Don’t you just hate things that look really cool but simply don’t work? You know: the fancy all-in-one kitchen gizmo that cuts nothing but your fingers, or the software that’s supposed to “boost processing speed by 64%” and has made zero difference in your productivity. In such cases the novelty quickly vanishes, and you’re left feeling disappointed, annoyed, even cheated by the entire experience.

That’s how I view the radar chart that someone recently shared with me:

When the user clicks on a wedge of the chart, a pop-up box displays information about a medical practice’s performance on a single CMS Physician Quality Reporting System (PQRS) measure. Imagine the dismay of the person who seemed so enamored of this nifty little device when I said, “yes, it looks cool, but it doesn’t really work — and the novelty will wear off after the third wedge.”

You may think me just a grumpy killjoy of all apparently cool charts (and I’m okay with that), but there is logic behind my crankiness: this type of chart is basically useless. Here’s why:

• The viewer has to hover a pointing device above a certain area to see the name and all other information about the metric and its benchmark. As soon as the hovering device is released, the information disappears. Since humans can only hold so much in short-term memory, users soon lose their grasp on which wedge displayed what measure.
• As the number of measures increases, the chart becomes unwieldy; there is more and more to remember; and more information is lost.
• No scale or directly labeled results are permanently displayed, making it impossible to easily and accurately compare measure performance.
• The measures in a radar chart cannot be ranked.
• Inverted measures (where a lower value is better) cannot be correctly displayed on it.
• The chart does not lend itself to showing additional information about a metric, such as trends over time in alignment with the measure.
• A radar chart makes categorizing the measure by CMS Quality Domain (for example) difficult. Colors could be used to code different Quality Domains, but viewers would have to use pre-attentive processing to match the colors to a key (I’m already exhausted just thinking about it).
• On a dashboard, radars take up a lot of space, whereas display devices such as bullet charts and small multiples use far less real estate while delivering much more information.
• Finally, if you try to publish this to PDF or print it, you will get only the chart, with no associated information (and you know as well as I do that folks still want to see a PDF or a print).

Now look at the same data (and a whole lot more) displayed using a simple bar chart with figures, lines, and an indicator.

Click on image to expand

In this display we are able to:

• Label each display directly with the measure number and name.
• Include, in the measure number, hyperlinks to additional detailed reports about the practice’s performance on the measure.
• Directly compare and rank performance results.
• Show an inverted measure (lower performance is better).
• Use figures to show the difference between a practice’s performance and a national benchmark.
• Display the number of eligible events required to meet target.
• Display performance over time.
• Use an indicator (red dot) to draw attention to those measures that fall well below the national comparison (we can even encode them to display in a darker hue as the difference increases).
• Free the viewer from repeated mouse-clicks to reveal information.
• Save the view as a PDF that displays complete information and can be printed and|or sent.

Remember, just because a software application can create charts like the radar type doesn’t mean that we should be seduced into using them. Stick with the best practices of data visualization, which are based on research and evidence that clarify how people see, understand, retain, and use information most effectively. Your viewers will thank you-and so will I.