Today, Nature Reviews Methods Primers published its first articles, including a Primer on Bayesian statistics and modelling. To provide readers with a behind-the-scenes look at the writing process, the editors spoke with Sarah Depaoli (UC Merced), Ruth King (University of Edinburgh), Mahlet Tadesse (Georgetown University) and project leader Rens van de Schoot (Utrecht University) about their experience writing this article.
How did the process of writing a Primer differ from writing a Review? What were your favourite aspects and what did you find most challenging?
Sarah Depaoli: My favourite aspect was working with such a diverse team with a broad range of expertise. It is common for researchers to read and understand the literature in their own fields to a deep and extensive degree. However, knowledge about other fields is typically at more of a peripheral level. It was very interesting to work with Bayesian experts tackling the topic from different angles than my own. The most challenging part of the process was keeping track of the many moving parts, but those moving parts fell into place nicely and created a good final product.
Ruth King: Writing for the Primer was focused at a wider scientific community than for Review papers that I have previously written so that it was generally broader and there was a clear need to minimise and/or explain what would usually be regarded as standard scientific jargon. It also involved a wider set of authors most of whom had not worked together previously, bringing together a wider set of perspectives and viewpoints — which itself was somewhat interesting. My favourite aspects were seeing different viewpoints and how pieces fit together that are wider than my own research area as well as collaborating with new individuals. Collaborating with many new individuals was also a challenge, as was balancing the level and depth of scientific details within limited page/word restrictions. (I would note here that this could have been more challenging than it was but Rens was fantastic in co-ordinating everything and everyone!).
Mahlet Tadesse: The process is a bit different because of the underlying differences between the two formats. Usually, when writing a Review, the target audience is already familiar with the fundamentals of the subject area and the focus is on digging deeper in a specific topic. The contributors of the Review are experts in that particular topic and are familiar with each other’s work. On the other hand, a Primer targets readers who are not familiar with the subject area and the goal is to give a broad overview. The contributors of the Primer are experts in different subfields of the subject area and may not necessarily be familiar with each other’s work. It was both rewarding and challenging to try to explain a fairly broad research area in a limited number of words without using many equations.
Rens van de Schoot: The process of writing a Primer is completely different compared to a ‘classical’ paper. Determining the exact audience determines how much detail is needed. For a classical review the typical audience is colleagues from all around the world, but familiar with the basics of your field. A Primer should be readable for a much larger audience which requires a more basic introduction without turning the paper into a tutorial. At the same time your colleagues will read the paper and they need to recognize the state-of-the art of the field. So, the challenge is balancing easy-to-read parts with highly technical parts without using too much jargon. You should really like finding this thin line if you want to write a Primer because during the review process we often had to start over again, which is a great experience if you are prepared for it. As the lead author, you need to keep calm when receiving the many rounds of feedback, and you have to be skilled in version control. But, it is all worth it, not only because a Primer truly is a team effort describing the current state-of-affairs in a particular field, but also the process itself is refreshing!
How did you find collaborating with such a global team?
Sarah Depaoli: My experience of working with such a diverse, global team was very positive. Each coauthor came to the project with a different set of skills, and reading about the different perspectives of Bayesian methodology expanded how I view the field. The primer is from the perspective of a team of researchers who have vastly different areas of expertise within the same global topic. This team-based structure allows for a more comprehensive treatment of the topic.
Ruth King: This was interesting — particularly as the collaboration was with individuals who I would not normally have interacted with. Of course working across countries/continents is significantly easier than historically.
Mahlet Tadesse: The primary author and the editor did a great job coordinating the contributions from each team and pulling the different sections together. All communications were done via the primary author who relayed the comments from the editor and from one team to another. I believe the process was fairly seamless for all contributing teams.
Rens van de Schoot: A truly international team with senior staff and some early career scientists, but all super-experts in their field is what makes a Primer unique. Every author was responsible for her/his subsection and we provided feedback to each other, which was yet another layer of peer feedback. This way of sharing responsibilities in combination with peer-feedback resulted in great discussion and I learned so much from my colleagues.
What was it like working with the editor?
Sarah Depaoli: The editor gave very helpful comments throughout the process and was very easy to work with. The editor took on a global view of the project, ensuring that the various pieces (and voices of the authors) were weaved together coherently. I also found the panel of reviewers to be very helpful in that they provided nice suggestions and commentary for improving text.
Ruth King: The editor provided very detailed comments — and it did involve thinking more about the ideas and how things were expressed. Not all comments were agreed with — but even this did provide further thought provocation and to step away from the article to see how others in very different areas would interpret and perceive the ideas and written explanations.
Mahlet Tadesse: The editor had great input throughout the process. I have to admit that at each iteration of the manuscript, I would think that my section was done, but then the lead author and the editor will come back with clarification questions, and my first reaction would be “this is obvious, I cannot explain this in any simpler way”. At the end, however, I think that their comments helped make the Primer more accessible for readers who are not statisticians or are not familiar with Bayesian inference.
Rens van de Schoot: It all starts with discussions how to organize the skeleton of the paper and whom to invite as co-author. During the process the editor is very approachable and really thinks along with you how to improve the text. I believe there is not a single line of text which was not discussed or changed during the many rounds of improving the manuscript.
Is there a section that you think was particularly valuable to the Primer? If so, what was it and why do you think it is important for readers?
Sarah Depaoli: As an overview of the implementation of Bayesian methodology, I found the Experimentation section to be an important overview of the key ingredients to Bayesian statistics. Readers new to the topic area will view this section as an important reference for the general topic.
Ruth King: It is difficult to pick a single section! However — perhaps biased by my own Applied Statistics focus — I think that the Applications section is particularly valuable as it demonstrates the depth and breadth of Bayesian statistics within the wider scientific community. In some sense it showcases the impact of Bayesian statistics in a range of diverse areas which should hopefully provide some motivation and impetus for researchers to find out more about the application of Bayesian inference within their own particular field of interest. Of course in order to perform an analysis of one’s own data then one should follow the “Bayesian research cycle” (Figure 1) — and this “simple” figure very concisely summarises each step that should be taken within such an analysis.
Mahlet Tadesse: I think all the sections are relevant to give a broad perspective of Bayesian inference and are valuable to the Primer.
Rens van de Schoot: I think the combination of main text, with additional boxes, extensive figure headings and a glossary makes a Primer completely different compared to more classical reviews or tutorials.
Is there anything that you gained in the process that you think you’ll take forward?
Sarah Depaoli: Substantively, I learned a lot about the application of Bayesian methods in fields outside of my own. Reading the work of the rest of the team was enlightening, and I left the project with an even deeper appreciation for the application of Bayesian methods across various fields.
Ruth King: Possibly thinking more about how figures can be used to more effectively convey information (“a picture (aka figure) is worth a thousand words”); and how best to critique one’s own writing by putting oneself in other individuals’ shoes who will have very different experiences/knowledge.
Mahlet Tadesse: It was interesting to see the process of pulling this project together in a relatively short period of time. The primary author's management and coordination skills are impressive. I hope to replicate those skills in my collaborations.
Rens van de Schoot: I very much appreciate the general shift from me-science to we-science and I believe these Primers are an excellent example of benefits of team-effort