Merging Graphics and Text to Better Convey Experimental Results: Designing an “Enhanced Bar Graph”
by William D. Berry, Florida State University, and Matthew Hauenstein, Florida State University
Over the last two decades, randomized experiments have become much more common in political science, magnifying the importance of clear communication of the statistical results of experiments—and their interpretation as treatment effects—to readers. We propose a format for presenting experimental results that combines a graph’s strength in facilitating general pattern recognition with a table’s strength in displaying numerical results. The format supplements a conventional bar graph with additional text labels and graphics. The resulting enhanced bar graph relies on graphics to convey general patterns about treatment effects, displays point estimates and confidence intervals for all key quantities of interest relevant to testing hypotheses (e.g., first differences in the mean of the dependent variable), and uses text labels to clarify how these quantities can be interpreted as estimated treatment effects. The last feature is especially valuable to readers with limited training in quantitative methods. Presenting information in a single figure avoids the need to devote scarce journal space to both a graph and a table. Moreover, an enhanced bar graph prevents readers from having to move back and forth between a graph and a table of numerical results—reducing their cognitive load, and facilitating their understanding of the findings.
PS: Political Science and Politics / Volume 50, Issue 3 / July 2017, pp. 831-836