Most scientific articles are written as if the research followed some kind of waterfall model. In other words, a typical paper proceeds pretty linearly from the idea to the results to the implications of the research, hiding the fact that one thing might have lead to another in a very non-linear way.

I guess the reason for the linear format is to attempt to make the research easy to digest for others, and to keep the paper as short as possible. However, what should one do when following the standard format becomes a too large burden (e.g. excessive cross-referencing), or hides a key insight about the research process itself, which could be valuable for others to know?

  • best strategy for me always has been to read papers in a different area of research, and look out for the way they structure things there. I find theory papers to be a good place to look for alternative structures. Seems that (from the outside at least) theoretical work is less prone to the streamlining. Is the waterfall model a US thing? I have never heard of it.
    – marts
    Jul 31, 2016 at 13:36
  • 2
    What field is this? Paper writing can vary quite a bit from field to field (I'm in math and my papers never are organized in the order: ideas, results, consequences).
    – Kimball
    Jul 31, 2016 at 13:38
  • AFAIK there's no rule that says that a paper has to be structured in any particular way (although they may be the occasional reviewer who thinks that there is and that it is their job to enforce it...). If the usual way doesn't work, do something different :-)
    – Flyto
    Aug 5, 2016 at 6:42
  • Sorry to be blunt (and no offence meant at all), but to me it looks like you are trying to give a cover-up reason for your lack of effort in structuring the paper. Your research may not have been "linear", but there is no reason why you cannot reformulate your findings in a clear and direct way. Aug 14, 2016 at 19:07

3 Answers 3


Whatever format you choose, you have to make sure that your audience can grasp what your paper is about within the first 4 sentences, that (s)he won't get "lost" while reading your article, and that it is easy to strategically skip much of the article and still get the main message. Remember that there is way too much being published, and that it is impossible to read it all. So most articles are only quickly glanced at and are only read if you convince the reader while (s)he is glancing at your article that it is interesting. This is the benefit of the standard format of articles.

While writing your article, you are obviously convinced that your article is worth reading, but keep that reader in mind for whom your article is just one of way too many and who won't read your article very attentively.

For example, the research process almost never follows the strick linear form in which it is presented in most articles. If that deviation is somehow relevant, then you can often use footnotes to discuss that without deviating from the standard form. For things that need more room, there are (web-)appendices.


I recently fought my way through unpicking a thicket of cross-references like this. By focusing on the hourglass model (broad problem, specific problem, new contribution, specific implications, broad implications) and pruning unnecessary tangents I was able to shuffle the content around into an order without any forwards references, though doing so took me months.

Your process in producing the research and paper are not the paper's topic. I would think twice before even referring to it.


Though most of the papers publish only mention their liner approach, some do try to mention the multiple path summery (mostly psychological research), but that's where citing comes, an author conducts his/her research in a single path with controlled conditions, and opens door to other authors to take other path.

Even if you plan to include all the possibilities of your research outcomes, you should first follow waterfall model, and just jot down other possible outcomes for future, because reaching the destination is initially lots of trial and error process.

PS: it will mostly depend on your area of research and the approach you are taking

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .