Wednesday, April 22, 2009

Writing a Results Section

I'm working on a paper, new and improved from the last rejection, and I've got lots and lots more shiny new data.

It's a lot, perhaps too much, and I don't know how to present it effectively. I'm having a really hard time.

My advisor advised me to work backwards - think about the points that I want to make and then make the graphs from there. But I don't have all the data finished yet, and it takes a long time to run them all. My fear is I will plan to make a point that won't actually be shown by any of my data (which is a bit irrational because I'm pretty sure what will come out, but that's just the way I work I guess).

How do you write your results section? How do you lay it out and construct it?


Mrs. CH said...

I have always just cranked through all the data before writing the results section. Granted, in astronomy it's probably a bit different because a) the data isn't as abundant and b) sometimes the results aren't obvious.

Maybe a good way to do it is go through some of the best data, see what you get, and then go from there?

PhysioProf said...

This is the order in which I write a paper:

(1) Figures.

(2) Figure legends.

(3) Abstract.

(4) Results.

(5) Introduction.

(6) Discussion.

Sometimes I'll do the intro before the results, but I always do the figs, legends, and abstract first. It is particularly helpful to do the abstract before the other sections of the main text, as that serves as an outline for all of the other sections.

PhizzleDizzle said...

MCH, I wonder if I could even parse an astronomy paper, and its results section. I should find one and try to read it. I bet it's super cool!! Any recommendations? Like, a paper that all first years should read or something?

CPP, that has unbelievably changed my day. I now feel like I have a grip on my shit. And my graphs look friggin awesome.

Science Bear said...

I'm currently having problems with that section as well and with my advisor at EB I'm glad you posted this!! I'm expected to have a rough draft to Dr Boss by tomorrow afternoon but have no one to ask for guidance.

Mrs. CH said...

I don't know about an article for first years, but here's one that my project is based on:

I'd have to think about it a little more to come up with some better options. I can always send you my paper too :)

Mrs. CH said...

hmm...let me put that in a link for you so it's easier to get too!

ScienceGirl said...

I start with figures and results, and finish with intro and abstract. I do think I tend to work backwards - from the interesting points to showing them on a plot, etc.

Psych Post Doc said...

Great question. I think it's so interesting to hear how other people do this.

I usually write up the methods prior to data collection, just because I've always written pretty detailed protocols.

I start writing the results section as I'm analyzing the data, with figures and tables.

Then I write the intro and discussion (usually going back and forth between the two) and finally the abstract. I find it fascinating that PP write the abstract so early on. I hate writing the abstract and absolutely can't do it until the rest of the paper is written.

microbiologist xx said...

My order:
Figures (I will even draw them if I haven't made them or finished the work).
Figure legends
Materials and Methods
Abstract (I loath the abstract)

Jane said...

Great question! Maybe I'm in the minority here, but I always start with the abstract. A very, very rough one, basically of the form "here is my hypothesis and here is what I did (in one sentence) and here is what I think the results are going to show." Then I write enough of the methods/experiments to remind myself of what I did to get and analyze the data. :) Then I write up the results and figures as I run the experiments and gather the data. I find that I then keep going back to the abstract and the methods and changing those as the experiments and analyses change. Once I have a fairly good results section and it's clear what my results are, I write intro, discussion, then background/related work. And then revise the heck out of it.

PhizzleDizzle said...

MCH - I started to read the link and my eyes totally crossed. The names they give to celestial bodies!!! ZOMG! They're crazy!

Jane - Your way is my general MO, except that I've never been comfortable with it. I think our common MO might have to do with the fact that as computer scientists we can run simulations and script up a graph at the very last minute :).

This paper is one of the first I have worked on where I will have a significant body of data *long* before the deadline, which is why I think starting backwards this time will work better for me.

But I'm gonna put the abstract last. I think only after I know exactly what I'm saying will I be able to summarize it briefly.

This really was so helpful today - my results section is feeling MUCH better...I nearly skipped and dance home from the office today! Thanks all!

PhizzleDizzle said...

edit: I'm going to put the abstract last, where I typically do it first and then change the crap out of it by the time I turn it in. trying a new thing this time!

Candid Engineer said...

PD, I'm sure you're already well on your way, but before I start writing anything for a paper, I ask myself what story do I want to tell. What is the best story my data can tell? What is the most important story?

And I build my figures around the story. And then, as others have commented, everything falls into place. Although I admittedly often write the Methods section first because it doesn't require a lot of thought and it gets the ball rolling. :)