We’re now at the stage where it’s time to start to think about turning the work I’m doing into a scientific paper, so I thought I would talk about how I tend to write papers. A lot of people have written about this before, and I highly recommend Sophie’s advice (she's actually a journal editor).
Step 1. Have some results
This step perhaps goes without saying, but before you start to write a paper you need to have something to write about!
Some people start with a brilliant idea for a paper that just hasn’t been written yet. For me, it usually starts with “I wonder…”. In this case, we knew that different reanalyses have different representations of East Coast Lows, so we wondered whether we could use satellite rainfall or wind data to assess which one is the best.
So that’s followed with a lot of writing code to analyse data, make a mountain of figures, and look at a whole bunch of different numbers. And at some point, you realise that some of what you’ve found is interesting, and is worth sharing with the world.
Step 2. Decide what the story is
The rookie mistake I made when I was writing my first paper, and that I think a lot of other young scientists do too, is wanting to show all your work. You’ve done so much work, and there are so many little things that are interesting, and you want to write about it all!
Unfortunately, this usually ends up with a paper that’s a bloated mess - paragraphs full of numbers, page after page of repetitive tables or figures. When it’s hard to understand what you’re trying to say, readers get really bored.
What you want is a tight, short paper that tells a single story, with only as many figures and as much detail as you need to convey the story. This means figuring out what your most important results are, and what you need to say in order to make your results clear.
In theory, this should also be the time when you decide what journal to target, because that helps you know how you should structure it. Although sometimes I don’t actually do this til step 4, oops.
Step 3. Write an outline
This isn’t for everyone, but I love writing outlines.
I start with the major sections - the different parts of the results/story, and what order I think they should go in. Then I fill it in with a few dot points about what each section should cover, and start putting in the main figures (these are not in a final form, just to start the planning.
From there, I will send the outline around to the co-authors to get their opinions. That way we can make sure everyone’s on the same page and discuss what results are unnecessary and what extra analyses to do, before wasting time writing something that may need to be totally rewritten later.
(As for who to have as coauthors? I tend to be on the generous side of things, because I’d rather have more people, and more eyes to find mistakes, than go fewer and take the chance of insulting someone. But everyone’s equation is different)
Step 4. Start filling in the details
By this point, my outline is usually like 5 pages long, from all the figures and dot points and tables. So the next step is to start changing those dot points into paragraphs. I can’t really give much guidance on how to do that part, I’m afraid. Once I have the scaffolding of the outline, expanding dot points into paragraphs is relatively straightforward for me.
I’ll typically write the results first, then the methods and introduction, then the conclusions, and leave the abstract to the very end. Then I’ll print the whole thing out and edit it with a pen, to really find all those places where the writing doesn’t make sense, before sending a draft around to coauthors. And iterate until everyone’s happy.
Step 5. Clean everything up
This is more important than you think it is. Do your spell checks, and check that the reference list is all consistent. Make sure the paper is formatted in the journal’s style, and that it’s not too long or with too many figures.
Go through the figures and make sure that they’re clear (and any text is readable), that they have labels if they need to, that they don’t use the rainbow colour scheme, and that they convey the story you’re trying to tell easily. For bonus points, it can be good to make sure as many as possible look okay if printed in black & white.
If you’re anything like me, and your code when you’re analysing data is a bit all over the place, now is a good time to clean up your code and make sure the bits you need to actually produce the figures and results in the paper are in one place and hopefully clear & commented enough that when you come back to the paper you’ll easily be able to follow & recreate what you did. This is more and more important with the current movement towards transparent & reproducible scientific code.
Step 6. Submit... and wait
Usually it takes 2-3 months for a paper you’ve submitted to come back. Most of the time you’ll get asked to do either minor or major revisions. This usually means explaining things reviewers didn’t understand and doing more analyses to make the paper stronger (or make sure your results say what you think they do.
I’m not going to go into how to respond to comments, but Sophie has some great tips. Just remember to thank the reviewers, because they usually make the paper stronger. And remember to review other papers & do your best to be both kind and rigorous, because otherwise the system crumbles.
And sometimes you’ll get a reject and resubmit, or just a flat reject. This can be disheartening - after all, you put so much work into this! But about 30% of papers are rejected, and it happens to all of us - I had a paper rejected just a couple of months ago. But usually, this just means you need to rethink how you designed your paper (or where you submitted it), and give it another go. Re-working that rejected paper is on my to-do list for April, now that I’ve submitted my responses to my thesis examiners.
Step 7. Celebrate!
If all works well, after a couple of rounds of review, your paper will be published, and you’ll have a moment of excitement… before starting to work on the next one. :)