How to Format a Science Paper

Science papers, the kind you find in professionalSay what you set out to do, and what you
journals, are usually write-ups of experiments, orexpected to happen, which is to say, lay out your
other formal investigations. They follow a sethypothesis. It's a good idea to lay out the null
structure, based on the standard architecture ofhypothesis too. E-mail me if you want this defined.
experiments themselves.METHOD (sometimes called other things, remember)
Science articles, by contrast, in magazines, are moreDesign: First describe the study's design, i.e., whether
relaxed versions of science papers, with much moreit's double-blind, randomised, placebo-controlled, or
discussion of the wider importance of a particularwhatever.
investigation, or group of investigations, and whatParticipants: Then describe the group you involved.
makes them so interesting.Say things like this: 'We included 8 patients (6
There's more than one way to do an article. (Yourwomen, 2 men: age range 21-35 years: mean
editor chooses which way.) For papers, however,± SD; 29 ± 6 years). All patients had
there is only one formula, with no real variation. I'll laypresented the same condition. None of the patients
this out for you here, without going into thehad any other medical conditions, or were on any
philosophy of scientific inquiry--I'll assume you knowmedications that could affect their response to the
why experiments are designed the way they are. (Iftreatment.'
you don't know, and you want to, send me an e-mailMaterials: List everything you used, even
about it.)questionnaires and handouts. (Blank specimen copies
Here are the sections.of these can go at the very end of the article, as
The Effect of X on Yappendices.)
G. Doctor and K. ResearcherProcedure: Write out EXACTLY what you did--in such
________________________detail that anyone can reduplicate your experiment.
From the Department of Planetary Sciences,This must be childishly explicit.
Gudger College, andthe Department of Rocketry,Data (or Data Analysis): Say how you crunch your
Universityof Blaupunktnumbers. There needs to be a measure of
[First comes an 'abstract', the paper written incentralizing tendency, and a measure of dispersal.
miniature - it's actually the last thing you write. JustifyAfter that, studies differ. Most of us just ask
the right margin for this little section.]statisticians what we're supposed to do.
Background: Say the known scientific mechanismResults: Lay out the raw data. You can do this in
you're dealing with. Next, say what aspect of it is nottext, but you should also include some tabulation.
yet understood. Suggest your mental model thatAfter that, you can interpret with graphs, though
could explain that not-understood aspect.some people like to leave these for the next section,
Hypothesize that if you're right, then if you do Xwhich is what you conclude from your results. Graphs
you'll get Y.are specialised. Make sure you know the definitions
Objective (or 'Aim'): 'To assess whether... [say yourof 'bar chart' and 'histogram' (they're not the same,
hypothesis].'even if they look alike). Don't use scattergrams
Method (sometimes 'Method and Materials',(sometimes called 'scattergraphs')--they're for
sometimes 'Procedure'): 'In a double-blind randomizedcorrelation studies, not controlled experiments. Pie
placebo-controlled study, 8 patients (21-35 years)charts are only for percentage shares of things, as a
each received [whatever it was]. Assessment was atrule.
baseline and at 14 days after treatment.' You canConclusion: Your verdict, expressed in terms of
add a word here about how you managed thewhether your results can or cannot support your
statistics.hypothesis. Keep this short.
Results: 'Assessment at 14 days post-treatmentDiscussion: This is the longest and most entertaining
showed a significant... [lay it out very briefly].'section. It's where you quibble with your own study.
Conclusion: 'We show for the first time that [X hasYou point out all the weaknesses in your sampling
an effect on Y].'strategy, your management of variables, your choice
[Now you start the paper itself.]of control group strategy, and so on. The idea is to
BACKGROUND (sometimes called INTRODUCTION)qualify your answer, so that no one else can. This
Say what's known about the big thing you'releads naturally into the last paragraph, where you
investigating. List all the relevant studies.normally say that 'future studies should consider...'
Say what's missing from all the studies so far.(whatever you left out or did wrong).