Writing a journal article

From Sustainability Methods
Type Team Size
Me, Myself and I Group Collaboration The Academic System Software 1 2-10 11-30 30+

In this article, you will find a range of material that is hopefully useful to you when you prepare your paper.

The target audience is research students and other early career researchers, especially (but not only) in the environmental sciences and sustainability. My own background is in the natural sciences, but I have also worked with social scientists a fair bit. My suggestions are necessarily biased by my own experience. If you work in a field that is very different from mine, you may find that some of my suggestions are not as useful to you as others: but hopefully you will still find that some of the material is useful.

First Steps

Before you write an article for publication in a journal, there are a few basic things that are good to think about.

Content: You need to know what it is that you are planning to write about. Once you know what you want to write about, there is still the question of how to best present the content. Many papers with solid content could have been published in much more highly ranked journals if they were presented slightly differently. The issue in such cases is not so much the main content as such, but how it is prioritized, structured, at which length and level of depth it is communicated, how it is embedded in current debates, and how it is framed.

Your target journal: You need to understand your target journal – what will appeal to the readers, to the editors, and what kinds of particular policies does the journal have. There are papers that are very good but are rejected because they are submitted to the ‘wrong’ journal. Moreover, there are brilliant papers that are accepted in a good journal – but then nobody reads them! That can happen if the paper is a poor fit relative to what the journal overall is about. Study your journals a bit, flick through recent tables of contents, find out something about the editors, read how the journal presents itself in its ‘scope’ section, and so on. You will find that the same content might fit quite different journals, depending on how it is structured and presented.

Framing your paper

For a given quality of science, how the paper is framed probably makes the biggest difference for where it is published, and how well read and cited it will be. Framing, essentially, is how you fit your paper into the ‘bigger picture’, or if you prefer a less neutral term, it is how you sell your paper. My general advice is that you choose the largest plausible frame for your paper that is reasonable. In other words, to make sure you’re widely read it makes sense to find a big hook off which to hang your paper – but you have to be careful not to oversell your work. This is subjective.

Typical choices for framing papers are current global issues, big theories and new theories, particularly those that are controversial. The trick then is to make the link from the ‘big issue’ to your study as swiftly and directly as possible at the beginning; and back to the same big issue (plus sometimes additional ones) near the end of the study. A good frame is one that appears reasonable and can be directly linked to your work, and one that your work directly speaks to. A bad frame is one that takes a lot of imagination to see how your study fits in with it, and that your results don’t have a lot to do with.

Often but not always your framing is partly determined by how you planned your study (as opposed to your paper) in the first instance.

Possible exercises:

  • Brainstorm a list of possible frames. Initially, don’t be too selective — just write down lots of possible frames that might be connected with your paper.
  • Read your discipline: where are things at? What’s currently a hot topic? What kinds of topics have recently been published in the most prestigious journals? Does your work fit in with any of those topics?
  • Generally: can you use a certain bandwagon for your benefit, and jump on it for a bit? Don’t actually change your focus too much — but can your focus be twisted ever so slightly so that it becomes interesting to a broader audience?

Planning your paper

I suggest several phases to a paper:

  1. Think about what you want to write. When I say ‘think about’, this thinking could take many different forms. Ironically, some people only think clearly when they write, others need a whiteboard, others need to talk things through with their friends. No problem! The strategy does not matter, as long as you come up with a sound logic. Note, however, that if you do write at this stage, don’t become too attached to your writing. This first step is really about clarifying your flow of the argument in your head. Communicating it effectively on paper is a separate step.
  2. Think about where you want to publish your work. This is an iterative process, with two main consequences. One, you may need to re-think how you choose to frame your paper (see section on framing), and therefore you might have to go back to the previous step. Two, you will find that there is a particular word count you need to fit within.
  3. Now that you know the word count, you need to fit the content to the length specifications that you need to work with.

Prioritising content

It is important that you prioritise what goes into your paper, and what you leave out. Especially when you start out in research, it seems appealing to pack in as much information as possible into your first few papers. After all, you did a lot of research, which was hard work – and you gained a lot of insights! Now surely it’s worthwhile communicating these insights as completely as possible.

That logic holds only in part. Especially if you need to write a short paper (e.g. because that’s what the journal requirements are) it is critically important you set yourself some clear priorities about what matters, and what doesn’t.

But even when the word count is not a problem, more is not necessarily better. Imagine a paper with three really good points (each of which makes you go “wow, very clever”), two quite good points (which make you go “that’s a fair point”), and two reasonable points (which make you go “sure, but not particularly exciting”). And now imagine an alternative, shorter paper, which just makes the three really good points. Which paper will leave a stronger impression on you?

This might depend on what sort of a reader you are. But it stands to reason that the short paper, with only really good points, will keep a reader engaged for every single paragraph. The longer version of the same thing, by contrast, might have the reader drift off, flick pages, or skip on to a different paper. So, don’t just think about the marginal contribution of an extra thought or extra text: of course, more text always means more insight. But also think about the average quality of your thoughts: sometimes, additional thoughts that are not particularly interesting just dilute the quality of your work, leaving the reader to feel your paper is “just average” when in fact you had some very good points – but those may have got lost among the less good points.

Fitting the content to the length

It is important that those bits of content that are really important to you are not overshadowed by other bits of content, which actually are less important to you. If the important parts are overshadowed, this effectively means they get lost among other, less important parts. Overshadowing of important content most commonly happens because the good points receive too little space in the paper relative to other, less important points.

