What a journal article actually does
A journal article is a story you tell about your research. Not a story in the classic sense, but in the sense that it goes beyond simply presenting data. You take your reader on a journey.
That journey isn’t just: I did this, and I did that. No, your reader wants to know why you went left here instead of right. To speak in metaphors: Why did you go to the beach instead of the mountains? Why did you take the train instead of the plane?
We can call this journey the ‘scientific narrative’. This narrative forms the backbone of your journal article — without one, the thing falls apart. This narrative must adhere to tight logic, because you are dealing with some of the most critical readers on the planet. If your logic isn’t clear, they’ll question the validity of your research (and probably, so will you, if not your inner critic).
What does that logic look like?
The scientific narrative requires that every step in the research follows logically on the previous one. What that looks like, depends on the context, so let’s go through it section by section.
Scientific papers often feel difficult to write — not because the IMRaD format (Introduction, Methods, Results, and Discussion) is complicated, but because the logic behind it is hidden. Once you understand the questions that drive each section, the structure becomes much easier to work with.
At the heart of a paper are three levels of questions:
🌐Broad, general question(s)
- Concerned with broad, often societal issues; usually one question, found2 at the start of the intro.
- Asks stuff like: How can we improve this (complex) process/situation/…?
- Is so broad that it can’t be fully answered in the paper (research will address a small subsection of the broad question), but will often be partially addressed in the Implications or Conclusions section.
🔍Research questions**
- This is the stuff you actually wanted to study. Will ask things like: How does x affect y? To what extent does a cause b?
- Usually (implicitly) asked2 in the middle/final part of the Introduction.
- The research questions are extensively addressed in the Discussion, using the results and literature.
📏Specific, measurable questions**
- Highly specific, measurable questions that focus on the actual (often quantitative) variables. Will ask things like: To what extent does x_study affect y_subset, given a?
- These questions are implicitly stated2 in the Methods and can be considered subquestions of the research questions.
- Answered in the Results section using the data.
Zooming in and out
These layers of questions create a hidden structure that mirrors the famous hourglass shape of IMRaD:
- Zoom in: from broad, general questions (Introduction) → to focused research questions (Introduction/Discussion) → to specific, measurable questions (Methods/Results).
- Zoom out: from specific answers (Results) → back to broader implications (Discussion/Conclusion).
Once you see this structure, IMRaD stops being a rigid template and becomes a logical flow of questions and answers. That’s the hidden scaffolding that makes papers coherent — and much easier to write.
Introduction
The purpose of the Introduction is to convince the reader that this research is worth publishing and/or reading. You must convince them of three things:
- This research is important, because it addresses an important societal issue or a major scientific unknown.
- This research is relevant, because it addresses one ore more knowledge gaps that are crucial that societal issue/scientific unknown.
- This research is appropriate, because is addresses the right research questions using the right approach.
These three reasons build on each other, so that the Introduction forms one clear argument that answers the question: Why does this research matter?
Methods
Most of the Methods don’t actually do that much explicit reasoning, and are simply a description of how the research was carried out, but there is one exception: the methods rationale.
You might not have heard this term, as it is rarely called by a specific name in STEM fields. However, you can usually observe the methods rationale in the first paragraph(s) of the Methods.
The rationale explains how the methods address the research questions, and why these specific populations/variables/models/locations/methodologies/datasets/etc. were used. This is especially important when using novel or unusual methods, where reviewers might be extra critical or even skeptical of your approach (in other words, if it is glaringly obvious why a particular method was used to people who read this journal, there is no need to write about it).
Results
The Results section is a bit of an odd one out. It’s job is mainly to report results in the most transparent manner possible. However, there I’d like to add a caveat: most data needs to be interpreted to form results, which requires reasoning. For example: statistical reasoning, benchmarking, model validation, etc. However, this type of reasoning is so extremely niche-specific, that it is not relevant for the purposes of this guide (which is on writing scientific journal articles).
Discussion
The purpose of the Discussion is to interpret the findings* and draw conclusions about them, which can be really fun — with the emphasis on can. However, simply drawing conclusions is not enough. You must also convince the reader that these conclusions are the most reasonable interpretation of the data.
In other words, you need to argue that your conclusions are solid. Now, this is hard. Because there is a chance that your conclusions might be wrong — after all, they are only the most likely interpretation of the findings. There no absolute certainty that your conclusions are correct!
This doesn’t mean you can’t write an airtight Discussion — but more on this in Stage 3 of this journal article roadmap (see below).
*This is different from interpreting the data_, which happens in the results section._