Monday, October 5, 2015

Data Presentation

The editors at AJP Lung have been thinking a lot about data presentation.  Many scientists present their results by showing bar graphs that plot the mean plus or minus the standard error of the mean (mean ± SEM).  The advantage of this approach is that bar graphs are neat and easy to read, but it is also appreciated that this method of presenting data can mask some important attributes and potentially may be misinterpreted.  Recently, there was a very nice article by Harvey Motulsky, from Graphpad software, the maker of a commonly used data analysis program that highlights some important misconceptions about data analysis.  The article is available as open access (Pharmacol Res Perspect. 2015 Feb;3(1):e00093) and the link is provided here: Motulsky article .  It is especially interesting to consider figure 5 of the paper, which shows the same data set graphed 6 ways.  Looking at this figure, it can be appreciated that plotting mean ± SEM gives little insight into the variability in the data set.  It also may mask different sample sizes in some cases.  Some journals have editorial policies to insist that data be presented as scatter plots.  While AJP Lung has not decided on such an editorial policy at this time, we do want to encourage authors to think about how they present their data and to do so in the most transparent manner possible.  In the meantime, Happy Fall!

No comments:

Post a Comment