Tables > figures yet again

I received the following email from someone who would like to remain anonymous:

A journal editor made me change all my figures into tables. I complied, but I sent along one of your papers on the topic of figures versus tables. I got the following email in response which I thought you’d find funny:

Yes, statisticians prefer figures over tables. However, you are not writing this manuscript for statisticians. Your audience will be clinicians, nurses, epidemiologists and public health professionals.

The funny thing is, I think of biomedical journals (Jama, etc) as being pretty good about using graphs to convey their main results. They’re not as good as physicists, but they’re often better than statisticians!

6 thoughts on “Tables > figures yet again

  1. One advantage of tables (compared to graphs) is that it is pretty easy for automated methods to extract the data from tables. Graphs require more context, or human intervention, for data extraction. For example, we have a project (http://neuroelectro.org) in which we’ve extracted data on the electrophysiological properties of neurons from thousands of journal articles using text-mining tools. This was really only possible because a subset of authors (a shrinking subset, over time) put some of their data in tables, which were straightforward to machine-parse. The more people move to graphs (over tables), the harder it is to collect vast amounts of data across the literature, at least until algorithms get smart enough to handle the vast diversity and complexity of graphs without human supervision. Thus, tables are a win for the data sharing community.

  2. In an era of online publication, I do not understand this. I’ve always wanted to see both figures for insight, and tables for for checking or further analysis. It seems easy enough to have supplementary files online that have tables or spreadsheets. I’d rather have a spreadsheet I can just open than have to parse one converted to a table in the middle of text.

  3. Coming from the other side, I have really hated having a figure for what is probably 8-12 numbers without a meaningful spatial pattern. It’s even worse when the data are interesting and they have to tell me the numbers anyway so that I can use / discuss them!

  4. It’s a surprising response, but we haven’t seen the plots. I doubt ‘clinicians, nurses, epidemiologists and public health professionals’ would appreciate them any less than physiologists.

    Two great series from Advances in Physiology Education. Some are wonderfully illustrated (e.g. Correlation), with minimal fuss and fully functional code as Supplemental Material.

    http://advan.physiology.org/cgi/collection/explorations by Douglas Curran-Everett

    With well considered perspectives on potential interpretation and representation of statistics, (e.g. Presenting data: can you follow a recipe?):

    http://advan.physiology.org/cgi/collection/perspectives by Gordon B. Drummond et al

  5. I am involved in a biological research project in which non-parameteric regression is used to model some interesting phenomena and for which the figures give quite interpretable results. Some form of parametric regression could be used and the parameters printed in a table. But there is no biological model on which to base a parameterization, so to compare results for different variables differences in parameters have no meaning except for the effect they have on the shape of the curves. I think it would be irresponsible to put parameters without scientific meaning in a table.

Comments are closed.