The editors and staff of The Journal of Physiology take great pride in the success of our journal. We strive to fulfill our mission to effectively communicate important advances in physiological principles in health and disease, and honour the esteem our journal has gained with the scientific community and general public. One of the keys to our success lies in effective scholarly communication in the articles published in our journal. We thank authors who strive to utilize best practices in communication skills. We are here to facilitate this process for our authors.
No doubt, grammar and linguistics lie at the heart of effective communication in journal articles. I direct a scientific writing course for graduate students at my institution where I stress the three C's of writing good research articles – to be clear, concise, and complete. Although it is obvious that this rule applies to writing, it equally applies to effective communication through illustrations. Indeed, we are all familiar with the adage ‘a picture is worth a thousand words’.
Effective illustrations are vital to good research. They must communicate data in a clear, concise, and complete manner. Generally, investigators use illustrations of data for two purposes: first, to visualize and explore correlations and predictions among data sets; and second, to illustrate interpretations of these data clearly and concisely to an audience. Recent advances in obtaining extremely large sets of data (e.g. with omics approaches), has highlighted the need for effective visualization of complex data. Many options exist to visualize underlying correlations and predictions in large data sets. In fact, it has become a major corporate industry. Illustrations such as heat (colour) maps, circular designs, data clusters, tree maps, etc. are examples of visualizing data to reveal associations. Often, however, these approaches lack clarity and conciseness, and attention must be given to whether they provide the proper focus in research articles.
In summative research articles, illustrations primarily serve to assist the reader in visualizing interpretation of data after they have been collated and statistically analysed. This second form of visualization generally requires different strategies from those for visualizing primary data sets. Statistical comparisons are usually linear (cause–effect), which explains the typical use of bar and line graphs.
Regardless of the purpose for visualizing data, the three C's of good composition need to apply to illustrations in a research article. When designing graphs, clarity is achieved by attention to details, such as the appropriate labelling of axes (with correct units of measure), proper and consistent scaling of axes, and clear differentiation of data among groups. The use of proper line width and font size also impacts clarity. Of major importance, clarity is largely dependent upon proper focus in the illustration. Extraneous information should be eliminated or held to a minimum so that main points can be easily seen visually.
We recommend effective use of colour to differentiate data sets within graphs for clarity and focus. Our journal facilitates this recommendation by not charging for use of colour in illustrations in the online version. However, we stress two important points when using colour in illustrations: first, use distinct colours that stand out on a white background and vary in tone or intensity so that individuals with colour‐impaired sight can see the differences; and second, use the same colours and annotations consistently through all illustrations.
Concise illustrations are most effective. Illustrations and graphs with several unrelated panels distract from focus. Keep the message simple. Our journal helps with this goal by imposing no limit on the number of illustrations in our research articles, eliminating the need for multiple unrelated panels in illustrations.
Illustrations need to be complete. Although extraneous information is distracting, incomplete information is disastrous. Axes must be labelled. Micrographs must contain a dimension scale. Gels and images must be properly annotated. Westerns must contain appropriate control lanes and a molecular weight scale. Lastly, images must not be retouched to enhance or distort data.
Graphs must also accurately and clearly visualize data analytics. We challenge the traditional and misleading use of standard error rather than the more transparent standard deviation bars for expressing variance in data points. Whenever possible, it is also a best practice to illustrate individual data points (scatter plot) in addition to a mean or median and standard deviation.
I provide an example of the impact of these suggestions in Fig. 1. The figure illustrates hypothetical data of mRNA levels for a protein (Prtn) across genotypes of XYZ in male and female mice. Panel A illustrates the data in a format that is unclear, distorted, and incomplete. The y‐axis is not labelled (mRNA expression level). Lettering is not proportionally scaled to the graph, lacking consistent use of font type and size. The groups are not clearly defined or easily differentiated by shade or tone. There are no visual clues of sample size, variance or statistical comparisons. The three‐dimensional cones are distracting and provide a misleading sense of the magnitudes of effects. Importantly the grouping of data masks the important comparison between mRNA expression in males and females with a +/− XYZ genotype, and suggests a false sense of a difference in mRNA expression between males and females with a +/+ genotype. Panel B illustrates the same data in a much more clear, concise and complete manner. Note also the importance of providing a clear description of the data in the figure legend.
Figure 1. Examples of bad (panel A) and good (panel B) illustrations of data.
The graphs illustrate hypothetical normalized mRNA levels of a protein (Prtn) in tissue from 10 male and female mice with differing XYZ genotypes (+ wild‐type allele; − mutant allele). Panel B, individual data points (circles) with means ± SD (horizontal lines). *mRNA levels were decreased in females compared to males with +/− XYZ genotype (P < 0.001, two‐way ANOVA with Bonferroni post hoc test). mRNA levels were decreased in the −/− XYZ genotype for both males and females compared to the other genotypes (P < 0.05, Bonferroni post hoc test).
In summary, authors are encouraged to apply the principles of good composition to their illustrations as well as to their writing – clear, concise, and complete. Attention to these details will help The Journal of Physiology maintain its reputation of excellent scholarship.
For more details on use of illustrations in our articles, please see:
https://jp.msubmit.net/cgi-bin/main.plex?form_type=display_requirements#figures
https://jp.msubmit.net/cgi-bin/main.plex?form_type=display_requirements#statistics
Authors of accepted papers are also encouraged to submit illustrations for consideration for the cover page of the edition.
Edited by: Kim Barrett