Publications: Figures & Tables - patrick-shamberger/PHATE_Lab_Handbook GitHub Wiki

Figures

Figures are incredibly powerful. Done correctly, a figure can immediately communicate a result or an outcome; done incorrectly, a figure can obscure the meaning of the data or could even improperly convey a different meaning

There are too many nuances to go into in great detail here, so we will just highlight the most important / most common issues. For more detailed examples/guidelines, see the google drive. Finally, remember that your audience will judge you not just by the quality of your data, but also by the quality of its presentation.

  1. Use uniform fonts in all figures (generally a san serif font - arial/calibri/segoe UI/myriad).
  2. When the figure is real size (e.g., 3.5" wide for a single column figure, 7" wide for a 2-column figure), axis values should be 10 pt font, axis labels should be 11 pt font, and internal labels should be 10 to 11 pt font. Generally, do not use <10 pt font.
  3. Use appropriate and consistent color schemes. Use a color palette generator like coolors.co to establish a uniform color palette for a paper. Avoid red/green combos that look equivalent to some eyes. Finally, print a figure in greyscale and make sure you can still differentiate different points.
  4. Avoid unnecessary dazzle or glam. This is not your 3rd grade trapper keeper.
  5. Never delete / shift / adjust data points or images you don't like. Reasonable adjustment of contrast and brightness is ok for images to clarify the image, but not if it is used to suppress or artificially augment features of the data.
  6. Images should always include a scale bar.
  7. Whenever possible different sub-figures that use the same axes (e.g., temperature, time, etc.) should cover the same range at the same scale.

image

Tables

Tables should be cleanly organized, should include appropriate uncertainty statements, should include an appropriate number of significant figures, should include a table title at the top and any footnotes at the bottom, and should include references for any data that is not yours

  1. Table title should be short but exact, communicating exactly what the data represents
  2. All data included in a table needs to have an uncertainty associated with it (see separate section on uncertainty of measurements). In many cases, this may include both a measure of sample-to-sample variation (e.g., you made 5 different measurements of 5 different batches), generally reported as a 2sigma (2 standard deviations), as well as some estimate of accuracy which is generally done by measuring known reference materials using the same technique.
  3. Include units for all values.
  4. Do not include more significant digits than your overall accuracy allows. If the accuracy of your measurement is +/- 0.1 [units], do not report values with more than 1 significant digit. Wikipedia discussion of significant figures is spot on.
  5. Generally, journals handle formatting of tables. But when you prepare a table, make it clean with a minimal number of lines, and labels and values lined up appropriately

image