Improving the Visual Display of Data

​Dr. Kevin Little

Informing Ecological Design
Madison, Wisconsin, USA

 

Here is an all-too-common scenario, with which health care professionals around the world will be sadly familiar: you are in a professional meeting, at a lecture, attending grand rounds, or in any one of a number of forums, awaiting explanation of the results of an important study. The presenter might be armed with a PowerPoint presentation, slides, flipcharts, or overheads, all packed with data. Just one problem: virtually all of the visuals are a mess. Too many lines of text is a virtually ubiquitous sin, but worse still are mysterious graphics, with zillions of pieces of data that have no apparently interpretable results. What, you wonder to yourself, is the take-home message here?

 
The careful acquisition of solid data in measurements of quality improvement is a fine thing. And visual displays of these data are grasped faster and more easily than summary statistics or analysis. But even the most compelling results can be nullified by poor displays of data.
 
Dr. Kevin Little (Informing Ecological Design, LLC, Madison, Wisconsin, USA) suggests a method for improving the communication of data. He recommends thinking of graphical representations as tools to help tell a story. In improvement efforts, these stories are often complex, so it is all the more critical to tell the stories with clarity, precision, and efficiency. Little suggests a number of questions that will help determine the type of data display used:
  1. What do the data say (description)?
  2. What story are the data trying to tell (exploration)?
  3. How should the data be summarized (tabulation)?
  4. Can the data be used to intrigue or motivate (invitation)?
 
The best graphics show the most information with the least amount of clutter (ink, text, markings, colors). They also emphasize comparisons, relationships, change, and patterns. Effective displays do not distort the data, use three-dimensional displays except for three-dimensional data, use extraneous art or extraneous embellishments, change scales or symbols in the middle of a graph, or mask important changes in the data.
 
For improvement efforts, in which demonstrating results of interventions is crucial, the most important method of visual display is the “run chart”— a graph that shows changes over time (summary statistics can hide information on pattern and outliers) and is annotated to provide specific evidence or markers of improvement.
 
graph_work-up done on floor.jpg 
Another important graphical technique is the method of “small multiples,” a series of graphics of the same type that show the same combination of variables, but are indexed by changes in another variable.
 
image_small multiples schematic.jpg 
Shrinking a lot of data into a number of small graphs shown together creates a rich display of data that has a great deal of integrity to it. Such displays make changes in patterns much more understandable. Further, Little says this method lessens over-reliance on averages or central tendencies, and helps to point out more clearly the influence of particular subsets on overall patterns. He likes to quote the late statistician John W. Tukey, the pioneer of “robust analysis,” who said that the more you know what’s wrong with a number, the more useful it becomes. Robust displays provide context and “traceability” of data sources, allowing researchers to better assess the quality and limitations of the data.
 
Little is working to develop a visual spreadsheet tool, in which data cells are replaced by pictures, creating displays that go beyond the current limits of typical spreadsheets. With such developments, especially when they can be easily used by ordinary clinicians and others at their desktops, reports of improvements efforts, scientific studies, and a whole host of other venues for displaying data can only get better.
 
Further Reading:

 

Average Content Rating
(0 user)
Please login to rate or comment on this content.
User Comments