I am wondering if I should use a bar chart to present the results of a one-way anova with three conditions? Or should the results rather just be described in the text? The field is social psychology (APA style).
I would reccomend a line graph, so that you can display any interaction effects.
Hi Samantha,
thanks for your answer. Sorry I fear the question was unclear, so I edited it a bit.
I am refering to a very simple design, that just compares the means of three different groups (so no interactions). It would be possible to do this in a text or via a bar chart with three different bars (see the attached figure as an example). I am wondering if a bar char would be appropriate here, or if this design is so simple, that it usually does not need a figure?
Figure from:
Koulmanda, M., Bhasin, M., Fan, Z., Hanidziar, D., Goel, N., Putheti, P., ... & Strom, T. B. (2012). Alpha 1-antitrypsin reduces inflammation and enhances mouse pancreatic islet transplant survival. Proceedings of the National Academy of Sciences, 109(38), 15443-15448.
If you want to compares the means of three different groups, It would bebetter to do this via a bar chart with three different bars,becasuse there is no interactions,just like your example.
Thank you everybody for your comments. According to the literature that I checked, it is rather uncommon to present it as an actual bar chart. So I will simply describe it in the text.
Corrected: I left out the square root in the original post which was late at night. "Do consider confidence intervals using the + or - 2*[sqrt( mean squared error )] as a 95% CI. But, do not become involved in the overlap-non overlap question (Mathematical statisticians will say it is not validly related to significance, and it has pitfalls in application)." [Square brackets around the change ]
However: when sleeping upon it I realized I do not know how many degrees of freedom are associated with the MSE, if they are less than 60, you should use Student's t instead of 2. If they are less than 20 the ANOVA will be hyper sensitive to any non-normality present and you should use a Friedman's test instead of an ANOVA, Then a box plot for each treatment would serve nicely for a graphical presentation.
- Badges
- Science topic
- Similar topics
- Psychology
- Social Psychology