Guest post by:
The Association for Women in Mathematics recently launched their new flagship research journal, La Matematica (LaMa), published by Springer. The journal’s title is inspired by the fact that the word for “mathematics” is feminine in many romance languages, including Italian (in case you wondered about the lack of a diacritic).
La Matematica is a peer-reviewed research journal set apart by its commitment to inclusivity and to a positive review experience. LaMa features high-quality research from all areas of the mathematical sciences: theoretical, applied, and computational, as well as mathematics history, education, and philosophy. As Editors-in-Chief of LaMa, we want to support the flourishing of all mathematicians by adopting best practices towards equity in STEM publishing as we currently understand them. For now, that means doubly-anonymous reviewing, a diverse editorial board, an equally diverse set of reviewers, a streamlined submission-to-publication timeline, clear expectations for authors, and constructive reviews.
This blog post describes some of the decisions we made and resources we used when developing our review processes and editorial board, along with some of the struggles we’ve had in living up to our lofty goals.
There seem to be many misunderstandings in the mathematics community about what doubly-anonymous refereeing means in practice, what its goals are, and why a journal would choose to implement that structure. Research is clear that doubly-anonymous refereeing often leads to more equitable outcomes in a variety of fields. It’s important to note that doubly-anonymous refereeing is not just a more equitable practice for white women and for BIPOC mathematicians; research shows that this practice can affect acceptance rates for papers relative to number of authors, prestige of the author’s institution, and country of affiliation. It mitigates “status bias”: inequities that favor famous authors, elite institutions, and those from high-income and English-speaking nations. In other words, a doubly-anonymous referee process can lead to more equitable outcomes for almost everyone in our profession.
Doubly-anonymous refereeing means, simply, that the referee receives an anonymized manuscript to review, with authors’ names and identifying information (grant funding, acknowledgements, and potentially some references) redacted. Some publications ask authors to refrain from posting their work publicly (for example on arXiv) prior to review, but La Matematica does not take that approach. We understand that there are good reasons for authors to post their work to arXiv, especially in very active research areas. So we leave it up to the authors to decide: posting on arXiv may compromise our ability to carry out truly anonymous refereeing, but we respect the authors’ choice. Similarly, we recognize that an author may feel their identity is important to the research presented. Authors determine the extent to which they anonymize the text of the submission.
We recognize that most mathematicians are not able to keep up with every article posted to the arXiv, so it is very likely that we can find reviewers who are not immediately aware of authorship when they receive the manuscript. Even reviewers who think they can identify authors based on writing style or the topic of the paper are at best guessing, and these guesses are quite often incorrect. We simply ask the referees to respect the review process and not seek out information about (or confirmation of) authorship until the review is complete. We trust in the referees’ ability to act professionally and to respect the equity-based goals of doubly-anonymous refereeing.
If editors are unable to find referees who are uncertain of authorship, we will contact the authors before proceeding with a singly-anonymous review process. It’s important to note that in the worst case scenario, when doubly-anonymous refereeing completely fails, the result is the singly-anonymous system, the more typical process for most mathematics research journals. There is nothing to lose and potentially a great deal in terms of equity to be gained by instituting this system.
Other review considerations
We all know the frustration of submitting an article, waiting 6 months (or more!), and receiving a rejection with minimal feedback, saying simply the article is not a good fit for the journal or not the right level. In an attempt to mitigate this frustration and streamline the review process, we have instituted a preliminary quick assessment (either by the editor’s decision, or after a quick expert opinion is received). Submissions that are judged unsuitable for publication in LaMa even if the mathematics is correct will be declined in a timely fashion, ideally within 2-4 weeks. This declination may be based on the article being incremental rather than a substantial contribution, the exposition not meeting the standards of a LaMa article (see below), the content being too specialized to be of interest even in broad strokes to a wide audience, or clear flaws with the approach or the results. A justification of the rejection will be given to the author, with suggestions of alternate journals when appropriate.
If the quick assessment indicates that the paper may be suitable for LaMa, editors will seek two referee reports. Editors will respect the author’s wishes of excluded referees, if this information is provided. Referees are asked to respond to the invitation within 7 days. If no response is received within 7 days, editors will seek another referee, and each referee who agrees to review a manuscript will be asked to send their report within 60 days. Editors will request and enforce that the referee’s reports provide sufficient basis for the decision, with reports designed to improve the submission rather than to criticize it.
On the author side of the relationship, we view research journals like LaMa that cover a broad range of research areas to be an opportunity for cross-fertilization between fields. We therefore ask authors to write papers whose introductions, conclusions, and significance are understandable to mathematicians who are not experts in the sub-discipline, while also providing the technical results and exposition desired by researchers in the area. Continuing on the theme of transparency as a key element in equity, we also encourage our authors to engage in transparency in their research communication. This might entail providing ample resources so that a reader can reconstruct all elements of a proof. For applied mathematicians and statisticians, this means sharing code and data used to generate their results, including all figures and tables in the paper. We realize that this is not always possible (for example, when data is subject to privacy protections) but we request that authors adhere as closely as possible to the tenets of reproducible research.
La Matematica is the first journal in the mathematical sciences helmed by women. But this is the year 2022; diversity, equity, and inclusion has to mean more than seeing more white women in positions of authority.
Research points to a homophilic interaction of identity (specifically gender and national origin) of editors and reviewers and acceptance rates. This means that less diverse editorial boards and reviewer pools lead to less equitable outcomes in acceptance rates. So we worked hard to build a diverse editorial board along multiple dimensions.