Two main strategies can help with this problem at a general level. First, see my suggestions for prioritizing your content – the gist is to focus on your most exciting ideas, and perhaps just leave some of your other thoughts untold. Second, you can sometimes use supplementary online material to explain things that are not of central interest to all readers.

In practical terms, I suggest you write a dot-point outline of your paper, with all headings and sub-headings included; and dot point summaries of the main content below. Such a skeleton outline can be very helpful because it’s all about the logic and the flow. The wording is completely irrelevant at this stage, because this outline needs to make sense only to you (and perhaps your co-authors).

Once you have an outline, think about which sections are particularly important to you. The relative importance of a section should – to the extent possible – be mirrored by its length. In other words, it is not a good idea to have 2000 words of background material, followed by 200 words of brilliance, before a 500 word conclusion. Much better, in this example, would be to have 700 words of background material, followed by (well-structured) 1000 words of brilliance, followed by a 250 word conclusion.

Go through these steps, and the general steps for planning a paper, several times in an iterative fashion. What you end up with is a dot-point outline summarizing your headings and main argument, including an indicative word count that each section will occupy. You can also add key references to this outline.

Such an outline is an incredibly powerful way to start "writing" properly. The idea is to leave the actual writing until the content is completely clear. This has the advantage that you now know precisely what you want to say – you just have to put it into words. Sections that are too long, rambling, or don’t fit become virtually impossible if you follow this approach.

Note, again, that if you are the kind of person who develops their thoughts as they write, that’s not a problem. But don’t think your initial writing (even if it’s nicely worded prose) should be the same as the paper in the end. Once you have done your first bout of writing, it’s still a good idea to condense things back into an outline, and then write again.

(Of course, with practice, you may not need this elaborate process, but it is a pretty fail-safe approach.)

Possible exercises:

  • Write and re-write an outline for your paper until the logic is smooth
  • As you go, consider different journal options — what does this mean for which content you may have to leave out, and how many words would you devote to different sections if you were to submit to different journals?


The content-to-length ratio accurately summarises what more and more high-end journals are all about: cramming as much exciting, new information into as little space as possible. Not all journals make this a high priority, but it is increasingly common. Journals that pride themselves of rapid turnaround, have short papers, have ‘Letters’ in the journal name, regularly use supplementary online material, and have high impact factors are very likely to like a high content-to-length ratio – whether they explicitly state this or not. If you get the sense that a high content-to-length ratio is expected of you, it is particularly important that you prioritise your content and fit your content to the prescribed length. Your writing style also becomes more important, because there is no room for extra words.


Throughout the writing process, you need to think about structure. Most people think of structure as something that applies to the paper as a whole. However, structure is also important at finer scales, such as the level of sections, paragraphs, and sentences.

Structure: the paper

No matter what type of paper you write, it needs to have a clear thread through it, and sections need to clearly link. One of the challenges is that writing is linear – it has a start point and an end point. By contrast, much academic content is complex – more like a website, where things are related in many different directions. The challenge of writing is to turn the multi-facetted nature of the content (where everything is related and linked to everything else, like the internet) into a simple, one-directional argument.

Some general principles hold. At the most general level, it’s a good idea to start your paper broad, have specific aims at the end of your introduction, and then have a detailed ‘meaty’ part in the middle. Towards the end, you need to get back to the big picture, preferably the same context that you started with. The other general principle is that you must not assume background knowledge beyond the obvious in your discipline. In other words, your chain of argument must not leave out steps that are actually important for the reader. Ask yourself how you would need to explain the general gist of your paper to your friend who studies a different discipline, or to your aunt. A good structure tends to make sense even to ‘uninformed’ people. If they don’t get the basic structure you’re proposing, it’s likely that you have left out important steps in your logic.

I distinguish between a few different types of papers. Standard empirical papers tend to follow a standard structure: introduction, methods, results, discussion, conclusion, acknowledgements, references. Sub-headings are common (and useful) within methods and discussion, in particular, but sometimes also in the results section.

Essay papers are fundamentally different. They have no set rules, which makes it even more important that they follow a logical, and clearly understandable thread. Often, it’s a good idea to outline what this thread is specifically at the end of the Introduction. Also, even though it’s an essay, sub-headings can be immensely useful. Check your subheadings – if, without any further information, they tell a logical story, you’ve probably worked out a useful structure.

Review papers are different again. Again, there is a lot of flexibility for how exactly those are prepared, and it depends a lot on which journal you’re writing for. It’s worth noting that in ecology and conservation science, journals are increasingly keen on quantitative reviews or meta-analyses (a statistical approach to extract patterns out of multiple case studies).

No matter which kind of paper you write, headings are critically important. Typically, you will want to use 2-3 heading levels. You need to clearly differentiate these in style, so that it’s easy for readers to see what you’re doing. E.g. use a bold 15 point font for main headings, a bold 13 point font for subheadings, and an italics 12 point font sub-sub-headings. It’s a good idea to match the journal style.

Possible exercises:

  • What type of paper are you planning to write?
  • What is a logical sequence of headings?

Structure: section and paragraph

A section consists of several paragraphs. The ‘rules’ that apply to structuring sections and paragraphs are pretty similar.

The first sentence and last sentence are disproportionately important. The first sentence needs to be as clear as possible about what the paragraph is about. Avoid first sentences that are just background, or pick up the idea from the previous paragraph. If, for example, your paragraph is about limitations of a particular conceptual model, a good first sentence might be: “The xxx conceptual model has several inherent limitations.” Or, if your heading already says as much, your first sentence can potentially go one step further, such as: “Inherent limitations of the xxx conceptual model arise from factor X, factor Y, and factor Z.” In both cases, the first sentence makes it very clear what must come next, namely details about the limitations. This helps the reader in several ways. First, she will know what the section will be all about; and second, this helps her decide whether to read the paragraph or skip straight to the next one. The importance of allowing readers to skim-read cannot be overstated. A good document is one that enables us to quickly grasp the content. First sentences of paragraphs help immensely, if they are well written.

The last sentence of a paragraph or section has to fully wrap up the content. Make sure that your thought is not left hanging, only 90 % complete. You need to fully finish your thought so the reader has no doubt about your intended ‘so what’. So, when you think your paragraph is finished, ask yourself: ‘So what?’ If you just need to read out the last sentence again, you included the so-what. If, however, you need to come up with a new, additional sentence, it shows you hadn’t quite finished your thought. This little exercise won’t always work, but it may help you to test yourself whether your take-home message would actually get through to the reader.

Apart from the first and last sentences, many paragraphs list things, explain things, discuss causal relationships, or contrast things. In all cases, simple phrases can help to make the argument clearer. Don’t be afraid of listing things as “First, …” and so on, or starting sentences with “Although …”. Simple words like these at the beginning of sentences can make it much easier for your reader to find the thread of your argument.

Possible exercises:

  • Both in existing papers, and in your own draft, check the first and last sentences of several paragraphs and sections. Do they introduce the content, and wrap up the content? Or are thoughts left hanging?
  • Can you structure your argument within a paragraph more clearly? For example, by using numbered lists, or simple words that link your sentences causally?

Structure: the sentence

Direct sentences are easy to understand, whereas indirect ones are difficult to understand.

This was an example of a very direct sentence. It has a few properties:

  1. The most important notion comes first and tells you what the sentence is about.
  2. It has a simple structure, in this case two parallel phrases with identical grammar.
  3. It uses simple words and avoids jargon. I used ‘direct’, because it is a short and simple word. And to contrast it with something else, I used ‘indirect’. I could also have used ‘convoluted’ or some other word, but the simplest word is the one that creates the sharpest, clearest, black-and-white contrast to the other one. Note that the era of big words is finished, at least in scientific writing! The simpler the better. You don’t come across as smart if you use (unnecessarily) big words, but as overly complicated.
  4. It uses unambiguous words. For example, I used ‘whereas’, which has only one meaning – I avoided ‘while’ because this has a temporal meaning as well as a contrasting meaning. Similarly, I used ‘difficult’ and not ‘hard’, because hard has multiple meanings.
  5. It is short! Many short sentences are much easier to follow than few long ones.

Two additional questions concern tense and voice. Most academic writing is in past tense, unless you report generic facts not related to your findings or other recently reported empirical findings. Most importantly, make sure you do not accidentally jump between tenses.

The use of passive voice was common in academic writing in the past. For example, an ecologist might have reported “All animals heard or seen were recorded by two experienced observers”. Today, this sentence would probably read “We counted all animals heard or seen”. Although writing in active voice is now preferred by many journals (some have it in their style guide!), this is not to be mixed up with necessarily using the first person. In the example above, ‘we’ is the first person plural, and so in this case, indeed, the active voice wording also uses a first person perspective. But take a typical sentence from a results section, such as “Companies used a variety of strategies to market their green credentials”. This is an active voice sentence (in third person plural), and there is absolutely no need to somehow try to turn it into a first-person sentence. For example, adding “We found that…” at the beginning of sentences is not necessary (though you will find that some authors now do just that) – in fact, it makes the sentence longer and unnecessarily complicated.

Finally, a warning for those operating in a German academic context…: modern academic writing in English is very different from classic academic writing in German. Avoid nested sentences, long sentences, hidden meanings or clever metaphors.

Phrases to use and to avoid

Through time, you will develop your own writing style, and that’s a good thing. However, there are certain phrases that help in writing, and others that should be avoided. Here is a short list of them; presumably very incomplete, but nevertheless a starting point.

Avoid Use
Like xyz such as xyz
(e.g. xyz)
Etc. spell out and end the list with ‘and’ or ‘or’
A number of Several or many; ideally list the actual number if possible
But (at beginning of sentence) However, …
… while … … whereas … (use ‘while’ only in a temporal context)
… as … … because … (use ‘as’ only in a temporal context)
In contrast, … By contrast, … (though ‘in contrast’ is not technically wrong)
For example, …For instance, …
Furthermore, …
In addition, …
Notably, …
Surprisingly, …
About, roughly Approximately

Who to cite, when and where?

One important aspect of academic work (as opposed to popular writing) is that you need to acknowledge where various ideas came from. As a tutor of mine once said: whatever you don’t wake up just knowing one day, needs to be either referenced or arise directly from your results. But with so many papers out there now – what does this mean? What about citing multiple papers for the one statement? When should you cite how many papers, and which?

One approach to this is to cite a lot of other papers. The idea is (I guess) that you somehow ‘prove’ that you really know the literature. In this school of thought, even relatively obvious statements often get three or even more references to other work. This approach is very common, but I don’t think it’s very elegant.

An alternative approach is to focus on the most important citations. A good choice is to focus on the ‘classic’ papers on a given topic, as well as the most recent, high-impact papers on that topic. A good way of finding the ‘classics’ on a given topic is to search in online databases, and sort the findings by the number of times a paper has been cited. The ‘classics’ tend to be highly cited.

In addition to the classics, the latest ‘high impact’ papers make for good sources to cite. Those are basically the ones that will be relatively highly cited in the future, but are not highly cited yet – it typically takes about 1-2 years for a paper to get cited after it is published. To find the latest important papers in your discipline, you need to be aware of which are the most important journals in your field, and it’s a good idea to (regularly) read the table of contents of these journals, so that you’re up to date.

Key parts of a paper where most of the citations end up are the introduction (where you demonstrate how your work fits in with others, and that you know who else publishes in your domain), and the discussion (where you relate your findings to other people’s findings). Note that with very few exceptions, there are no citations in the abstract.

Sections relevant to all papers

Although there are many different types of papers, some types of sections occur in most papers.


The abstract is the first thing your reader will see. It’s also the first impression that editors will get of your paper. A good abstract will leave people satisfied that they know what you did, why you did it, and what you found out. A good abstract most likely means people will want to read the rest of the paper, and it greatly increases the chances of people remembering your paper later on. This, in turn, means a good abstract is important for people to cite your paper. If your main message is clear from the abstract, others are much more likely to recall what your paper was all about when they pick it up a year after first reading it – for example when they are in the process of writing a paper of their own and are looking for appropriate citations.

Abstracts differ greatly between journals, in both style and length. It is critically important that you follow the instructions for authors for your particular journal. Here is an overview of some of the differences that you might encounter.

Descriptive abstracts give some background information, and summarise some of the argument. Their analogue in the movie world is a trailer. They tell you enough about what the paper is all about so that you want to keep reading it – but they don’t give it all away. They often don’t give you the final take-home messages, but rather, they end in statements such as: “The implications of these findings for policy development are discussed.” Some disciplines use these kinds of abstracts a lot, but personally, I find them frustrating. Take my previous example just above. If there are important policy implications, wouldn’t it be much nicer to know what they actually are, rather than just knowing that there are implications? My sense is that descriptive abstracts should be avoided, unless you are dealing with a journal and discipline where it is expected of you to write such an abstract. Otherwise, readers get a lot more out of ‘real summaries’.

Real summaries is my slightly clumsy term for the more common type of abstract, which tells you about what you did, why you did it, what you found, and what the implications are. It’s really the latter that sets them apart from descriptive abstracts. Real summaries fully reflect the scope of the paper from motivation to take-home-messages. So, for example, a last sentence of a real summary might be: “Our findings suggest that a more participatory approach is needed to improve citizen acceptance of the suggested reforms to water policy.” This doesn’t just tell you that there are policy implications, but it tells you something (of course only very briefly) about what those implications actually are: in this case, the need for more citizen participation.

Structured abstracts are a type of abstract that some journals use. They tend to use a series of sub-headings or numbers; and authors are supposed to follow this structure when they write their summaries. The Journal of Applied Ecology is a good example of such a format. Typically, the different parts of the abstract refer to different sections of the paper. For example, there might be one part of the abstract that summarises the background and motivates the paper, one point that summarises the aims and study location, one for the methods, and so on. In the case of the Journal of Applied Ecology, the last sentence needs to be a clear summary of the take-home messages.

Although structured abstracts are relatively uncommon, almost all abstracts work well if they are first written as if they were structured abstracts. Typically, abstracts will mirror the structure of the overall paper. It is particularly important to have a clear first sentence that motivates the need for the paper and gives background information; and it is particularly important to have clear take-home messages. Perhaps the least important part of an abstract (in terms of length) is the methods. A summary of methods should be included, but often this can be quite short. Given that not everything can be included in an abstract, the methods can be dealt with relatively briefly, whereas it is critical that the motivation and take-home-messages are long enough that they are clear.

Abstracts vary widely in length. Some journals allow as little as 120 words, whereas others go up to 400 or so. Longer is not necessarily better, or easier to write. In all cases, it’s important to try to appropriate fit the content within the prescribed length.

Finally, it works best if you write your abstract last – to summarise your document, you first need to have your document finished.

Possible exercises:

  • Collect several abstracts from existing papers, both within your own expertise and outside your own expertise.
  • Analyse these abstracts: Do they follow a clear structure? Which abstracts tell you a lot, and which tell you very little? Which do you like, and why? Which are easy to follow, and why? Is the ‘so-what’ clear?
  • Write a short list of attributes that you found frustrating in other people’s abstracts, and a list of things you found really useful.
  • When you write your own abstract, use this list as a reference point.


The introduction of a paper is critically important. Even if your results are quite good, unless you introduce your work well, interesting results can come across as boring or meaningless. This issue is related to that of framing, which I discuss separately elsewhere.

Introductions need to start with a broad motivating statement, which takes up one or two sentences. The scheme for these sentences typically is something like this: “XYZ is a really important issue because of A, B, and C.”

Following such a broad statement, there is usually some review of existing knowledge. This serves two purposes. One, you need to embed your work in the context of other people’s work. Two (a more political reason), you need to demonstrate that you know about other important players in the area in which you are publishing. Reviewers get very annoyed if you have missed their very important work, even though it’s on a related topic. You can’t (and shouldn’t!) cite everything, but it’s a good idea to be familiar with who else is working in your area, and give them credit for their work.

Introductions then become increasingly specific, like a funnel. At the very end, you present specific aims. The logic of an introduction thus typically flows something like this:

  • XY is an important issue
  • For example, it has these effects, and these other effects
  • These have been investigated in a number of ways
  • Author A came up with this explanation
  • Author B proposed an alternative explanation
  • To date, it is unknown what the role of the phenomenon Z is.
  • Phenomenon Z could be important because of this, that, and something else.
  • Here, we investigate the role of phenomenon Z in the context of …
  • Specifically, we addressed three aims:
    • First, we tested whether …
    • Second, we compared our findings …
    • Third, we applied our insights to …
  • Optional last sentence to summarise the key finding: “We show that …”

The last optional sentence wraps up the introduction. It basically states the main finding. While that might seem unusual, it gives a strong ending to the introduction, and makes it very clear for the reader in which direction you are heading. Some leading journals now encourage a summary sentence at the end of the introduction.

How long should an introduction be? This depends on the discipline and journal you are writing for, and on your overall strategy to fit your content within the prescribed length. You might want to copy and paste the introductions from some papers that you like (which are broadly similar), and do a word count on them to get a rough idea for what is appropriate.

Possible exercises:

  • Collect the introductions from several published papers, both from your own area of expertise and from outside your own expertise.
  • Analyse these introductions: Do they follow a clear structure? Which introductions lead you clearly to key aims? Which take big detours? Do you think those detours are useful (because they add depth), or are they a distraction? Why? After the introduction, are you inspired to read the rest of the paper? Are the aims clear — do you know what’s coming?
  • Write a short list of attributes that you found frustrating in other people’s introductions, and a list of things you found really useful.
  • When you write your own introduction, use this list as a reference point.


Many papers don’t have conclusions. My personal view is that this is not a good trend – I think it’s a good idea to have a short conclusion that summarises your key finding, and key take-home message. This shouldn’t be too long or repetitive, but is worth having so that your argument is not left unfinished. Importantly, don’t start any new thoughts in your conclusion.

Possible exercises:

  • Find recent papers that are relevant to your own work. Do they even have conclusions?
  • If there was just one thing that you want your reader to remember from your work, what is it? What if there were two, or three things? List these things: they probably should be in your conclusion.


The referencing style differs between different journals. No matter which journal you submit to, you save yourself a lot of work if you use specialized software for managing your references and for formatting them (such as Endnote or RefWorks). This software can do three things for you: (1) make sure you don’t cite things that then are missing in the reference list, and (2) format all your references consistent with a particular journal style, and (3) re-format your references for a different journal style should you want (or have) to submit your work elsewhere. Like all software this takes a little time to learn, but if you intend to write several papers, the learning curve is absolutely worth it.


Before your references section, you should list your acknowledgements. This includes funding sources, colleagues who helped you (but are not authors), and it also includes referees if they gave you useful suggestions for how to improve your paper. For example: “Comments by two anonymous reviewers greatly helped to improve an earlier version of this manuscript.”

Additional sections to empirical papers

Some sections are particularly relevant to empirical papers.


For non-essay papers, the methods section describes everything the reader needs to know to replicate what you did. This means that all procedures need to be described in sufficient detail that somebody else could implement them again. Methods are usually written in past tense. The only references in the methods section tend to be to previous, related work (which the reader needs to know about) or technical papers outlining particular protocols. In the latter case, it is appropriate to summarise the technical procedure briefly, as well as point to a specific reference for further details. The only procedures that need no references are those that are accepted as ‘standard techniques’. It will depend on your discipline what is considered a standard method. An example might be simple linear regression. If you say that you used simple linear regression in an empirical science context, nobody will expect you to provide a reference. If you say, for example, that you used canonical correspondence analysis, however, people probably will want you to give a reference, so they can look up the details. So what is considered a standard technique and what is not is context-dependent.

Possible exercises:

  • Do you have lot of methods? If so, you probably need to think of a series of sub-headings.
  • What can you assume is general background knowledge in your discipline, and what requires explanation? Hence, which of your methods can you describe briefly (perhaps with a reference), and which need more space?


The results section of a (non-essay) paper is where the findings are presented. This is typically done in past tense, and without any references to other work. You simply summarise key points from your figures and tables, and explain additional findings in the text. You provide no interpretation of findings. The results section is purely descriptive.

Possible exercises:

  • List your results, prioritise them, and decide which to include in your paper
  • How can you best communicate your particular results: in text, tables, or figures? (Figures work best for many readers.)


Standard (non-essay) papers have discussion section after the results. This section is about interpreting the findings, placing them in a bigger context, relating them to other work, and presenting take-home messages.

Unlike for methods and results, you have a lot of freedom how you write your Discussion section – probably even more so than for the Introduction. Broadly speaking, there are two types of discussion.

The ‘boring’ type of discussion goes through your main results, one after the other, interprets the result, and relates it to other literature. The method goes something like this:

  • (Short re-statement of main result, e.g.:) “Many large companies integrated sustainability considerations in their daily operations.”
  • (Interpretation, e.g.:) “This may be because of X, or because of Y.”
  • (Relate to other people’s work, e.g.:) “These findings underline the importance of ZZZ (reference), as also shown in Japan (reference) and the USA (reference).”
  • (then the next main result, and so on)

Most discussions have at least some sections that follow this boring format. There’s nothing wrong with this, other than that it is, well, boring…

Towards the end of your discussion, and if you’re bold all the way through, you need a more creative way of putting things together. Even if you’re using the ‘boring’ format, you still need to get to the bigger picture, the big so-what messages. That is very hard to do unless at some point, you go for a more creative format.

It is harder, but more interesting, to start with a more creative format from the outset. A freely structured discussion needs to make a clear argument, and it can be very useful to use sub-headings to structure this argument. Even though the structure might be free and creative, you still need to draw on your own results under each sub-heading, not just on other literature – a discussion is NOT a literature review, but your work must be at the centre of the argument.

A discussion often picks up similar themes as discussed in the Introduction, and thus is a bit of a reverse funnel. Whereas an Introduction starts broad and becomes focused, a Discussion focuses on your results initially, but then ultimately must leave a clear impression with the reader how the results contributed to the resolving the broad issues that you outlined at the beginning of the Introduction.

Visual Items and extra information

Here, I briefly outline some of the main considerations for the use of figures, tables, boxes ans suplementary online material


Most people find it easier to read figures than tables. Figures therefore are an effective means of summarizing key findings. A convention is to describe the overall pattern in your findings in the text, and then refer to a figure for details.

For example, you might write: “Species composition changes systematically along a gradient of land use intensity (Fig. 2)” – and Figure 2 would then show the details of this gradient.

Or you might write: “Large companies often had explicit policies in place to enhance sustainability (Fig. 3a). By contrast, many small companies had policies for various aspects of sustainability (e.g. governing social issues) but these were rarely combined in an explicit framework for sustainability (Fig. 3b).”

In both of these examples, the text tells you the basics, and you can look at the figures to find out more.

Here’s a couple of bad examples: “Figure 2 shows changes in species composition” – Figure 2 should not be the central part of the sentence, but something referred to at the end. “Large companies had different ways of dealing with sustainability than small companies (Figure 3)” – this doesn’t tell you enough about what those different ways actually were. In both cases, there is too little emphasis on writing good text to go with a good figure.

There are a few things to consider when preparing the figures themselves:

  • Use simple fonts (sans serif) and avoid ‘graph clutter’. So, for example, Times New Roman is not a good font for graphs, whereas Arial is a good font. Similarly, three-dimensional bar charts like Microsoft Excel can produce are full of graph clutter. Keep things minimal – whatever doesn’t contain information shouldn’t be in the graph.
  • Keep your things black and white, unless you are writing for a journal that uses colour figures (and note those often cost extra!)
  • Keep in mind the size of your figures. Make your figures as small as possible, and make sure that the fonts are readable in the size the figure will ultimately be printed in. Also, think about how many columns the table should take up. Journals tend to arrange their text in one, two or three columns. If you write for a two-column journal, it makes sense to come up with a figure that either takes up one column in width, or two columns in width.

Finally, figures need to have informative legends. Good legends are often quite long. Try to write legends, which (together with the figure) can pretty much stand alone. Legends that are very short often contain too little information. Especially for somebody who reads your paper quickly, it is very useful if your legend contains quite a lot of information. Note that legends for figures are placed below the figure. In a manuscript, they typically go on a separate page from the figures themselves.

Possible exercises:

  • Initially, don’t worry about software. Draw by hand what you want your figure to look like.
  • Then ask yourself if you can implement this figure yourself, or if you would need additional technical skills, or if there is a person you can ask to help with the figure.
  • If your ideal figure can’t be done easily, ask yourself if there is a slightly simpler version that you can do, which is almost as good. Don’t waste too much time on ‘the perfect figure’!


Tables are an effective means to summarise summary information, both numerical and otherwise. When you prepare a table, also consider the alternative – figures. Figures tend to take up more space for the same information, but some patterns can be summarized much more effectively in pictures than in a table. But especially when the data you are trying to visualize is quite simple, tables are sometimes almost quite as good as figures, but take up a lot less space.

Once you have decided a table is what you want, think about how to present your information within the table.

Most journals will allow multiple nested column names, but otherwise, note that the formatting in published tables is usually pretty minimal. So, while in your word processor you can use different types of highlighting or colours, this is not the case in most journals. The best is to format your table using only horizontal lines (like in published tables) – and then make sure the information is clearly understandable.

Keep in mind the size of your tables. While you can fit a lot into a table in your word processor, journals typically have less space – so make your tables as small as possible. Also, you might want to think about how many columns the table should take up. Journals tend to arrange their text in one, two or three columns. If you write for a two-column journal, it makes sense to come up with a table format that either takes up one column in width, or two columns in width.

Finally, tables need to have informative legends. Good legends are often quite long. Try to write legends, which (together with the table) can pretty much stand alone. Legends that are very short often contain too little information. Especially for somebody who reads your paper quickly, it is very useful if your legend contains quite a lot of information. Note that table legends go above the tables (unlike figure legends).


Not many journals allow boxes. In case you write for a journal that does have boxes, they are really useful for a few things.

Examples. A classic use of boxes is to provide short case studies or examples of a concept, theory or situation that you outline in the text.

Additional depth. If there is something you find exciting, but it’s not really central to your argument, a box can be good for that.

Cross-linkages throughout your text. Sometimes boxes can help to link different sections of a paper in the one point. For example, if you have several sections that build different parts of an overall argument or theory, a box can be an effective way to summarise how those parts come together.

If your journal does allow boxes, check carefully how they are formatted. Some contains figures and tables as well, others include just text. Also check the typical word count in a box in your target journal; this varies widely, depending on the page layout of the particular journal.

Suplementary online material

Supplementary online material is now being used by many journals for things that will only be of central interest to a small number of readers, but nevertheless should not be left out. It is essentially what used to be appendices. Supplementary material often can be several additional pages of text, and sometimes up to 10 additional figures. Different journals have different restrictions on this.

Typical items that can go into the supplementary material are things such as (1) in-depth methods, (2) additional results, (3) further justification for a particular approach, or (4) photographs related to the study. In all cases, the main text needs to refer to the supplementary material (e.g. “see supplementary online material for further details”, or by referring to “Fig. S1”). It is also a good idea to summarise briefly the supplementary material in the text. For example “A null model was used to test whether communities were randomly assembled (see supplementary online information for details)”, or “Additional scenarios were explored and showed broadly similar patterns (Figures S1-S5)”, or “Our study assumes that communities are influenced by institutions at multiple scales, including local, state and federal governments (Fig. S1)”.

Supplementary online material should not be seen as a deposit for everything that didn’t otherwise fit. Rules of prioritizing your material and fitting the content to the prescribed length still apply. Rather, supplementary material is for things that are necessary or informative, but not of central interest to the majority of readers.

Possible exercises:

  • Collect supplementary material from a few different journals that are relevant to you. Get a feeling for what people put into their supplementary material. Which kind of format is useful for you, the reader, to understand how the material is organised?

The process of publishing

It’s helpful to understand the process of publishing. Writing a good paper is probably the most important, but it’s also helpful to understand some of the peripheral issues surrounding the process of publishing.

Formatting your manuscript

Most journals expect that your manuscript is formatted in the following way:

First page is a cover page, containing title, keywords, author names and institutional affiliations, plus the email address of the corresponding author.

Second page is where the abstract should start.

Third page onwards comes the manuscript text, WITHOUT any tables, figures, or table or figure legends.

At the end of the text, list your Acknowledgements, followed by References.

On a new page after the references, list your figure legends.

On separate new pages after that, present your tables. The table legends go directly above the tables.

On separate new pages after that, list your figures, one figure per page. Some journals also want you to upload the figure files separately (in which case you need to follow their instructions file type and resolution).

In addition, your entire manuscript should be double-space. It is also helpful if you add line numbers, for example every fifth line, consecutively throughout the entire manuscript. This makes it much easier for referees to comment on your work.

Should I care about impact factors?

If you find your work worthwhile, you will have an inherent interest that many people read it. This means you need to place your paper in the right journal (e.g. matching the topic with what else is in the journal), and preferably in a widely read journal. The impact factor of a journal is a measure of how widely cited recent articles from the particular journal are. In this sense, it does matter. In addition, a high impact factor publication will look better on your CV than a low impact one.

Although the impact factor matters, the most important thing is that you publish in a journal that is highly visible to your target audience. A certain journal might have a slightly lower impact factor than another one, but if it is still among the ‘standard leading journals’ in your field, it’s not worthwhile worrying too much about the impact factor.

Ideally, think of the impact factor as a means – to measure if a journal is widely read. Publishing in a good journal, by contrast, is the ultimate objective. The impact factor will help you make an informed decision where to submit your paper, but it should not be the only criterion.

Cover letter

Cover letters have become quite important in some academic disciplines. Ask around in your discipline how common it is for papers to be rejected without peer review. If that is a common phenomenon, this suggests that editors themselves have a lot of power – they might decide that a lot of papers don’t even go for peer review. Rejections without peer review tend to be quick, but are frustrating because you get virtually no feedback why your paper was rejected.

Cover letters are intended to convince the editors that they should send the paper out for peer review. If you are dealing with a journal that sends almost everything out for peer review, a minimalistic cover letter is fine. But if you are dealing with journals that reject a lot of work without peer review, you need to explain why your particular paper is worth considering.

In a (more than minimalistic) cover letter, you might want to explain:

  • That you’re submitting the work to the journal for consideration for publication;
  • Why the work is new, exciting and important;
  • Why you think the particular journal is suitable;
  • Any additional information that is important to the particular journal.

Many journals have additional specific requirements for their cover letters. For example, they may require a statement that all co-authors have read the manuscript and agree to its submission; or they may ask you to suggest potential referees. Check the journal’s instructions for authors to make sure you follow the instructions.

In some cases, it’s a good idea to contact the editor by email before you submit your paper to ask if the paper might be suitable for a journal. This is not something that all journals are happy with, but some actually encourage such pre-submission enquiries. If you do write a pre-submission enquiry, it’s a good idea to keep it brief; send the abstract, and maybe a key figure; and explain why you think it might be suitable for a particular journal. Make sure you specifically ask whether the editor thinks the work may be suitable for the particular journal. Sometimes editors get back to you with a helpful hint: for example, they might say that your work sounds interesting, but needs to have a clearer international relevance for it to be considered. If that’s the case, you already know that there’s something you should work on some more before even submitting to that particular journal.

Personally, I think a cover letter should be about one page, and if possible, it looks better if you write it on the official letterhead of your institution.

Open access

For many journals, you need a subscription to read the articles. This costs money, and it is one of the key reasons why people in poor countries do not have very good access to primary scientific literature.

In an effort to make scientific literature available more widely, an increasing number of journals offers an ‘open access’ option, and some journals are entirely open access. Both typically require the author to pay money (out of research grants, typically) when a paper is accepted for publication. However, both types of model typically also have an option that stipulates that if you truly cannot afford the so-called page charges, they will still take your paper.

Some people have speculated that making your paper open access might increase your citation rates. Whether that’s true or not, it is certainly easier to access a paper if it is open access. I suggest whether you choose an open access journal or not should depend on who you think your main audience is (and hence what the benefits of open access are), and whether you can find a journal that has a good reputation and is open access in your field. My personal view is this: If you have two equally prestigious journals, one allows you to make your paper open access while the other does not, and you have funds to pay for this – then you may as well make your paper open access because you have nothing to lose, and you might gain extra readers. If your work is directly relevant to poor countries or to institutions who typically do not have access to journals (e.g. in the business sector), you might place an even higher priority on publishing your work open access.

Revising a paper after review

When you first write your paper, of course you hope for it to be accepted. But reality is that many papers are rejected, and those of those that are not, many require revisions, often major. As an example, leading international journals in applied ecology now reject 80% of papers – if you look at leading journals like Nature, the percentage of rejections is much higher still. Given this situation, being asked to submit a revised version, even if the revisions are major, is often a ‘success’. How can you make sure that when you are asked to submit a revised version, you maximize the chances of your revised paper being accepted?

The first general rule is that you take the criticisms of all reviewers seriously. As the author, it’s your job to communicate your ideas clearly. Even in cases where the criticism you received seems unjustified, ask yourself “What can I do better so that this same reviewer will be more likely to buy into my argument next time around?” Even if you are completely satisfied that your argument holds, ask yourself if you can do a better job of communicating it clearly.

You then need to go through the comments by the editor and all reviewers one by one, and address every single one of them. Broadly speaking, you have three options for how to respond to a particular criticism:

  1. When, upon reflection, you agree with the reviewer: You implement a change that does what the reviewer asked for. For example, she may have asked for additional explanation, an additional reference, or the complete re-write of a section, including a different conclusion.
  2. When, despite reflection, you disagree with the reviewer: You might not want to implement some particular suggestions, especially if they relate to subjective matters, which you simply feel differently about. (For minor suggestions, I suggest just do as the reviewers suggest, because there’s nothing to lose; but if you truly disagree, just implementing the suggestions is not a satisfactory solution.) Still, in this case, you can often communicate more clearly what your position on the issue is. So re-write your text in a way that might convince the reviewer of your position, lends more weight to your argument, or is simply easier to follow. In this case, you’re still making the same argument as before, but you explain it more clearly.
  3. When, despite reflection, you think a reviewer is simply wrong: In this case, you change nothing in the text, but you explain in your cover letter to editor why you did not make the change.

As a general rule of thumb, if you want your paper to be accepted, and especially if it goes back to the same reviewers, your response to most criticisms needs to be type 1, followed by type 2 – only rarely will you get away with response type 3 above.

Your revision ultimately consists of the changed manuscript, as well as a detailed letter outlining point by point your responses to the reviewers. Put yourself in the position of the editor: she wants to understand as easily as possible what you changed or did not change, and why. Making this job as easy as possible for the editor is what your response letter needs to do.

The following format works well:

  • An introductory cover note, which follows a logic like this:
    • Thank you for your decision letter from xyz.
    • We greatly appreciate the constructive comments.
    • We have now addressed these comments, and this has strengthened the paper.
    • On the following pages, we outline point by point responses to the comments by the reviewers (and the associate editor, if applicable).
    • We hope our revised version will be received favourably and look forward to hearing from you in the near future.
  • On the following pages, detailed point by point responses, to each of the comments, in a format as follows:
    • Paste in a given reviewer’s comment in italics
    • Then explain how you have addressed it, e.g.:
      • “The authors assert that biodiversity is declining in their study area, but they do not substantiate this claim. At the very least, this statement requires a reference.”
        Response: We appreciate this concern and have now added a reference (Smith et al. XXX; see page 4 of the revised manuscript)
      • “The methods used on page 6 are unclear. Which type of regression model was used?”
        Response: In the first paragraph of page 6 of the original manuscript, we had actually specified that we used generalized linear models. Arguably, our statement was hidden among other information, which is why the reviewer may have missed it. We have now moved our statement to a more prominent position at the beginning of the section on ‘Data analysis’ (page 5 of the revised manuscript). This should make it more obvious which methods we used.

Don’t be surprised that response letters sometimes end up very long! Still, it is better to explain too carefully what you have changed and why, than not carefully enough. The objective of a response letter is to show the editor that you are taking the review process seriously, and that you have put in a very genuine effort to address the comments, including the difficult ones. It’s a good idea to approach this process from a position of intellectual humility, while also making sure that you stay true to your core argument.

Topics not covered

There are additional topics that could be listed here, but which for now, I have not included. Here is a short summary of some of those issues.

Statement of author contributions: a short summary saying who did what, such as in Proceedings of the National Academy of Sciences USA (PNAS).

Writing for Nature or Science, or other journals with unusual formats: Nature and Science often mix results and discussions, and journals such as PNAS put the methods in small print at the back of the paper. To familiarize yourself with how to write for these journals, it’s best to check a number of recent articles, as well as read the author instructions in detail.

Reporting statistics: Different disciplines have slightly different conventions on how to report statistics, such as significance levels or which methods were used. It’s best to check other recent papers from the particular journal you are submitting your manuscript to for examples.

Links & further reading

Here are some other sources that you might find helpful.

For social scientists

The author of this entry is Joern Fisher. This article was originally published on Joern Fischer's blog.