Degree In Sight

Graphs are often the last thing graduate students think about as they're writing their theses, dissertations or other research projects, says Penny M. Pexman, PhD, co-author of "Displaying Your Findings: A Practical Guide for Creating Figures, Posters, and Presentations" (APA, 2003).

That's a mistake, warns Pexman, associate head of the University of Calgary psychology department. "People look at your figures and charts and implicitly judge the quality of your research by the quality of your data presentation," she says.

And it's often the graphics that grab readers' attention, adds Pexman. "You have to expect that a reader may just go straight to your figures," she says. "These tend to be the things people focus on as they're discussing your work around the table during defenses, for example."

Pexman and other experts offer these tips on how to make striking graphs, charts and other visuals:

  • Begin with the visuals. Start writing your paper the way many senior researchers do—with the graphic elements. "First graph your data and then use that to help you think about and interpret the patterns," suggests Wendy A. Rogers, PhD, editor of the Journal of Experimental Psychology: Applied and a psychology professor at the Georgia Institute of Technology in Atlanta. "It enables you to see what the patterns really are."

  • Create your own. Students often rely on the default visuals automatically generated by their statistical software, but it's better to make your own, experts say. These programs often do things you wouldn't—or shouldn't—do. Potential problems include tiny font size, variations in scale from chart to chart and a reliance on keys that readers must consult rather than simply labeling the lines or bars.

  • Have a goal for each graph. No matter what kind of visual you're creating, your aim should be to make the data explicit and accessible. Research always begins with a question, says Stephen M. Kosslyn, PhD, author of "Graph Design for the Eye and Mind" (Oxford, 2006). Your graphs should help readers focus on the data that answer that question, he says. Only include data that are relevant to the point at hand, emphasizes Kosslyn, dean of social science and the John Lindsley professor of psychology at Harvard. Often, he says, students will include data that are interesting but irrelevant—such as differences between men and women.

  • Choose the right format. "Certain kinds of data invite certain kinds of display," notes Kosslyn. Pie charts are a good way to represent proportions of a whole, he says. Bar graphs are a good way to show point values. Line graphs are the best way to show trends or interactions. For further guidance, check out the "Publication Manual of the American Psychological Association" (APA, 2001).

  • Create clear pictures. Don't try to cram too much information into one graph, says Rogers. If you're using a pie chart, for instance, don't make the slices so small they can't be deciphered. The goal is to help readers make critical comparisons easily. "If you're comparing younger and older adults, put their data close together in the graphs," says Rogers. And make sure the graphic can stand on its own, adds Pexman. Readers shouldn't have to refer to a list of acronyms to figure out what the chart's about, for instance.

  • Label carefully. As you draft graphics, imagine someone seeing them for the first time. Maximize this opportunity to get across what the data are about, says Pexman. "Don't just say 'comprehension scores,'" she advises. "Say 'mean accuracy per trial.'" You may also want to indicate the standard error in both your bars and captions.

  • Be consistent. "A pet peeve for many readers is having different axes across different charts," explains Pexman. "Also make sure the scale of your axes is preserved across multiple figures." Data consistency not only clarifies your findings but also helps convince readers you're being upfront about what you found, she explains. Finally, now is not the time to get creative, says Kosslyn, emphasizing that you should follow research conventions as you create your visuals. Typically the y, or vertical, axis shows the dependent variable and the x, or horizontal, axis the independent variable, for example.

By Rebecca A. Clay

Rebecca A. Clay is a writer in Washington, D.C.