Some words about Open Science (OS)

If not replicated, can we trust Science?

https://unsplash.com/photos/ZYBl6VnUd_0

In the past few years, the research in management (but also in other fields) is facing what is called a reproducibility crisis, as several researchers tried to replicate published studies but fail. This crisis is creating a problem for authors, for research groups, and for those that use academic studies (i.e., the society). If scientific results cannot be replicated, can we trust science?

One huge step forward is commonly called Open Science (OS). OS involves a broad range of practices with the primary goal of making science transparent and accessible. The most famous concept is Open Access, which is the practice of making studies free to anyone.

But there are other practices too. Here is a shortlist:

Open Data

It is the practice of sharing all data that is used or generated by a study. Data is commonly published alongside the article (sometimes, even with a doi, which can be cited in future studies).

Open data is, perhaps, the most debatable practice of OS, since some data might have property rights or might be generated under strict legal conditions. One middle-term practice is to share the data with a small group of editors so they can replicate the study results to ensure they are trustworthy.

Data Reuse

It is derivated from Open Data and is the practice of using the data generated by a previous study to extend, replicate, or estimate alternative hypotheses using the same data. Usually, authors that reuse data are not the same authors of the original study that generated the data. By definition, the data reused must be published in a public depository so that everyone can access it.

Open code

It is the practice of making all codes and research protocols available in a public depository. Users can download the code to analyze the empirical steps made. This is important to avoid errors in the analysis of data and, therefore, in the interpretation of results.

One criticism about this practice is that it configures an additional demand for authors. Research codes (and research data management as a whole) are usually not very structured in academia (at least not as it should be). Thus, structuring them demands extra effort from authors. Apart from this criticism, every author should have a personal protocol of how to create clean codes (at least for him/herself), so they can be reaccessed several times in the research process. Thus, Open code only demands that authors are conscious and try to write codes as clean as possible to share in the future.

Open Peer-Review

It is a variation of the traditional peer-review process. Academic studies are usually published after the peers' approval that its content is relevant and does not contain errors. The peer-review process is traditionally double-blinded, so authors do not know the identity of reviewers and vice-versa.

Open Peer-Review makes reviewers and authors learn the identity of each other by the time the article is published online. Thus, the blindness of authors and reviewers is secured until the publication. Alternatively, reviewers' reports and author’s responses can be made available online alongside the article, providing the public access to the reviewing history.

Preregistration of research

Authors can submit research to preregistration whenever they have a precise plan of how to conduct it. Thus, preregistered research is reviewed before data collection and results analyses. Preregistered research may be submitted, peer-reviewed, and accepted by journals, despite future results. As such, journals commit to publishing the accepted preregistered research even if theory or hypotheses are not confirmed.

The primary benefit of this practice is that the hypotheses are generated before they are tested, thus decreasing what is known as Harking (hypothesizing after the results are known).

Tutorial-articles

This is a personal addition to this list, so you may not find tutorial-articles in traditional OS lists. A tutorial-article is an article that aims to introduce and to provide a discussion and an example of a specific topic in research, usually a methodological one. More experienced and trained authors commonly are those that write a tutorial-article, and they are written in a way that beginners can understand.

Open data and Open code are almost always used in a tutorial-article. Thus, readers can learn about the topic and test it with their own hands at the same time. Because the tutorial-article and the code go through the peer-review process, the code is more likely to be free of errors. Thus, beginners can adapt the code to their research and make fewer mistakes than they would without the tutorial.

Final words

This is a short description of Open Science practices. I hope you enjoy it!

Henrique Castro Martins
Henrique Castro Martins
Assistant Professor of Finance
Next
Previous

Related