In general, we associate light tones with “less” and saturated tones with “more.”

The deep blue squares appear to be the busiest months for air travel. However, the legend reveals that the opposite is true! Because this heat map’s colors subvert our expectations, viewers may be confused or draw the wrong conclusions.

When adding context to data visualizations, consider what to say and how to say it. Use neutral language that is clear, unbiased, accurate, and accessible to your audience.

This graph from Best Deal Airlines shows customer comfort ratings based on ticket status and flight length.

Terms like “ruby” and “sapphire” are unconventional descriptions for airline ticket status. If this graph is for Best Deal’s employees, the labels are okay, because internal teams are likely familiar with these terms. However, if the chart is meant for a wider audience, the labels for ticket status may cause confusion. Additionally, the annotation uses the term “cheap seats” in a way that feels judgmental and unhelpful.

Graphical complexity occurs when too much information is crammed into one chart. Viewers may be unable to process the takeaways, so an overly complex chart is ineffective.

For example, Best Deal Airlines wants to compare its prices with top competitors. This graph includes 14 airlines that are organized alphabetically rather than by price. The graph also color codes the airline as budget, standard, or luxury. With so many details on one graph, we can’t easily observe how Best Deal’s prices compare with other companies.

While it’s not necessarily wrong to include so many airlines in one graph, it distracts from the insight we’re trying to convey. To compare Best Deal with its top competitors, we should focus on budget airlines. This will cut down on the details and make it easier to process the information.