How to Write a Scientific Paper
Academia is full of hidden curriculum. These are the skills and lessons we’re meant to learn and may be evaluated on, but rarely receive formal training in. How to write a scientific research paper is one of them. The scientific paper is a genre of academic writing that comes with its own set of standards and “rules.” But when writing papers, proposals, or theses, most people have to learn through mimicry and trial and error. Knowing how to ask for feedback can be tricky itself, and mentorship is not evenly distributed.
This piece aims to make these literary conventions more obvious and accessible, so that writing about your science may be easier and more enjoyable.
Broad Themes to Consider
Intended Audience and the Expectation of Critique: A scientific paper is first and foremost a communication tool, and with any form of communication, it’s important to consider who your audience is.
The purpose of a paper is to share newly discovered information with the world, so we commonly refer to our audience as “the broader scientific community.” But depending on the topic you study or the journal you’re writing for, you may find yourself writing to a more specific group of scientists in your niche field. Knowing this will help you determine the appropriate tone and level of detail for your writing.
On the other hand, many authors actually find themselves writing for an even more specific audience: the journal editor and panel of reviewers. Maybe you also have a particular colleague or competitor in the field that you intend to impress. But it’s the opinions of these few individuals that can really shape the way you write. A research manuscript is not only explaining its claims, but also justifying and defending them to convince reviewers to accept the paper. And even when the paper does reach the broader scientific community, they too are trained to read your paper with a critical eye. Readers look for gaps in your logic and evidence, which “better” science may be able to fill. This is how we learn and grow and advance our fields, but it also has direct effects on our writing style. We’ll explore how this expectation of critique shapes the writing conventions in each section of a scientific paper.
Neutral vs. Promotional: The cultural values of the scientific community are reflected in the way we write, and let’s just say, we contain multitudes.
On the one hand, we intend to honor the scientific method and uphold its sense of objectivity and neutrality. We present our data as merely a reflection of reality, not biased by or indebted to the person who collected it. We want our papers to uphold these scientific ideals, and this seeps into our writing style. The use of passive voice (“X was done”) in Methods sections is a common example of distancing the science from the scientist (more discussion on this later).
On the other hand, the intense competition and critique within the scientific community requires authors to perform a certain level of self-promotion. Active voice (“We did x”) can be used in the Results section when an author wants to subliminally take credit for an impressive experiment. Introduction and Discussion sections are especially likely to mention the innovation, impact, elegance, and novelty of the study. The tools are state-of-the-art, the datasets are unprecedented, and the findings will transform our understanding. Our papers are thus never neutral reports of neutral observations.
It is important to note that there is no “correct” way to approach this dynamic. Each author must strike their own balance for every paper. For example, when stating a claim, scientists must walk a fine line between acknowledging the technical and logical limitations of their study, while also asserting their findings are real and important. Authors must comment on how their study advances the work of other scientists, without critiquing previous research enough to cause a controversy (unless contradicting an existing idea is a main goal or outcome of your study). Even when selecting references, you want to cite enough similar findings to support your conclusions, but not so many that your results no longer seem ground-breaking. All of this takes practice, and seeking feedback from your mentors and peers is highly encouraged.
Sections of a Scientific Paper
Title: The article title succinctly (in 10-12 words) answers the question, “What did this project study and/or find?” This can either be a direct statement of the main finding, or a brief description of the topic that may hint at the findings. The latter could make the reader more curious and let them read the paper with an open mind, as opposed to being told what to believe from the start. In the end, though, the main goal is to use keywords that will attract as many people in the intended audience as possible.
Abstract: The abstract is a summary of the paper, and often determines if and how thoroughly a reader will consume the rest of the piece. It begins with factual statements that orient the reader to the topic and its significance. It then states the main question of the study, and references the methods used to achieve this goal. The last sentences summarize the main two or three findings and their implications or impact on the field.
Introduction: The main goals of an Introduction are to explain the main research question and why it is important to study, review what is known and unknown, and state how the current study advances this line of work. The structure and length are flexible, but may be shaped by the journal’s preferences. Deciding whether to start broad or jump right into your specific focus will depend on how familiar or interested you expect the audience to be with your research topic. The last paragraphs are usually either a preview of the methods and hypotheses of the study, or a summary of the findings.
Methods: This section describes the resources, methods, and analyses used within the study. While your results and interpretation of the data may evolve throughout the writing process, the methods are less likely to change, so many authors find it useful to write this section first.
In theory, this section is meant to function as a summary protocol that other scientists could use to replicate the experiments themselves. For this reason, it is typically written in past tense and passive voice, so statements like “we did x” become “x was done.” This intentionally removes humans from the scientific process and promotes the illusion of objectivity. The sentences are focused on the actions that were done and not the people who did them.
