The first person in a research article
Whether a research article says we is decided section by section: the first person suits a statement of the authors' own opinion or a distinction between their work and the work of others, while the methods and the results usually keep the reader on what was done and what was shown.
Whether a paper may say we
It is a widespread myth that words such as I, my, we and our are not acceptable in scientific papers, and by 2007 it was unusual for journal editors to veto a personal style in appropriate parts of a paper. In 1999, a review of a random selection of 100 life-science journals found first-person pronouns in all of them, and in 2001 a review of 500 sets of instructions for authors from medical and life-science journals found only two that specified an impersonal, passive style in articles.
Personal style is not appropriate for all sections of a paper. Descriptions of materials and methods and statements of results are meant to be straightforward presentations of facts, in which readers want emphasis on procedures, actions and outcomes rather than on agents, so it is usually irrelevant, and sometimes obtrusive, to mention there who did or observed what. It is appropriate to mention agents in an introduction or a discussion, where it is frequently essential to distinguish clearly between what others have thought or done and what the authors thought or decided to do.
Personalism and thematic focus
First-person personalism results from a story-line narration rather than a thematically focused description: agents are the subjects of sentences or clauses and their actions are the verbs, instead of themes represented by noun subjects linked by the verb to be to a subject complement. One view holds that personalism in a journal article may be reader-friendly but is distracting, because it deviates from the expected professional formality and comes at the expense of thematic focus.
The introduction
A well-written introduction normally includes a mixture of personal and impersonal, active and passive constructions, as in a passage that moves from "Methadone is used increasingly for controlling pain in the elderly and in palliative care settings" to "In this paper, we report the case of an elderly woman with…". In an introduction, explicit markers of the conceptual components a reader is anticipating, such as an objective, are thematically focused, especially in contrast to the investigator-focused we. For example, "We hoped to develop an injection therapy for women whose cancer cells over-express this oncogene" can be revised to "The objective was to develop an injection therapy for women whose cancer cells overexpress this oncogene". Likewise, "We were unsure about the mechanism for CKI cessation enabling the initiation of the cell cycle" can be revised to "An unanswered question was the mechanism for CKI cessation enabling the initiation of the cell cycle", a revision that creates a nominal conceptual subject, uses the descriptive verb to be, and makes the mechanism the subject complement.
Background from an author's own published or unpublished research can be indicated by a reference rather than by "we did this and we did that", especially where citations give author name and date rather than a number. But without an explicit reference to the authors, their identity as the authors of the cited article is not obvious, which justifies a statement such as "We previously identified a cell line with a single base pair deletion in exon 11 of the Smad2 gene". An impersonal structure can also hide where the work of others ends and the authors' own begins: after "lipofuscin formation by interactions between oxidative stress and autophagocytic degradation was proposed by Brunk et al.", the sentence "It was hypothesised that H₂O₂, mainly generated by mitochondria, permeates into…" led one reader to assume that Brunk et al. did the hypothesising, a confusion "We hypothesised that…" would have avoided. Writing we there is not self-promotion: it claims no credit for something the authors did not do and throws no irrelevant emphasis on the actors rather than on their activity or its outcome, but simply makes a clear statement of fact.
Materials and methods
In most situations it is not important to the reader of a materials and methods section to be informed who performed the experiment, and the personal pronoun we is often disrespectful to the technical staff who actually performed the experimentation but whose names may not appear as authors. Revising "We exposed the auditory cortex in the right hemisphere by craniotomy and durotomy" to "The auditory cortex in the right hemisphere was exposed by craniotomy and durotomy" deletes we and turns the active exposed into the passive was exposed, and with it moves the thematic focus to the auditory cortex. A materials and methods section is likely to be best encoded in mainly passive, impersonal forms, but an occasional first-person, active construction may be appropriate, such as "In contrast with previous studies, we used a recirculating design".
The results
In a results section, the author-focused, often egotistical "we did this and we did that" can be avoided by shifting from who did what to what was shown. For a data-based preliminary interpretation, "We demonstrated that epidermal growth factor-activated Ras-signaling enhanced pol I transcription (Table 1)" becomes "As shown by the data (Table 1), the epidermal growth factor-activated Ras-signaling enhanced pol I transcription": the agent subject with its noun-clause direct object is replaced by an elliptical adverb clause of orientation, and the noun clause is reduced to the noun phrase that becomes the sentence's subject. Results presented as narrative text are usually appropriately presented in impersonal constructions, preferably active, such as "There was no significant difference in mean peak PID scores between oxycodone 5 mg/ibuprofen 400 mg and ibuprofen alone".
The discussion and conclusions
In the discussion, authors are expected to say what they think their results imply: they review their facts, give their opinions and, where appropriate, discuss the ways in which their findings confirm or differ from the work of other authors. So it is entirely appropriate, indeed often essential, for that section to use personal pronouns, to distinguish clearly between what the writers think and what others think. A discussion will normally include a complete mixture of first-person and third-person, active and passive constructions, such as "We have used a set of restriction endonucleases" among impersonal statements about the methods.
Where personalism is treated as a distraction, we still seems most appropriate for accepting responsibility for an opinion, such as a hypothesis, a conclusion or a preliminary interpretation, although personalisation connotes some assertiveness, which may be less appropriate than a depersonalised statement. Of all the sections of a journal article, the discussion seems the most likely to contain first-person pronouns, because accepting personal responsibility for an inference or conclusion modulates the presumptions. Yet "We conclude that this algorithm can be applied to many geometry-based systems" emphasises the authors at the expense of the information, and it can be revised to "In conclusion, this algorithm could be applied to many geometry-based systems". Conclusions may be presented as a numbered list of impersonal, positive statements, but more usually they are asked for in narrative form, again as impersonal, positive statements.
Impersonal statements of opinion
Impersonal encoding all too often leads to loss of accuracy. In English, impersonal constructions such as "is considered to be" are generally understood to imply that the community at large accepts the assertion. So writers who mean "We do not consider such a calculation is meaningful" but write "Such a calculation is considered not to be meaningful" seem to assert that no one considers the calculation meaningful. Likewise, "This phenomenon is of unknown significance" asserts something its writers did not, and perhaps could not, support, where "We do not know the significance of this phenomenon" would have been accurate and honest.
Thinking, believing, considering and hypothesising are all subjective activities, and constructions such as "it is thought that", "it is believed that", "it is considered that" and "it was hypothesised that" do nothing to ensure objectivity. As a broad generalisation, every such construction should be followed by a reference telling readers who did the thinking or hypothesising, and without one the writer is probably making an unclear assertion. Writing "The authors postulated that…" is no more objective than "We postulated…", since postulating is a subjective activity.
To write personally is not necessarily to introduce bias, which is introduced mainly by what is said rather than by the way it is said. Interpretation of results is a matter of judgement, so "We consider that…" is no more or less biased than "The authors consider that…", and the we version arguably has a more open, credible tone of confident ownership of the opinion than the arm's-length, evasive tone of the other.