How a research paper is built

Research papers can look forbidding, dense with methods and numbers. But most of them follow the same underlying plan, and once that plan is familiar a paper becomes much easier to read, and much easier to judge.

The IMRaD structure

Wikipedia's article on IMRaD describes it as a common organisational structure for scientific writing: Introduction, Methods, Results and Discussion. The article states that IMRaD is the most prominent norm for the structure of a scientific journal article of the original research type. It frames each part as the answer to a question. The introduction asks why the study was undertaken and what question or hypothesis it addressed. The methods ask when, where and how the study was done and with what materials. The results ask what answer was found. The discussion asks what the answer might imply and why it matters.

The same article describes a wineglass model of the structure: wide at the top, where the introduction starts from the general field; narrow in the middle, where methods and results deal with the specific study; and wide again at the bottom, where the discussion returns to the broader picture. The shape is a helpful way to remember where to look for what.

Each section, and what it is for

Title and abstract

Before the four main sections come the title and the abstract. The IMRaD article notes that a brief abstract is usually required and should work as a stand-alone text. The abstract is what most people read, and often all they read, so it is written to state the question, the approach, the main result and the conclusion in a small space.

Introduction

The introduction sets the scene. It reviews what is already known, identifies a gap or a problem, and ends by stating what this study set out to do. A good introduction lets a reader judge whether the question matters and whether the authors have understood the earlier work on which they are building.

Methods

The methods section is the recipe. It describes the materials, organisms or samples used, the procedures followed, the equipment, and the statistical analysis. Its purpose is to let another researcher repeat the work. It is often the least read section and the most important one for judging a study: weak methods cannot produce strong results, however striking the numbers.

Results

The results section reports what was found, usually with figures and tables, and ideally without interpretation. Well-written results sections lead the reader through the data in a logical sequence, each paragraph pointing to the figure or table that supports it.

Discussion

The discussion interprets the results. It compares them with earlier work, explains what they might mean, acknowledges limitations and suggests what should be done next. This is where authors are most tempted to claim more than the data support, so it rewards a sceptical reading.

Figures and tables

Figures and tables carry much of a paper's evidence. A graph can show a trend, a spread of values, or a comparison between groups at a glance; a table can set out many exact values side by side. Each should have a caption that explains what is shown well enough to be understood without reading the text. When reading, it pays to look closely at the axes, the units, the number of samples behind each point and whether any measure of variation is shown. A difference that looks dramatic on a graph with a truncated axis may be tiny in reality.

References

The reference list shows what the paper is built on. Every claim about earlier work should be traceable to a cited source, and a reader can follow those references to check that they say what the paper says they say. The quality of a reference list, whether it cites the main earlier studies, including those that disagree, and whether it relies on reliable sources, says a good deal about the care taken in the paper as a whole.

Reading a paper efficiently

Few experienced readers go from the first word to the last. A common approach runs in passes. First, read the title and abstract to decide whether the paper is relevant. Second, look at the figures and tables with their captions, which often tell the core story. Third, read the introduction's final paragraph, where the aim is stated, and the discussion's first paragraph, where the main finding is usually summarised. Only then, if the paper matters, read the methods and results in full, with the questions from the first passes in mind.

Throughout, it helps to keep asking three questions. What exactly did they do? What exactly did they find? And does the conclusion follow from the finding, or does it go further? Those questions, applied patiently, are most of what critical reading of science involves.

Other kinds of paper

Not every paper follows IMRaD. Review articles survey a field and are usually organised by topic. Short communications compress the structure. Methods papers focus on a technique. Knowing which kind of paper you are reading helps set expectations: a review, like those cited throughout this guide, weighs many studies, while an original research paper reports one.