Computers may be evil but they can help you find common flaws. Here are some things to do when you have writer’s block but still want to improve your text.
Checking your text
- Run the spell checker!
- If you want to write in active voice, search your document for the string “ing”. That string pops up a lot in passive voice. Look at the sentence “ing” is in and see if you could rewrite the sentence to be more active.
- To check for clarity, search for “it”. “this” and “that”. Look at the containing sentence; is it clear what “it” or “this” refers to?
- In scientific writing one often describes previous work by others and then one’s own work. Make certain that it is clear where the previous work stops and yours starts. “This” could be the previous or the present work. For example, “This paper” may need to change to “The present work …” to remove ambiguity with a discussion of previous results.
- To clean things up, look for strings like “In order”. Often “In order to … “ means the same as “To … “ and the “In order” is not needed.
- Even when using LaTeX you can export a pdf to word and use the Microsoft spelling and grammar checker to find errors. Just get used to telling it that intro- and -duction are words it can ignore.
- And don’t put blanks or special characters in file names to make certain you can compile, send and generally deal with them later.
Plots and tables:
- Figures need to have large enough labels on axes to be readable.
- Axes need to have labels with units! And a legend on the plot or the caption describing the content is a must.
- Generally, I suggest that captions fully describe (but not necessarily interpret) the plots. If someone just looks at the plot or table and the caption, they should be able to figure out what everything is. If your work is good, people are going to reuse your plots and you want them to do so correctly.
- Pro-tip – When comparing data with other groups, your data are the big filled circles and thick solid lines and the other group’s are the open triangles and thin dashed lines.
- Log scales can hide many sins. (So can linear scales If your data varies over orders of magnitude.) Use them wisely.
- Suppressed zeros in plots are the last refuge of scoundrels. (Except in ratio (not relevant) and log (impossible not to) plots.)
- Be aware that a significant proportion of the population is color-blind to some extent. Check out https://thenode.biologists.com/data-visualization-with-flying-colors/research/ for some suggestions.
Other Hints
- Use 12pt type so people can read it.