Tag Archives: scientific

Please do not try to tell a story but report your results

When I prepared a lecture last year on scientific paper writing I have found countless advices how to tell a story – it made it even into elife.

Don’t do that – there are lies and damned lies (Disraeli) while you are easily running into a trap when trying to “tell a story”.  Preregister your study plan, tell the world what you did right from the beginning, what did not work, why you repeated an experiment or why changed your opinion.

Writing a story from the backend distorts the proportions and misdirects attention. Ulrich Dirnagl highlighted this problem in an earlier talk here in March using the following two slides.

How we present results in a basically linear way – story telling. By NIH QUEST/Dirnagl.
How it would be fair to tell the story with numerous inputs, non working assay (primer, antibody, other samples), new literature and conference talks, where we constantly change our opinion that could have led to alternate papers. By NIH QUEST/Dirnagl

 

Better figures

A recent paper identifies 10 rules for better pictures. As I have also given several lectures on that topic, I was excited what the authors think…
1. Know your audience. This is trivial as you never know your audience.
2. Identify your message. True and not true at the same time. True as it makes your findings more evident – not true if you are allowing a reader to find his own message.
3. Adapt the figure to the support medium. Trivial. May be very time consuming.
4. Captions are not optional. Absolutely true, I also suppport good captions – mini stories for those who can’t read the whole text.
5. Do not trust the defaults. Trivial. No one does.
6. Use color efficiently.  Not really,  avoid colors for those of us who are colorblind and to avoid expensive page charges.
7. Do not mislead the reader. Why should I?
8.  Avoid Chartjunk. Absolutely. Most frequent problem.
9. Message trumps beauty. Sure, form follows function.
10. Get the right tool. Maybe correct while the further recommendations look like a poor man’s effort to make his first graphic at zero cost: Gimp, Imagemagick, R…