Representing Error bars in within-subject designs in typical software packages

The addition of error bars in graphs can greatly assist readers in determining where differences lie between groups or conditions. The use of error bars to represent the standard error of the mean or confidence intervals is relatively straightforward for between-subject (independent group) designs. However, confusion still abounds as to how best to graph these same measures for within-subject (repeated-measures) designs. This paper serves both to consolidate previous recommendations regarding error bars for within-subject studies and to provide clear instructions as to how to implement these recommendations using typical statistical packages.