In the previous column we realized, with the help of Dwight Atkinson's article, that scientific articles have changed a lot over these three centuries of science [I]. The question that remains is whether they continue to change, which is difficult for those who write these articles to understand: busy writing them, we are not aware of the changes in the time scales of our professional activity. Furthermore, at least in natural sciences, the articles we read and cite are, most of the time, very recent, that is, very similar. Rare, older classic articles can even be strange when reading, but they are the stars of the time, who perhaps do not represent the majority of players of their time, if I may use the football metaphor.
Another question is about who induces changes and Atkinson's article also provides a clue. “The role of editorial policies in long-lived and influential magazines such as Transactions – and the relationship of these policies to the textual development of the scientific research article – are interesting questions, but difficult to study, as few explicit statements of these editorial policies appear to have existed prior to the last 50 years”. So we have one possibility: read these statements. I choose one, the preface to a recent edition of the magazine Scientometrics (yes, about scientometrics), in which the editors express their desire to no longer publish case studies, recommending the publication of articles with new methodologies and/or views on the topic. Case studies would be of more restricted interest and would be cited less, claiming, however, that impact should not determine editorial policy. Finally, they mention that the final decision would be made by the editors (not the magazine, who write the preface, but the publisher). This explains a curious dynamic: the editorial policy of a scientific journal is the result of a conflict between the editors (members of the scientific community) and the publishers (and your business interests).
If case studies must have their spaces restricted, how about thinking about a set of case studies? A change proposed now not by editors, but by authors: a work of crowdsourcing reported by Raphael Silberzahn and Eric Uhlmann, described in Nature in 2015. What would it be? Silberzahn and Uhlmann propose a research question and make a large data set available to research groups recruited through Facebook and Twitter. They were joined by 29 groups around the world with a total of 65 researchers involved. The question: “does the color of a football player’s skin influence the frequency with which he receives red cards?” The data set: compilations of team lineups, positions in which each player plays, referees, number of games, yellow cards, red cards, game results, etc. of four major European football championships.
Each separate analysis would be a case study, but in this case crowdsourcing they were carried out jointly, with anonymous criticism (led by the duo named above) of the proposals between the groups, discussion of the suggestions, also anonymous analysis of the results and, finally, joint publication. It's worth reading: there are 29 different answers to the same question and the same set of data; 29 different methodologies for different data collections. There in the middle of the text is an appreciation of the preparation and review of scientific articles that deserves to be highlighted: “in some cases the authors use an analysis strategy because it is the one they know how to use, instead of being part of a specific rationality. Article referees can comment on the results and suggest changes to the chosen methodology, but comments rarely emerge from the analysis of the data set used. Similarly, it is not uncommon for referees to assume the methodology as correct, without questioning it, and focus exclusively on other aspects of the manuscript.” O crowdsourcing, which should not be taken as a panacea, is a response to these questions, returning the methodological issue to the center of attention. The collective paper “Many Analyses, One Dataset: Making Transparent How Analytical Choices Affect Outcomes” is...a new type of paper [II].
Speaking of data and remembering the shifts in rhetorical focus described by Atkinson, we now have a type of article with the rhetorical focus away from the methodology of an experiment or theoretical description and rather focused on the data itself! In some areas of knowledge, data collection is increasingly extensive and sophisticated. The research group responsible for collecting it may not be able to fully explore the data, so one of the variants of open science is the sharing of this data. As? Published in open access journals. And with that, data article magazines appear. One of them is the Scientific data, connected to Nature (https://www.nature.com/sdata/). What is the objective of this magazine? “It is an open access, peer-reviewed journal for descriptors of scientifically valid datasets and research that advances the sharing and reuse of scientific data.” Finally, the magazine proposes the format of this new type of article and the central focus, for which there is no size limit, are the collection methods, the recording of the data itself and its technical validation. Anyone who wants can (access is open!) browse these articles freely.
The articles have changed over three centuries, in contemporary everyday life we may not pay attention to this, but, to paraphrase Galileo Galilei: and yet, they change.
[I] http://www.unicamp.br/unicamp/ju/artigos/peter-schulz/matou-cascavel-e-publicou-o-pau
[II] https://psyarxiv.com/qkwst/
