Customer Service

bogus data visualizationsYou’ve heard of fake news. You may not have heard of fake data visualizations. How charts and graphs display information can mislead viewers, either intentionally or unintentionally. The display formats can skew data and, as a result, support a particular point of view or agenda.

For most business professionals, charts are commonplace in the daily work flow. Every day, we in public relations and marketing view charts on the economy, the industry, and the company. We view charts from vendors, pitching products and services. We frequently create charts and graphs for corporate superiors, employees, media outlets, customers and the general public. These days news releases and marketing materials without charts and graphs just don’t get much attention. Media monitoring and measurement tools for PR and marketing analyze media mentions and other information in data visualizations.

“There’s nothing new about using statistics to draw a conclusion the data doesn’t support, but data visualization is making it harder to spot these lies,” says Shel Holtz of Holtz Communication + Technology. “We need to learn to identify whether a graph is accurate.”

PR and marketing, especially, must be careful that the graphs they create do not misinform, misinterpret or mislead.

How to Spot Bogus Data Visualizations

Nathan Yau, a statistician writing for Flowing Data, explains how to spot misleading data visualizations. Here are some of them.

data visualization lies

Image souce: Flowing Data

Truncated axis. Truncating the axis in bar charts to make the bar length shorter dramatizes differences in values. The value axis of bar charts should start at zero.

Dual axis. By using dual axes, the magnitude can shrink or expand for each metric. This is typically done to imply correlation and causation.

Totals greater than 100. Pie charts and other charts that show parts of a whole add up to more than 100 are implausible. Pie charts represent 100 percent of something.

Limed timeframes. Cherry-picking dates can show outrageous increases or decreases and support favored viewpoints. Long-term views provide more truthful insights.

Broad binning. Grouping data into larger categories can over-simplify complex data and prompt misleading results. While broad binning can be useful, be wary of oversimplification.

Although these features don’t prove that a data visualization is bogus, they should set off alarms. A basic guideline: Scrutinize shocking and especially dramatic charts.

“Statistics are simply numbers – how we (choose to) interpret them is up to us mere mortals and the key mental tool of critical thinking,” says Thomas Maydon, head of Credit Solutions at Principa. “It’s time to think about thinking.” Maydon delves into logical fallacies, or mistakes in reasoning, in the Principa blog

Advice for Creating Data Visualizations

Here’s advice on for creating data visualizations that interpret and relay data accurately, are understandable, and get your point across.

Choose the appropriate chart. Using the appropriate chart type is an essential element of data visualization. For showing trends over time, line charts and bar charts are best. Bar charts are ideal for comparison and ranking because they include values on baseline, making it easy to compare values. Scatter plots are ideal for showing correlations between two factors – but remember that correlation does not prove a relationship.

Pick your numbers carefully. Specific numbers add credibility. In addition, odd numbers are more believable than even numbers, writes content expert Pat Friesen in 8 Tips on How to Make Your Copy More Engaged with Numbers. A statistic of 27%, for example, seems more exact than 30%. You can add a decimal point to make numbers appear more precise.

Label all chart elements. Clearly title the chart, label each axis and appropriately label each trend line or other chart component. The latest software for presentations and PR measurement enables pop-up labeling of chart elements.

Present data with context. For instance, compare numbers to a previous time period. If needed, call attention to the percentage change to highlight the comparative change. Tell viewers what the data means. Explain both what happened and why.

Follow best practices. Follow the industry best practices when using statistics, as published by the Public Relations Society and American Statistical Association. Be clear on how the information was collected and disclose who performed and paid for the work. If survey results are included, show the survey questions.

Bottom Line: Although data visualizations help viewers quickly understand complex numbers, they can lie just as words can. Charts and grafts can deceive viewers, deliberately or accidentally. Spotting data visualization cons and avoiding mistakes that mislead viewers when creating your own charts or graphs have become important skills for all marketing and PR professionals.