Write the methods section of a research paper
The methods section of a research paper describes the study that was actually done, in enough detail for another investigator to repeat it. It follows a default order, from the study design through ethics, the participants, the conduct of the study and the sample size to the statistics, which each manuscript adapts to its study and its journal. It is written in a descriptive voice and the past tense, and it has its own conventions for materials and for data.
What the section has to achieve
The methods section, sometimes called participants and methods or methods and materials, should describe a logical experimental approach. Fundamentally, it needs to describe three things: the subjects, what was done to them and the statistical methods used. With the results, it is one of the experimental sections of a journal article.
After the first draft, the writer asks whether the section allows reproducibility: whether a trained investigator in the field could replicate the study from the information provided. The section's rhetorical intent is to convey exactness and sufficient detail for another research expert to evaluate and duplicate the study. The methods and results sections should be written in exquisite detail, because if other people cannot reproduce the experiments, the study will not last. They should be clear enough for another scientist to read the methods, perform the same experiments under the same conditions and arrive at the same conclusions, and readers should never have to guess at what the authors are trying to communicate.
A reviewer equates the author's attention to written detail with the exactness with which the research was designed, performed and interpreted. The amount of detail often shows what was unfamiliar to the author rather than to the expected reader, so established methods familiar to experts in the discipline need no explanation.
The methods and results sections are problem areas for most authors. Editors face challenges when trying to infer information from the two sections. Authors often understand the information they communicate, but readers do not. To write methods and results sections that appeal to editors, the advice is to study the masters, the people whose publications have survived the decades test.
The methods is the part of a manuscript most likely to cause rejection, mostly because it frequently does not give enough detail for others to interpret the true significance of the results. Inadequate methods can be, or at least may be viewed as, a sign of problems in the study design. Content guidelines such as CONSORT and the ICMJE recommendations were developed to help authors prepare articles whose methods are complete, and the journal's instructions for authors may also set detailed requirements for the section. Writing a clear, well-organised methods section that satisfies all of them can be a challenge. To avoid rejection for an inadequate methods section, the section must include all the information the appropriate content guidelines and the journal's instructions for authors require, and everything readers need to put the study in context and interpret the results. The methods is like a recipe, and the guidelines are useful to go back to as a check that all the information is included, including the ethical statements and sources of equipment that writers often forget.
The order of the section
One general structure for the section fulfils the requirements of complete reporting of methods. It is only one possible way to organise the section and in no way rigid, and a different flow may work better for a given writer, but it is a good place to start. A similar flow can serve all other kinds of article or study, although some of the information will be irrelevant and the specific guidelines for each article type need to be followed. Whatever the article, the flow should be adapted to its specific needs.
Most of the methods should come from the clinical study report. Listing the main sections of the methods by the guidelines is relatively easy when writing from a report, because most of it can be copied across. Where there is no report, the first advice is to look at how the journal the author would like to publish in sets out its methods. A journal usually follows the same pattern each time: the study design and the recruitment and selection, with the inclusion and exclusion criteria, tend to come first, and things such as ethical statements, sources of equipment and statistical analysis tend to come at the end. This is a generalisation, and each journal has its own specific style. A less experienced writer can start with a word for the study design and a word for patient recruitment and selection.
Because the section presents a number of topics, subheadings are often used. The methods section of one published report on aircraft noise and fatal myocardial infarction has five: Study Population, Outcomes, Exposure to Aircraft Noise, Exposure to Air Pollution, and Statistical Analysis. That of one on blood storage and mortality after transfusion has six: Study Design, Patient Population, Randomization and Intervention, Outcomes, Data Collection, and Statistical Analysis.
What each part contains
Study design and key details come first, as an overview of the kind of study performed and when and where it was performed. The part opens with a sentence describing the overall design of the study, with the clinical trial registration number if there is one, followed by a sentence on the dates and location of the study, and then the study objectives, with an indication of the primary and secondary outcome measures. For example, a phase II randomised, double-blinded, multicentre study in adults with severe Crohn's disease gives its ClinicalTrials.gov number, then the dates and the 6 centres in Austria at which it was performed, then its primary objective of showing whether one dose is superior to another as measured by the CDAI, and its secondary objective of comparing the safety of the two doses.
Ethics follows: approval by ethics committees, the ethical guidelines followed, and a statement about informed consent. It can be described in a separate section or combined with the study design.
