Lowering the bar (well, probably removing it completely…)

Are univariate scatterplots the key to successful data presentation in scientific writing?
Image: Wikimedia Commons.
Image: Wikimedia Commons.

I’m not noted for my contribution to debate upon statistical topics, but I do like to air matters of botanical education concern. Hence this item about data presentation (which is about as statistical as Mr P. Cuttings likes to get). And far from arcane relevance, this item is aimed at all of us who try to teach the next generation of plant scientists something about science writing, e.g. producing a scientific paper, particularly the not-unimportant matter of data presentation.

As Tracey Weissgerber et al. emphatically state, ‘Figures in scientific publications are critically important because they often show the data supporting key findings’. However, their investigation found that for those studies with small sample sizes (which are probably typical of many undergraduate investigations constrained as they often are in time, etc., since they are but one component amongst many in a botanical higher education setting), the prevalent data displays of bar charts and line graphs* do not help readers fully to understand the data.

To overcome this deficiency in what is arguably one of the most important aspects of science communication, they advocate presentation of univariate scatterplots (also known as ‘rugplots’) as a much more appropriate and alternative display for small-sample-size studies (and helpfully provide Excel templates to permit construction thereof – which must surely encourage take-up of their suggestions!). To improve this data presentation situation, the authors recommend three changes to existing practice: (1) encourage a more complete presentation of data [Who could argue with that? – Ed.]; (2) change journal policies [!! – Ed.]; and (3) train investigators in data presentation [i.e. not just data analysis – Ed.]. There isn’t the space here to go into the study in detail, but I can thoroughly recommend reading the full – open-access – paper (if only to share in the feel-good factor of being able to read – and understand – a ‘statistics’ paper). If you agree with Weissgerber et al.’s analysis and assessment, then go forth and influence your students.

 

* By which term I’m guessing they mean line charts, since a line graph is apparently a mathematical concept from graph theory…