Always Write the Results First

DrugMonkey has an interesting discussion going about whether or not to “Cite While You Write“.  And yes, I am a big enough nerd that I find this type of discussion “interesting.”

On a bit of a different tangent, the best paper-writing advice I ever received was to stop thinking of a manuscript as a linear document with a beginning, middle, and an end.  Instead, create the tables and figures first and organize them until you can use them to verbally tell the story. Then outline the key points of the story and organize the critical citations that support the story.   Then write the results and the discussion.  Then the methods.  Then write the introduction and abstract last.  When you write the introduction first, you lock yourself into a narrative that may not be the best fit for your data.  I suspect this is how people end up writing papers where they have some data that sneaks into the results, but was never mentioned as part of any hypothesis or methodology.

Break yourself of linear manuscript writing.

Also, always include page numbers.  I don’t think I realized the crucialness of this move until I first watched a mentor review a paper by printing it out, separating out different sections, and spreading it all over his desk. 

On an even more unrelated note, how did I forget how funny Triumph the Insult Comic Dog was?

19 responses to “Always Write the Results First

  1. All of the papers that I have ever written begin with the figures and legends. It’s the best way to start to figure out what holes there are to fill, and how in the heck to finish it up. After the figures, I start with results, intro/discussion, methods and very very last: abstract. Even with grants, I start with the Specific Aims page, then start on the Research Strategy, but I always have that SA page open, because it will change as the RS evolves……

  2. Personally, I always start with the methods, because that’s the bit I’m most confident of. This is what I did and how I did it. then the figures, then the results, then watch my PI argue about the discussion, introduction and abstract (often substantially changing the percieved outcome of the paper from one revision to the next). Then go back to the methods and make sure I’m talking about only the things that lead to the results I’m going to show.

    Granted, this hasn’t ever gotten me a paper published, but that’s to be expected when the PI runs away with all the data. Writing the method just proved I was actually working.

  3. In my opinion, the only possible way to write an abstract is after you have finished the paper, because it needs to be a concise summary and you can’t do that until you have something to summarise.

    And next-last must be the Intro for the same reason, and so that you don’t go on and on with stuff that (ends up being) not relevant to that particular set of results, though you thought it was interesting at first. (Or, if you need to write the intro first for your own purposes, ie to overcome the inertia of getting started, then at least give it a severe edit after you have written everything else).

    So many times when i have reviewed papers, I have suggested that half the intro be deleted, as it waffles on in an entirely different direction from the main point of the paper.

    Sometimes abstracts do that too – don’t encapsulate what the conclusions really are.

    And how many times have you skimmed something that is only of mild interest to you, and ALL you have done is look at the figures, tables and conclusions? The figs and tables therefore should tell their own story, with adequate captions to enable this to happen for those not interested in the nitty gritty of the methodology.

    After that rave? Yes, just adding substance to Isis’ initial thoughts. I agree.


  4. PI insists we give him the *abstract* first, which continues to stump me. I can’t conceive of writing a nice summary blurb before I even know what my Aims are. When I was writing my NRSA, he changed the Aims after I’d written up the Research Strategy, with less than a week before deadline.

  5. also, i’ve been meaning to say that I’ll take your walking around at 4cm over my 20 hrs of pitocin and 1cm any day!

  6. I don’t even read papers linearly. Most scientists are not very good writers, so even if the story is compelling, the way they write it usually isn’t.

    I assume reviewers aren’t going to read my papers linearly either, so I try to make it clear what the story is—or at the very least how to go back and pick up the narrative—no matter where someone drops in.

  7. Funny, my PI always told me to make figures and the story would flow. I truly saw this happening when I was writing my thesis. It was much easier to do a literature search and organize things around the figures and tables once they were in some sort of shape. Great advice!

  8. Pretty much the same in my lab. We usually start the methods while we are doing them and they are fresh in our minds. But the results are the foundation of the paper – all the figures and captions are pretty much done from the beginning, although sometimes we move forward while doing some additional experiments/analyses that will improve the results. Sometimes our PI will change his mind on some detail of a figure after the paper is mostly done – usually color/style, axis range, etc. I would have a lot of trouble writing anything for a paper without the results ready to go. How do you know what story you have to tell without figures done?