Patients or subjects then describes how they were selected, in a single paragraph that first makes clear who was considered eligible and then gives the reasons for exclusion. The part is structured in the same way for an observational study. For a systematic review or meta-analysis it can be replaced by a description of how the articles were selected, and for a study in cells, animals or tissues it should describe what these are, how they were handled and how or from whom they were obtained. The description gives the subjects' age, gender and other important characteristics that may be pertinent to the study. The ICMJE recommendations state that authors who use such variables as race or ethnicity should define how they measured them and justify their relevance. The description also says whether any potential subjects were excluded and why, if there is a meaningful reason, as when a study of probiotics to reduce infant colic excluded subjects with chronic illness or gastrointestinal disorders. The writer can draw a picture of the population, whether they were young or old and how sick they were, the picture a clinician reading the article wants so as to recognise a patient they usually see, and paint it in words from the data the writer has.
Study conduct forms the middle part, because the study design and population need to be described first. For an interventional study, it starts by explaining what was done to the patients or subjects: how they were split up or randomised into groups, what they were treated with, how the treatment was administered, and what assessments were made. The writer may wish or need to describe the treatments in detail in their own paragraph or section, such as one entitled Vaccines, especially if they have not been described before or are not commercially available. This part describes what was actually done to the subjects, with any data collection tools such as survey instruments where appropriate. Any apparatus is identified by item and manufacturer, and every drug by its generic name, with its doses and routes of administration. Adding the trade name is optional but useful for the practising clinician.
The assessments, measures or assays come next. A previously published technical method needs only a single sentence giving the citation, as in Quality of life was assessed on day 28 using the HAQ (12), although a sentence or two summarising it can be included if the writer thinks it important, while an unpublished method should be described in full. In all cases, the limits of detection and sensitivity of the method and the source of any materials or equipment used should be described. A more detailed description, such as one of blood sampling and hemagglutination inhibition titres, should be presented as a separate paragraph or section, entitled for example Immunogenicity. For a clinical study where safety was assessed, the writer may want a separate section called Safety describing in detail the assessments of adverse events, severe adverse events and the scoring of solicited reactions, which are the expected adverse events.
Sample size follows. For interventional studies, describing the sample size calculation is essential, because it puts the results of the statistical tests in context: the relevance of the tests will be unclear if too few subjects were included to detect a meaningful difference. Even if no power calculation was performed, an explanation of how the sample size was selected can help put the results in context. The sample size can be combined with the statistics section, but it can also be effective as a section of its own, especially when it has an important bearing on the interpretation of the results. One writer places it just before the section on statistics.
Statistics should explain the software used, the statistical tests used, the specific populations or subgroups, and general statistical considerations, such as how statistical significance was defined and whether and how missing data were replaced or imputed. The ICMJE recommendations ask for the statistical methods to be described with enough detail to enable a knowledgeable reader with access to the original data to verify the reported results. That generally means identifying the specific tests used. The ICMJE recommendations also ask that the statistical software packages and their versions be specified, that a report avoid relying solely on statistical hypothesis testing such as P values, which fail to convey important information about effect size, and that quantifiable findings be presented with appropriate indicators of measurement error or uncertainty, such as confidence intervals, when the data allow.
Voice, tense, materials and data
Five characteristics of the methods section are its rhetorical intent, its thematic focus, its tense, its inclusion of materials and its inclusion of data. Attention to them may lessen the difficulties of preparing the section.
The sentences of the section are not instructional, in the second person, but descriptive, in the third person, focused on what was done rather than who did it, so they use the passive voice. We isolated the protein needs revising to The protein was isolated, to avoid the egotistic repetition in other sentences whose subjects could also be we. Gel filtration chromatography was used to isolate the protein needs revising to The protein was isolated by gel-filtration chromatography, to de-emphasise gel filtration. The past tense is the least controversial tense for the section, justifiable by its retrospective perspective. For some information, verbs, and so tense, can be avoided by using a list.
Materials that are not multilevel are usually embedded in the text in parentheses, as in The source of proteins A and B was serum of Sprague-Dawley rats (Jefferson Laboratories). Multilevel information, a grouping and subgrouping of many related details, contains too much information and is too distracting to be melded with the descriptive methods. A semi-table can be considered for it instead, with entries such as patients, individuals or experimental animals with their criteria for inclusion or exclusion and characteristics, a questionnaire with its categories of questions, and data management with its statistical tests.
Sources disagree on whether the methods may contain data. One guide holds that the methods should not include numerical data, which should be presented in the results. Another holds that data may be included in the methods if they are intermediate and not the focus of the article. Such intermediate data may include calculations, which are common in biostatistics journals, where calculation is a prerequisite for the subsequent statistical analysis, the purpose of which is to show a lack of bias in data collection, and they may even be expressed in tables. In a survey study, for example, the number of completed questionnaires and possible reasons for incompletion may be stated as intermediate data.
When to write it
The section is ostensibly the easiest to write, probably because its information is technical rather than conceptual and its sequence chronological rather than thematic. These characteristics may be why many authors begin writing a journal article with it. Beginning with it could be premature, however, if the data are inadequate to support publication.