On Thursday, I’m going to be holding a talk about how to structure a research paper. In preparation, I thought it would be a good idea to write a sort of prose outline of the talk, summarizing the function of each section of a paper in a few simple sentences.
In the introduction you evoke a world, invoke a science, and propose a thesis. You present a familiar but interesting context for your study, frame it with a standing consensus or ongoing controversy in your field, and state a conclusion, based on your research, with significant practical or theoretical implications.
In the background section you inform the reader about the practical context of your research question. You provide the reader with the references to the most reliable public sources that you are aware of, so that the reader may become as knowledgeable as you are about the conditions from which your research object emerges.
In the theory section you tell the reader what you expected of your object of analysis before your did your research. Or, perhaps better, you remind the reader what the reader would have expected your analysis to show if you hadn’t already told them in the introduction. Alternatively, you shape the reader’s curiosity about your object, their curiosity about what your data will reveal, about how the data will support your thesis.
In the methods section you explain what you did to collect your data and why you did it that way. The aim is to win the reader’s trust, respecting their natural skepticism and awareness of typical sources of error.
In the analysis section, you present your data in a way that either challenges the reader’s expectations or satisfies the reader’s curiosity. You offer your interpretation of the data, supported by the observations you have made.
In the discussion section, you explicate the implications of your analysis for either theory or practice (or in some cases both). Now that we believe your results (having seen the data you gathered by a trusted method) what changes to our ways of seeing (our theories) or our ways of doing (our practices) are we rationally committed to? How ought we to proceed from here? What can reasonably be asked of us?
In the conclusion you restate your thesis plainly and simply, with all the presumptuousness your theory allows you and all the confidence your method affords you. And you return the reader to the world or the science with which you began, set in a slightly different light and given a slightly different weight.
A typical research paper in the social sciences is about 8000 words long and consists of about 40 paragraphs, each stating a single claim. Since each paragraph consists of less than 200 words and takes about a minute to read, it should be possible to read a paper reasonably carefully in about three quarters of an hour. For comparison, this post is 500 words and should take you two or three minutes to read.
8 thoughts on “Towards an Outline”
Thank you, Thomas.
I look forward to hearing your talk.
Any special recommendations for theoretical papers?
Hi Sonia, you’re welcome. I wrote about theory papers many years ago on my old blog. See this post and this one. I’ll take your question as a spur to migrating those ideas over this blog soon. Best, Thomas
Thanks a lot!
This is as good a summary of this genre as I can imagine. But I have two questions/comments.
1. This says what to do but not how to do it. It would be useful to give some examples of both good and bad ways of doing this. Often, many people will think they are doing something like: “inform the reader about the practical context of your research question” but they will still leave the reader confused.
2. This outline presupposes the ‘ideal’ or idealised reader who reads the paper from beginning to end, word by word, sentence by sentence. But that is not how research papers are generally read. If you look at advice on how to read for university students it is to do exactly the opposite of reading linearly. And in fact, when I speak to students that is the thing I recommend they do. My mantra “The logic of discovery is not the same as the logic of presentation.”
I always want to ask: How many papers cited in the average academic paper have been read all the way through? When we see the enormous amounts of miscitation, we can conclude not a lot. How many of them have not been read at all?
So I think a part of that recommendation should be to think of every section as if it’s the only one being read.
I’ve noticed a new trend in some journals to start with bullet points of the key takeaway lessons that comes before the abstract. I think that is definitely a good practice to get into. The writer should ask themselves the question, can I summarise the point of the paper in 5-7 bullet points and then, have I done enough work to justify those 5-7 bullet points? Would the reader come up with the same bullet points?
Also, I wonder whether what you’re describing is in fact best described as an outline? This is more of a framework or a mold. Many beginning writers actually struggle with the outline of their argument more than with how to fit it into this mold.
Thanks for the comment, Dominik. There’s enough there for a whole post, so I’ll just note a couple of things for now.
1. Yes, this post was just laying out the what, not the how. It now serves as TL;DR for the one-hour talk I did last week. I’m of two minds about giving examples. More on that in a post sometime.
2. I’m not sure my outline presupposes a linear reader. But I do think papers should be written to reward (and certainly not to punish) the ideal reader. “How many papers cited in the average academic paper have been read all the way through?” you ask. “How many of them have not been read at all?” I agree that the answer may be disappointing. But I don’t think we should write papers to be cited but not read. I will certainly never recommend that.
My ideal introduction requires the writer to be able to summarise the paper in three simple, declarative sentences (the key sentence of each paragraph) and paragraph three ideally consists of eight sentences that go into a bit more detail. I think that’s as good (and in fact better, because it’s actually prose) than 5-7 bullet points.
Finally, I agree that I’m not here providing an outline. It was an outline of my talk about the “structure” of a paper (arguably a “mold”). These general ideas indicate a path toward an actual outline. I provided a heuristic on Twitter recently. Answer each of these in a single declarative sentence or two:
Where are we?
What do we expect?
What did you do and why?
What did you find?
Why does it matter?
That’s the social science research paper in a nutshell. Background. Theory. Method. Analysis. Discussion.
Thanks for the quick reply. Two points I found interesting.
First, I did not mean to suggest that people write papers not to be read but just cited. Rather, I would suggest that writers acknowledge that their papers may not (actually will not) warrant everybody’s careful attention. They should therefore structure them in such a way that more of them are read and when cited, the citation is in the appropriate context and with the right emphasis.
One way with which to do that is to provide more structure within the structure. Include subsections within the main section, when there’s a list of something, present it as a list (numbered when appropriate).
For instance, the answer to the question what did you find or why does it matter may require more than a single declarative sentence. But probably less than a paragraph. Thus I suggest a list – or even a structured list.
You say ‘prose is better’ than a list I would say that that is not true in all contexts. Many readers will appreciate a list of the main points without all the unnecessary connectives which are often doing as much genre signalling as actual cohesion building.
Lists are easier to scan – particularly when the person doing the citing is coming back to the paper (or only coming for a quick footnote and have no intention of reading the whole thing).
Second, I would certainly like to see a post on the pros and cons of giving examples. Personally, I try to err on the side of more examples rather than fewer. Particularly in case of procedural knowledge such as writing. But I’ve also seen (and perpetrated myself) really bad examples that are so trivial as to be misleading.
I quite liked Stanley Fish’s book on ‘How to write a sentence’ which is very ‘example forward’ but more importantly it provides a direction for the reader to gather more examples. (Until it loses its way in the second half – certainly a book that would have benefited form less prose and more lists).
But often examples given are very vacuous. Pinker’s book felt that way to me or some of Helen Sword’s. So where to find the balance? There are many who rail against passives (or other pet peeves) and yet the examples they give are either not passives or not the passives as actually used. Their examples do not help anyone write better but at least they reveal their ignorance. So examples keep us honest, too.
A very quick response: the reason papers don’t always need to be read in their entirety is that the reader already knows a great deal about the subject under study, the theories that were applied, the methods that were used, etc. This knowledge allows the reader to quickly skim over some parts and focus on others. I don’t think it has very much to do with how the paper is written. The only way to make sound decisions about a paper is to compose it one paragraph at a time, with the intention of occupying one minute of the reader’s attention at a time, and then to arrange those paragraphs/minutes in the most effective way possible.
The post about examples is now up here.