From Alexander Martin
Comments are most welcome!
A group of young researchers has started an initiative to improve scientific communication, focusing specifically on data visualization and the pitfalls of the ever-present bar plots. While there exist a wide array of ways to display data, many people continue to choose to use bar plots, a simple graph depicting a group mean and standard error (or deviation). Unfortunately, most data aren’t as clean as bar plots make them seem, and since bar plots reveal very little about the distribution of the data, this kind of visualization can be misleading. To finance a campaign aimed at journal editors, the group has created a Kickstarter : https://www.kickstarter.com/projects/1474588473/barbarplots/ proposing stickers and t-shirts and has created a hashtag #barbarplots so discussion can be followed on social media platforms. Join in the revolution and discussion about data visualization by supporting this initiative!
In a recent article, I have experimented with some (I hope) novel tricks to visualize data and significance: http://link.springer.com/content/pdf/10.1007%2Fs10849-013-9172-x.pdf (esp. Fig. 4 on p. 165 and Fig. 6 on p. 167). Feedback is welcome!
When I shared the above kickstarter link with colleagues in psycholinguistics, alongside the appreciation for a push towards greater statistical literacy, there was considerable resistance to the notion that one should *bar* barplots, and there were also some suggestions that they may be useful in some circumstances. It appears that similar concerns have been raised by others – there is a reply at the below link: