4. Elements of a paper

Scientific writing comes with certain structure-related constraints that are good to know. There is structure-related nomenclature that you should be aware of while you should also understand so-called ILMRaD structure (Botchkarev, 2025). Besides these higher level concepts, I will cover next how to work on the structure of your paper, how to construct your paragraphs, and how to develop a narrative for your paper to make it easier to read.

Depending on your template and where you aim to publish, there are several keywords you should understand and I have listed them before:

Chapter: is the highest level structural concept especially in a long work like a book or a thesis. Occasionally it may be missing entirely from a work and then highest level concept is called a section instead.
Section: sits one level below a chapter although in short works, such as conference or journal papers, it might be the highest level concept have available.
Subsection: is contained within a section and if the highest level concept of your work is a section, you will use subsections likely a lot.
Subsubsection: is contained within a subsection and usually something has gone wrong if you have to use it. Some journals have disallowed its usage for this reason. Another option is that your work has big enough scope to warrant the usage of subsubsection.
Paragraph: allows you to give small titles to paragraphs, but unlike others, it does not show up in the table of contents typically. When used appropriately, paragraph-level titles can give an extra level of readability for your paper and it can be effective even in short papers this way.

After you know the nomenclature of your paper, it is important to use it correctly within the paper itself as you refer to the concepts. Typically you use title-case (i.e., "In Section 4, we discuss..."¹) these concepts as you write to make them easily visible to the reader. Most likely your journal has a related recommendation although this seems like a common practice. Note that I will use the term chapter to refer to the highest level concept within a structure although within a journal context your would refer to chapters as sections instead so adjust the term to fit your needs.

4.2 Structuring a paper#

In Ecarnot et al. (2015), the authors discuss the so-called IMRaD (Introduction, Methods, Results, and Discussion) model for structuring a paper. Botchkarev (2025) expanded the model further by adding Literature Review to it as an element and arrived at ILMRaD². Botchkarev covers many other variants of IMRaD in his work but to keep it simple, I cover ILMRaD expanded with one more chapter type in this context. You can see this structure in many papers and it is a good one to follow³. Occasionally, you may have to tune the basic structure to fit your purposes and constraints given by your target publication. When surveying where to send a potential paper, it is a good idea to check out what kind of structure has been used in papers published earlier since that gives you idea of the preference of the publication and helps you to adapt the structure of your work accordingly. For example, in short surveys I explain that I am doing a survey in the introduction and then replace the Methods and Results sections with survey-specific chapters. In addition to the standard ILMRaD structure, I also include a Conclusion chapter at the end of the paper to recap my findings so I have arrived at what you might call ILMRaDC.

The general rule for structuring is that you approach the topic from an abstract angle to motivate the work and bring the reader to the problem in the Introduction, which you then study through a method before starting to go from concrete to abstract again through Discussion, and finally Conclusion. To give you a better idea of each possible chapter, I have condensed them to the table below.

Chapter	Purpose	Writing approach
Introduction	Bring the reader to the topic	Start from a broader context, link to related research, and zoom to your research problem before explaining your approach and the structure
Literature Review	Cover key topics related to your research so that someone not well versed with the topic can understand your work	Usually you should write background only after you have done your research so you know what you should cover. Occasionally it makes sense to split literature review into multiple small, well-scoped chapters.
Method	Explain how you approached your research problem on a methodical level in case your method was complex	Cover the method at such level that another researcher has no trouble reproducing your research
Results	Capture your main findings if the paper warrants it (i.e., there are quantitative or qualitative results to explain). Ecarnot et al. (2015) cover well how to approach writing the results chapter.	Besides text, use visual aids, such as graphs or tables, to condense information to make it easier to digest
Discussion	Reflect on your key observations	Compared to other chapters, discussion is the most free by its form. In discussion you can consider how your work fits in the broader context.
Conclusion	Bring the reader out of the topic	Remind the reader of your research problem, cover your key findings, and suggest what kind of further research should be done

Chapter types, their purposes and suggested writing approaches as ILMRaDC model partially adapted from (Botchkarev, 2025). Note that occasionally you can skip some of these if you don't need them, especially the middle ones.

The same structure, from abstract to concrete and back to abstract, is often visible at the section and paragraph levels. It is a good practice to begin each section with a brief introduction to bring the reader to the topic. Consider using short conclusion sections for additional clarity especially in longer texts to recap the main points for readers in a hurry.

4.3 How to work on a structure#

A good way to work with the overall structure of your article is to consider the table of contents and how it reads. While looking at your table of contents, it is good to consider hierarchy, keywords, structuring, and overall flow. Occasionally you might notice that something feels imbalanced. For example, there might be sections that feel too short, too long, or even too detailed. That is when it is a good time to see if you can perhaps split or merge something or eliminate subsections in favor of simpler paragraphs. Some of these decisions depend on the scope of your work and partially it is a question of style since some authors prefer shorter and more specific sections while others prefer to writer in longer format. The critical thing is to remember that the structure of a paper should make it easy for the reader to understand the main elements of your work. Although there are some rules on how to set up a structure for your work, as hinted above, I have found the following techniques helpful:

Consider the abstraction level of your titles. Occasionally, more concrete is better especially at the lower levels of the title hierarchy, and sometimes, I have replaced the conventional background section with a more specific title or even split it into a few shorter sections.
In case you notice that a section has deep nesting, that might imply you may have to split up something.
If a section might feel overly long in the first draft, consider if you should split it up somehow.
Occasionally, a certain order makes more sense than some other. Usual hints of this are the need to explain a concept before other explanations make sense. Especially early on, many basic definitions may be needed, and that is a good chance to introduce abbreviations to make writing and reading easier.

Depending on your writing style, you may benefit from planning a structure upfront. You should at least put the publication specific chapters into place early since you know you will have to write those eventually. It is a good idea to leave space between the beginning and ending chapters for content to appear especially if you work with multiple separate files as I discuss in A. Getting started with LaTeX appendix.

4.4 Why to develop a narrative#

Although scientific writing is about bringing together known facts, it is also about developing a narrative.

While scientific papers are not stories, they tend to have a point of view and certain key content. As you expose new content and ideas to your reader, you are bringing new actors to the play so remember to use them as you develop your ideas. Especially early parts of a paper establish the stage and fill in details required to understand what is to come. Specifically early portion of your paper sets up the context and pulls the reader into your world. Once you have established the stage, it is natural for you to explain your findings, before considering their implications, and ending the play with a cliffhanger as there is always more to research.

4.5 How to write an abstract for your paper#

Abstracts are used to communicate the contents of a scientific paper to other scientists effectively. Typically abstracts tend to be short and up to a couple of hundred words. The purpose of an abstract is to capture the essential information from a paper and act as a marketing text for it. Apart from the title, it is often the first thing read, and it already acts as an initial filter as readers decide if they should keep reading or skip the paper. Based on this background, you should effectively introduce the topic to the reader, motivate your problem, cover your study method, and explain the main findings. I have described a basic recipe for covering the necessary bases in your abstract below:

Describe the context of your research
Explain why it is important to study your topic
Describe the research problem
Explain how the research problem was approached by covering your research methods
Cover the main findings of your research

4.6 How to write a Discussion chapter#

The discussion chapter of your paper is the one where you can reflect on your findings in a relatively free form. I use the following structure when writing a discussion: 1. brief introduction with a main point or two, 2-n. individual points to discuss, n+1. optionally discussion of limitations depending on the type of the paper.

Introduction phrases out your main argument

The introduction to a discussion is a good point for phrasing the main argument of your paper in a clear manner. For example, you can point out the main impact of the topic you discuss. The introduction is the perfect place for high level commentary that foreshadows your specific discussion points.

Consider the types of your discussion points

I tend to divide my discussion arguments into three categories with a purpose of its own:

Observations
Questions
Contrasts

Observations are perhaps the easiest category out of the three as they are derived from what you learned during your research process. For example, you might have noticed specific benefits, challenges, or opportunities related to your topic and it is good to consider these in detail.

Questions are a good way to consider whether some proposition related to your topic might be true. Questions allow you to interrogate your topic in a broader context and consider for example the implications of your findings. A good way to derive questions is to consider what kind of questions your reader might have in mind when reading your paper and anticipate them.

Contrasts allow you to put your research next to other related topics and discuss with adjacent research while placing your results to a broader context. A good way to do this is to have a title along "Other ways to..." as that lets you bring other perspectives to your paper easily.

Addressing limitations

In a well written scientific publication you should consider the limitations of your work. In other words, you should consider the generalizability of your work and what might have affected this area. For example, it is possible that due to your methods or other factors, your results might hold only under certain cases. It is important to be transparent about the limitations of your study as doing this helps other scientists to frame their studies to address them for example as they help with positioning of any further work.

Addressing validity

Besides limitations of your study, it is a good idea to consider validity. Typically it is a good idea to consider both internal and external validity. By internal validity, I refer to the validity of the test setup itself and how you set things up for your measurements. External validity includes factors beyond your control in case you rely on external vendors or instruments that you do not understand completely. In addition you can consider construct validity, as in what you built for your research, as there may be factors related to whether it was built correctly or not. It is a good idea to consider validity because it lets other researchers know where your results might or might not apply and it could be a nice research topic for someone to try to replicate the results under different constraints.

4.7 How to write a Conclusion chapter#

Compared to Discussion, Conclusion is generally a faster one to write as it does not have to achieve a lot. The purpose of Conclusion is the opposite of Introduction as instead of bringing a reader to the topic, you bring them out of the topic. A typical way to write a Conclusion is to first remind the reader of the context and the problem at hand. After introduction, a simple yet effective approach is to create sections containing your research questions and answer to them based on what you learned. Finally, you should have a section where you discuss open research problems or directions as there might be further research to be done. The main thing to keep in mind is that the conclusion of your work should read as standalone and the reader should not have to read anything else from your paper to understand it. Although this sounds simple, it may take some effort to write a good conclusion as you have to find a good balance between length and amount of detail to include.

4.8 Summary#

Although scientific papers tend to follow a specific kind of standard structure, there is still work to be done as a writer. The general structure tends to go from a broad to specific and then back to broad.

You can expect to write chapters, such as Introduction, Discussion, and Conclusion, always, but the chapters that exist in between tend to vary based on what is needed. It is in these chapters where you can stand out by choosing your titles carefully to communicate the intent. Choosing good titles helps to scope your writing and think about what content is needed and how your work flows from a concept to another.

Scientific writing is not only about following these types of conventions but also developing a narrative that allows you to highlight your key messages to the reader. Papers tend to include some level of repetition on different levels for this reason as you examine the topic from different angles. As a writer, your task is to make it easy for the reader to follow your thinking and understand your key findings.

A common mistake is to use "we" instead of "I" when you are authoring a paper or a thesis alone. ↩
According to Botchkarev (2025), ILMRaD is used commonly in top tier journals over standard IMRaD and therefore we discuss the expanded model here. ↩
Interestingly enough, the basic structure of scientific papers follows the overall structure of progress, promise, and payoff used often in storytelling. Brandon Sanderson goes into greater detail about the topic in his lecture available at https://www.youtube.com/watch?v=-hO7fM9EHU4. ↩