Search This Blog

Friday, October 5, 2012

Writing a Journal Article. Part 3. Tables & Figures

Tables and figures attract the eye. Busy readers may get most of the detail they remember from your figures, so spend time thinking about the stories they tell. What you choose to show in your tables and figures is a statement about what you think is most important. And your tables and figures may well turn up in scientific presentations as slides.

The formatting requirements for tables varies among journals -- check what the instructions to authors say, and look at some recent published articles in the journal (preferably studies that are similar to your own) in order to get an idea of how readers and editors expect to see data presented, and what types of data belong in the paper.

Choosing figures is an opportunity for creativity. Think beyond your first choice. What can you include in the figure? What questions will readers have, and how can you make your figure answer those questions quickly and easily?

For example, in a paper I am currently working on, we are doing a case-control study. We are using the control group to generate regression equations to predict the expected score for our cases, and then comparing the expected scores with the observed scores.

So, the question is whether the difference between the observed and expected values is different from zero, and, if so, in which direction.

Our choices for figures include:
  1. Mean &  SD for residual (observed-expected) values. 
  2. Mean &  SD for residual values, by a mediating factor (in this case, sex).
  3. Dot plot versions of each of A and B, showing the spread of the individual points (because our sample is small, we have space to show all of the points). 
  4. Scatter plot of observed values (y) against expected values (x), with the x=y line drawn in, to illustrate the expectation of equality. 
  5. Scatter plot of observed values (y) against expected values (x), with x=y, and individual points coded to show details (for example, to show men and women separately). 
Choosing something like figure E will provide all of the data in A, but adds considerable information about the distribution of data, the difference between men and women, and whether the degree of deviation from expectation is different across the range of the expected values.

Finally, when you are writing your table heading (as always, following the convention of the particular journal you plan to submit to) and choosing the axis labels and legends for your figure, make it easy for your reader to understand you, and hard for them to come to common misunderstandings. Spend some time to wordsmith so that you are concise and clear rather than wordy.

This is Part 3 of a 3 part series of articles. I wrote it because I was about to write all of this to send to a junior colleague as feedback about her article, and realized that it would be of more general interest/use. If you find this useful and wish to share it, please credit me.

(c) 2012 Christine L. Hitchcock, PhD.

No comments:

Post a Comment