Graphing Within-Subjects Effects

Abstract

Most graphs in psychology articles fail to show distributional information other than the mean. Although many articles have suggested solutions to this problem for between-subjects designs, there has been relatively little discussion on how to show distributional information for within-subjects designs. Graphs of within-subjects data should be constructed so that between-subjects variation does not appear as uncontrolled error. This article presents a variety of methods for graphing data from within-subjects designs including jittered dot plots of difference scores, sum-difference plots, box plots of difference scores, and plots of trend components. These graph types show descriptive distributional information while controlling for between-subjects variation.

Keywords

• graphing data
• within-subject data
• jittered plots
• sum-different plots; box plots
• plots of trends.