The key ingredients of an academic paper

It’s writing season!! Or at least, it seems to be in my group.

Writing is not easy. I LOVE writing, but this is after many years of trying, doing, getting feedback, rewriting, and getting more feedback. Bizarrely, I was never taught how to write, and I have the impression that it’s still not generally taught very well, because a lot of students hate writing, and because of the common – but easy to avoid – mistakes that they make.

There are a lot of fantastic resources on how to write well. I have Josh Schimel’s as well as Stephen Heard’s books on my bookshelf, and I recommend my students to read this useful post on Dynamic Ecology, and the British Ecological Society’s guide to scientific writing.

But I still feel that while these resources are very good, they might be too lengthy or too detailed for some of my students, and I end up explaining the most important ingredients of a good article or thesis again and again.

(OK, so this post turned out to be quite lengthy, but you can just skip it and jump to the pictures.)

When reading through a few student papers, it suddenly dawned on me that what I really need is not a lengthy or detailed explanation of how to write an article, but a brief and intuitive overview that shows the key ingredients, and the links between them, at a glance (similar to this graphic that I made about the ingredients for a proposal). Because yes, writing gets easier the more often you do it, but it is also crucial to know the ‘recipe’ of what makes a good introduction and discussion. And yes, also what makes a good methods and results section.

So, I made the following overview of what are, in my opinion, the key paragraphs in the introduction and discussion of a scientific paper or thesis, and the links between them. I also made one for the methods and results sections – this one is less prescriptive, but it does include some important do’s and don’ts, and there are some links between these two sections – for example, I like to structure my results roughly similar to my methods section. 

The introduction 

An introduction doesn’t have to be long. It just needs to introduce the knowledge gap – the why of the study – and the key relevant concepts and current knowledge. The introduction is really there to introduce your study! One very frequent mistake that I see is that not all parts of a hypothesis are introduced – after reading the introduction, it should be very obvious what your hypothesis will be. Really, there should only be one possibility. Given the information in your introduction, how could the hypothesis be ANYTHING else? For example, when you hypothesise that soil microbial communities underneath slow-growing plants will be better able to cope with drought than those under fast-growing plants, you need to introduce all those aspects in the introduction, and explain the links between them. So, you’ll need to talk about microbial communities, about slow-growing vs. fast-growing plants, about their response to drought, and about the links between microbial communities and plant growth strategies. You should not have any part of your hypothesis unexplained.

It’s important not only that the introduction starts broad, highlighting the major problem and knowledge gap, and then increasingly goes into detail about specific information and knowledge gaps, but also that the paragraphs link together. It needs to have flow, and for that, you need to incorporate logical transitions from one paragraph to the next. It’s too much to also give examples of these transitions here, so I might write a next post about this.

Another of my pet peeves is that the hypothesis needs to have a direction. So, for the example hypothesis in the previous paragraph, don’t say “microbial communities under slow-growing plants will respond differently to drought than those under fast-growing plants”. Instead say “microbial communities will be better able to cope with drought under slow-growing plants than under fast-growing plants”. 

And then another of my pet peeves: don’t be vague. No handwaving. Be clear, and be specific, throughout your writing. And that’s especially true for the discussion. Really, if you don’t understand what you are writing, and you try to get away with being vague, do you think the reader will get it? There is simply no place for fuzzy and vague writing. 

The discussion

I’m not going through these sections in chronological order! I think that the introduction and the discussion should really map directly onto each other, and that’s why I put them in one graph.

I like it when the discussion very briefly reiterates the aim of the study, and, in the first paragraph, directly outlines the main findings and whether they support the hypotheses that you wrote in the final paragraphs of the introduction. Then, you can discuss the findings in more detail in the following paragraphs, starting with the most important findings, and ging into more and more specific findings, along with explaining them and putting them into context. 

Don’t just end your manuscript abruptly! Include a concluding paragraph that – VERY – briefly summarises the main findings and what they mean. How does this knowledge help us, and what are the implications? This doesn’t have to be world-changing! It could also simply be an advance in our understanding, which will inform future work.

The methods

The methods and the results are the most technical part of a paper, and in that sense they should be easier to write. It’s all very factual and there are no arguments or interpretations to be made. But they are very important! Your methods need to be written in such a way that the reader can repeat your experiment. They need to be clearly and intuitively structured, and it helps the reader if your results section broadly follows the same structure. 

When writing about your statistical methods and models, it greatly helps if you specify what each model is used for. For example, you can write “The effect of individual glacier and time since deglaciation (soil age, this excluded the never-glaciated Reference sites), and their interaction, on soil, vegetation, microbial properties, and 15N pools was tested using linear mixed effects models with an error term for block to account for the non-independence of the five replicate samples per soil age. This was done using the function lme in the R package nlme (Pinheiro et al., 2020).” (Text taken from De Vries et al. 2021.) In my opinion, you can never really be specific enough, and of course, you will make your code and data available when the manuscript is published, for example through a repository such as Figshare. 

The results

It should not come as a surprise that the results should clearly describe your… results! But the devil is in the detail. I find it extremely important that you are specific, refer to the specific figures and tables that support your statements, and include the relevant statistics. So, never say “drought had an effect on root exudation”, but say “drought decreased root exudation across species (ANOVA, main effect of drought F1,24 = 7.8, P = 0.002, Fig. 1). 

It helps the reader if you structure your results clearly, either following your main questions/ processes, going from the most general ones to the more detailed or complicated ones (from main effects to interactive effects), or following the same structure as your Methods. 

You should number your figures and tables as they appear in the text, and everything you say needs to be supported by your display items (figures and tables together). If you can only include a certain number of display items, you can include more in the Supplementary Material, and also number these as they appear in the text. Be aware though that for the reader it’s not very nice if entire sections in your results rely on Supplementary Material (many readers will not go through the effort of looking at those!) – if this is the case, you might consider moving those display items to the main text.

Make sure your figures are clear, intuitive (don’t make the control treatment red and the drought treatment green, for example!), visible for colour blind people, and (ideally) understandable on their own without having to read the main text. 

Also – bar plots are a thing of the past! Use a box plot or overlay the actual data points to show the spread of your data.

A good, beautiful, intuitive figure says more than a 1000 words, and also, you can use it in a graphical abstract or in your tweets about the paper!


I think I forgot one thing. The abstract!

Well, I always write this at the very end, because only then you need the final contents, the final conclusions, of your manuscript. How to write an abstract? I think that might be my next post!


Leave a Reply

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

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

Facebook photo

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

Connecting to %s