  9. I used to collaborate with a lab that always had great lab meetings because whenever anyone wanted to write a paper, they’d make up the figures, present pretty much only the figures to the group, and everyone would discuss the logical flow, what else was needed, etc. It was a fun way to get feedback, before you’d written the paper and gotten all attached to the text.

  10. Write it the way people are going to read it. Figures first IS linear…

  11. Thanks for the advice, Dr. Isis! That’s brilliant!

  12. Pingback: Sewing together a manuscript « The Tightrope

  13. Figures and tables first. I’m on the last legs of an MS thesis, and currently I have 40 pages of text (not including references) and 45 figures. The text talks about the figures. Once I figured that out, it was a breeze to write the text. Editing the text has taken longer, since my advisor is King Of The Blue Pen… and I write mostly by ear. (Yes, I’ve read the Chicago Manual of Style; it just hasn’t taken hold.) I have a good ear, but not a Scientific Paper ear. But figures -> tables -> methods -> results -> interpretation (of my own results) -> discussion (of my results relating to others’ work) -> conclusions -> introduction -> abstract has worked for me.

    In my first graduate class I was taught how to read scientific papers: abstract, figures, conclusions. Want more detail? Read the rest. So, make your figures as clear as they can possibly be, and spend TIME making the abstract as definite and straightforward a summary as it can possibly be. Your reading public will appreciate you.

    Now, off for more work on my $%#& abstract, which has to fit, double-spaced, on a single page…

  14. Isis the Scientist

    You have 45 figures? In a masters thesis????

  15. Well, I’m in the process of writing my Introduction before all of the analyses are done, but that’s because we’ve got figures that show more or less what the results are going to be, and I’m writing the Intro while my collaborator does the analyses. I always start with methods since they are straightforward. I sometimes have an abstract already, because I’ve submitted early results to a conference before all of the data is collected, but of course that has to get much more specific once the paper is completely done. Also, depending on the number and size of the experiment(s), there may only be one or two figures, and there is a lot more to my analyses than just those figures. Often the Introduction is pretty easy to shape because it follows from the discussion of my previous paper, and some of the elements will be the same no matter what the results show.

  16. “You have 45 figures? In a masters thesis????”
    It’s geology. Maps. Lots of analysis diagrams. More maps. More analysis diagrams. Extrapolations of what my study area looked like 800,000 and 400,000 years ago. Personally, I thought that “read the freakin’ figures” would have been a sufficient thesis text. Alas, my advisor disagreed with me.

  17. SadGradStudent

    This method makes so much sense. My PI also believes in laying out figures in order to know what the story is, but is way too fond of assuming that certain data will support the hypothesis and then using purely theoretical expected results as a placeholder figure while the text is written. Makes me crazy. How am I supposed to write up the results (and then the discussion and intro!) before all the experiments have been done? (The brilliant idea behind this is that we don’t want writing the paper to “hold us back” once the experiments are done; we should be ready to drop the data in and send it out.) Ugh.

  18. I haven’t written any papers of the same nature as you have mentioned, although I production-edited a colloquium proceedings with such papers, where few of the presenters followed instructions sent them regarding copy that could be cleanly copied into the production master, with an electronic copy, and copies of any photographs. I swear, most of the first two months’ work was composed of hounding authors and trying to get obsolete WP programs’ texts off floppies of varying sizes–3.5″, 5″, and 8″! I had to learn Mathematica in order to reproduce equations that were out of my league, but at least there were scientists on the same floor who could read them when I found something ambiguous. I actually had a blast.

    However, in high school and college, they taught that you were supposed to be linear in writing a paper, and I was not terribly linear even in those days. I did it my own way, after laying out an *initial* outline, and re-arranging things as necessary, producing the final outline/TOC at the end, as any reasonable person would do. Mind you I didn’t even have a Selectric–just an old manual typewriter, and no white-out or correction tape…and I have never been a fast typist without errors.

    I took to WP programs so fast my head spinned–where had they been all my life? Alas, in the brains of the inventors of personal computers of any stripe.

  19. Write the Results First was some of the best advice I received as well. I’ve been trying to find a resource for grad students that explains this non-linear approach to writing the sections of a manuscript. Are you aware of a publication that has addressed this writing sequence?

What do you have to say about that?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s