In practice, however, a Methods section alone can rarely be used as a real protocol. Instead, it serves as another tool to defend your science against critique. Over time, authors learn what experimental details are conventionally included for the types of experiments they do, but beyond that, the choice of how much detail to use can often come down to what will be most advantageous for you. Authors typically include enough detail to a) help the reader understand the data, b) convey that the experiments were done thoroughly and responsibly, and c) not reveal so much that a critical reader could find something to pick on. You want this section to help you and not hurt you. Methods that were either developed by another group or have already been thoroughly explained in another paper are not fully described and instead rely on a citation.
Results: This is the bulk of the paper where the findings are demonstrated through text and data figures. Figures use a tiered structure which is then mirrored by the text: individual plots may be grouped to support a finding, and multiple findings are grouped within a figure to convey one central message. Each subheading in the Results section thus typically represents one figure (i.e. Figure 1), and states its overall message (though two closely related figures may be grouped under a single subheading). Each paragraph under the subheading focuses on one of the findings, which can be supported by multiple plots (i.e. Figures 1a-c).
The paragraphs take on the following structure: “To learn x, we did y, and found z.” Guiding the reader through the logic of the experiments is key. The amount of detail used to describe or justify the experiments is left to the author’s discretion, but we usually assume the readers have enough technical knowledge to understand why the chosen methods will answer the current question (this can be a challenge for less experienced readers who are still learning about research tools).
Any sentence reporting a result ends with the associated figure label. P-values or other statistical measures may be referenced (i.e. “p<0.001, signed-rank test”), but specific measurements and sample numbers can be saved for the figure legends. These figure captions are written as descriptions of each plot to help the reader process the data, but they do not say what the figure is meant to prove. It’s the text’s responsibility to state the findings and suggest how the data should be interpreted.
Discussion: This section serves as the conclusion of the piece, and gives the author the most freedom over content and structure. It’s used to interpret the main findings, emphasize their significance, and anticipate the reader’s comments or concerns.
Each paragraph typically focuses on one main finding, and guides the reader’s interpretation of the data more than the Results section does. This is the space to demonstrate your critical thinking skills. You can relate your work to a broader theory, offer alternative explanations, and attempt to explain unexpected results. This section also directly addresses how your results relate to or contrast with the results of previous studies. Discussing contrasting findings allows you to either highlight an important distinction or advantage of your study, or preemptively state your defense if the other paper is significant enough to make someone question your work. It is important to state the limitations of your research and suggest follow up experiments, but reiterate that your current results are still valid. The final paragraph may highlight the novelty of your study and the types of future research it will inspire.
Citations: No science is done in a vacuum (well, some is). A paper should function as a stand-alone piece that tells a complete story, but it also contributes to a larger conversation and investigation taking place in the field. Authors must address the position of the current study relative to the previous work of their own lab, and the work of other labs.
References to other publications can be included in any section of the article, but are especially common in the Introduction and Discussion sections. Citations are used for several reasons: a) to support background information, b) to reference a methodology developed or well-described in another paper, or c) to mention a study whose findings are similar or different from yours.
When selecting citations, authors tend to choose papers from top tier journals or well-known labs, under the assumption that readers might trust these more or associate your work with them. We’re also often taught to cite the earliest paper that convincingly demonstrates the point you’re trying to make, but papers using modern methods may be preferred. Scientists are looking to be cited, so this practice can become a very social and political endeavor, wherein colleagues tend to cite each other’s work and well-known labs become even more well-known (a.k.a. funded). Given this context, you may develop your own strategy and values around selecting papers to cite.
Terminology: While some terms become established within a field, other terminology is often developed within the lab itself. This might include labels for experimental or control groups, names of specific tasks or assays, or new terms assigned to things discovered during the study. These phrases may be accepted as standard terms within your research group, but may not be as familiar or intuitive to the reader. Try to define your language as clearly as possible so the reader can easily interpret your data and relate it to the work of other research groups.
Journal Specificity: It’s important to be aware that certain aspects of your article’s organization and text will be determined by the requirements of the journal you submit to. For example, journals may have preferences for the number of figures in each article. Authors must make judgments about which data to include and how to group it based on these requirements, as opposed to crafting your ideal scientific narrative. In general, papers published in higher profile journals tend to be more succinct, self-promotional, and in places, use more plain or broad language to attract a wider audience to the topic.
Most scientists learn these concepts and conventions through the (at times high-stakes) system of trial and error. By demystifying this process, we help more people to feel a sense of belonging in the field, and allow them to focus their energy on the research itself. That being said, everyone is encouraged to look for additional conventions specific to their discipline, and define their own writing practice and personal preferences over time. And if you’ve learned anything from this article or your own writing experience, remember to share this knowledge with others. Happy writing!