In addition to using our existing contacts through the AWM and our various Research Networks, we searched through valuable online resources including Mathematically Gifted and Black, Lathisms, and the Spectra Outlist. We read all of the biographies to identify candidates for the editorial board. After compiling a giant spreadsheet of potential Associate Editors with expertise in all areas covered by La Matematica, we invited a highly-qualified group of Associate Editors to join us on this adventure. The rest of our enormous spreadsheet of incredible researchers became a resource for quick assessments and refereeing.
We are deeply grateful to everyone who agreed to join the board of a startup journal and who has dedicated their time and energy to help us launch the journal this year.
In order to measure our success, as well as to understand and address the under-representation of great swaths of our community in research publications, we will be collecting data from LaMa editors, authors and reviewers periodically. This data will be shared with the community on the AWM website. We see this as an ongoing learning process that will help us — and perhaps others — develop practices that encourage communication and visibility of the work of all mathematical scientists.
What is written above is a set of goals, and we have not yet fully achieved them. For some goals, such as accountability and reporting, we haven’t been around long enough to collect meaningful data. For others, such as a timely and constructive review process, we are still working out some kinks in our system. We are working within the system of a major publisher. And we — all of us, including the Editors-in-Chief, the Associate Editors, the referees, and the authors — are still doing our best to muddle through the lingering pandemic and the havoc it has wrought on our lives and our ability to work. We know that we have often fallen short of our goals, especially with regard to clear communication and timely reviews.
The process of starting this journal has been a lesson in humility, and we are committed to working toward these ideals and to continue to learn and grow in our understanding of best practices in equitable publishing. We are grateful for those who have provided resources to make our job easier, in particular those who have published about equitable review processes and who have gathered information about mathematicians from specific demographic groups (Lathisms and Mathematically Gifted and Black most notably). What we do know is that the journal only exists because of the creativity and hard work of the authors who submit to us, and the insight and hard work of those who review and edit for us. For that, we thank you.
We invite you to become a part of the LaMa community, as an author, reviewer, editor, or reader. You can submit articles on Springer’s LaMa Editorial Manager, and you can contact the Editors-in-Chief to be considered for future openings on the editorial board or as a referee.
We settled on the processes outlined above after extensive research into equitable review processes and helpful conversations.
- N. Reid, Hubble Cycle 26 TAC and Anonymous Peer Review: Anonymous peer reviews in proposals for Hubble projects switched the proportions of successful proposals led by men and women PIs.
- D. Peters, S. Ceci, Peer review practices of psychological journals: The fate of accepted articles, submitted again: Previously accepted articles were resubmitted with fictitious names and most were rejected, with the most common reason for rejection being “serious methodological flaws.”
- A. Meadows, Eight ways to tackle diversity in peer review, This article provides a list of concrete steps to take to address bias in peer review, including anonymous (or open) reviews, training and mentorship, data gathering, and diversity in editorial board,
- D. Murray, K. Siler, V. Lariviére, W. M. Chan, A. Collings, J. Raymond, C. Sugimoto, Author-reviewer homophily in peer review, Demographics of reviewers/editors and submitting authors affect acceptance outcomes, with similarities (homophily) leading to higher rates of acceptance.
- B. McGillivray, E. De Ranieri, Uptake and outcome of manuscripts in Nature journals by review model and author characteristics. Journal type and author institution and country play a role in whether or not an author chooses doubly anonymous review. Papers are more likely to be rejected through doubly anonymous review than through singly anonymous.
- T. J. Tanenbaum, Publishers: Changing the names of trans people in their own work is not enough This blog post describes some challenges and proposed solutions related to trans authors in scientific publishing.
- N. Buchanan, et al, Upending Racism in Psychological Science: Strategies to Change How Our Science is Conducted, Reported, Reviewed & Disseminated. This paper presents a broad analysis of the process of science from an anti-racist perspective. It presents a list of concrete actions to take, including around peer review and publishing, and recommends a set of benchmarks for measuring progress.
- T. Jefferson, E. Wager, F. Davidoff, Measuring the quality of peer review. An absence of consensus about the objectives of peer review makes it almost impossible to effectively define and then assess.
- I. Skelmakh, N. Shah, A. Singh, H. Daumé III, A Novice-Reviewer Experiment to Address Scarcity of Qualified Reviewers in Large Conferences. Good recruitment and mentorship practices improve the quality of reviewer pools.
- N. Shah. Tutorial: Systemic challenges and solutions on bias and unfairness in peer review. These slides present many of the challenges of peer review, propose some solutions, and list open issues yet to be addressed. The primary focus is computer science conference reviews, but many ideas are more broadly applicable.
- A. Ragone, K. Mirylenka, F. Casati, M. Marchese, On peer review in computer science: analysis of its effectiveness and suggestions for improvement. This study analyzes peer review for quality (reliability, fairness, validity), robustness (e.g., agreement between reviewers), and the ability to predict which papers will have significant future impact. Two noteworthy findings are the lack of robustness for mid-tier papers and lack of ability to predict significant future impact.
- G. Smith, Labels Matter: Methodology and and Visualization. In this recorded webinar hosted by the QSIDE Institute, Gaelen Smith discusses best practices in collecting data on identity, and things to consider when visualizing the data. The main takeaways are: free-answer questions are much better than multiple-choice questions when inquiring about identity because they allow everybody to be seen, make space in data visualizations for the unseens (i.e. the “none of the above” or “prefer not to answer” category), always question your assumptions and be willing to make changes: continue to do research, context matters: who is asking and who is answering?.
- S. Thornton et al. Best Practices in Collecting Gender and Sex Data. A research paper written by a group of statisticians giving advice on how to construct a set of questions for collecting data about gender. They emphasize the importance of articulating the context of the questions before writing the questions.