I have performed a mixed design ANOVA with time (pre-post) and group (treatment vs control) as variables. I'm wondering if it's sensible to use between subjects difference adjusted CIs (from Baguley, 2012) in the interaction plot below. What I want to emphasize is the group difference at post-test (and of course non-significant difference at pre-test). The error bars seem to convey the statistical significance of the time x treatment interaction, which was p = 0.0451. The error bars are calculated separately for pre- and posttest. Also, I'm thinking it'd be incorrect to interpret these errorbars as evidence that, say, treatment-pre is significantly different than control-post, is this correct?
Lastly, I'm aware that these error bars are not appropriate in interpreting differences within subject (and the figure caption will tell what the CIs represent). But is it sensible to use them like I've explained here, or is there some better way to graphically convey the significant group x time interaction?
Thanks in advance for any input on this matter!
Baguley, T. (2012). Calculating and graphing within-subject confidence intervals for ANOVA. Behavior Research Methods, 44, 158-175.
