Be sure to check out the PLOS Biology article I mentioned. What do you think of their arguments? journals.plos.org/plosbiology/article?id=10.1371%2Fjournal.pbio.1002128
Great video, thanks! Recently I have been thinking about how to overcome this issue with barplots in my own figures - currently I am making boxplots with geom_point overlaid to better demonstrate the distribution of the samples, but when I have figures with a lot of data points it can look a bit messy and harder to interpret. Looking forward to a future episode on bar plot alternatives!
Hey Grace - I agree. It isn’t easy and there’s probably no perfect choice for larger datasets ... but I think most are better than bar charts with +/- se 😂
Dude, most of my master's degree has been online due to Covid. You're helping me a great deal with these tutorials and your posts on the Mothur forum! Stoked to see what you recommend instead. Maybe errorplots?
Thanks Pat. I just think that SE is not the right thing to show in dinamite plots (that already aren't a great plot for means). SE is the standard deviation of the means of samples taken form the same population, so, it is not appropriate to show the variation for 1 single (or pooled) sample. Don't you agree?
Totally agree with you, I do not like bar plot either. But the real problem is the original data they use, very less chance to get them even by a kind ask with an email. Open science and reproducible research is still a long way to go.