Ron Wasserstein, ASA Executive Director
This is the second in a series of articles about the future of the ASA’s electronic publications. Last month, we began the series with a discussion about the future of journal structure, the article submission process, and peer review. The purpose of this series is to collect feedback from the statistics community about a variety of issues related to what our journals should become over time.
In this month’s installment, we seek your comments about the desired content of the electronic journal of the future. The easy answer, of course, is that we want all of the content and functionality discussed below, and more besides. But all this comes at a cost, and cost issues will be addressed in a future article in this series.
Article Length and Structure
In the print world, page cost dictates journal size, which dictates page length for articles. In the electronic world, “page limits” are a non-issue. A side benefit of page limits, however, has been that authors and editors have to think hard about the essentials of articles. This forces a certain discipline of brevity, but also means important information is sometimes left to readers to interpolate. What should the article of the future include that is often omitted now?
Supplementary Content
Related to the article, itself, may be many kinds of supplementary material: data sets, computer code, graphical information, additional examples or case studies, photos, audio or video, etc. These materials were used by the author in conducting the research, and likely will be helpful to others who wish to do further research. When should such materials be included in the publication? Always? At the author’s discretion? Should these materials be included in the peer review process of the future?
Another form of supplementary content is reader comments (and, possibly, ratings provided by readers). Discussion threads may then emerge as readers comment on the remarks of other readers. Should an electronic journal provide the option for readers to respond to the articles they read, and should those comments and discussion threads (moderated or otherwise) reside in perpetuity with the article? What is the best way to find and remove unprofessional comments?
Errata
In print-based publications, errors an article may contain reside in perpetuity with the article. Errors, sometimes found many years later, are published in errata documents in future journals and may or may not be discovered by a person who has come across the original erroneous journal article and is unaware of the error(s). Electronic journal articles can be corrected when errors are found, while preserving the original version and subsequent revisions. Is this the way errata should be handled in journals?
Nontraditional Content
Many important documents are written that are not in the purview of traditional journals. Often such material is called “gray literature,” and, for statisticians, could include technical reports, design protocols, unpublished weighting schemes for federal surveys, replications of experiments, lecture notes, old qualifying exams, and more. Should the ASA provide a home for gray literature in statistics?
Interaction with the Content
For all this content to be ultimately beneficial, readers need to be able to find it and engage with it. What search capabilities are needed beyond those currently available in ASA journals? And should readers be expected to search, or should ASA journals provide notifications when new materials are presented based on user interests and preferences?
When interesting material has been located, what should we expect of that material? Graphics can be interactive and code can be easily accessed and used with new data brought by the reader, for example. Are such capabilities important?
Please Send Us Your Comments
We are eager to receive your comments about these matters. Please leave a comment at the end of this article or email ASA Executive Director Ron Wasserstein at ron@amstat.org. All comments will be read by a panel appointed by the ASA Board to review and summarize feedback.
There are several features of software development that could be
translated to writing articles, one of which is the idea
of versioning and version control. A journal article could be
stamped with a version and have all previous versions available
via a history. This would allow for updating of errata.
Rather than trying to provide a wide range of additional material
for an article, the publisher should concentrate on making
the core materials available in such a way that further resuse
of the article is possible. For example, rather than asking the
author or publisher of an article to make an interactive version
of a figure, concentrate on making the article available in such
a way that it is possible for a third party to construct an
interactive version of a figure.
Think of the publisher as a “platform”
(c.f. Government As a Platform)
Loosening of page limits is an advantage to electronic publications, but editors and referees can still encourage brevity by having authors remove truly extraneous material. To avoid overly long papers, detailed explanations and derivations that are relevant but long can be moved to an Appendix, or to Supplementary Content (the latter contained in a separate file from the basic paper and Appendix). Thus, simple page limits are replaced by making decisions about what is appropriate for the paper versus the Appendix versus the Supplementary Content.
When errors are corrected this should be noted by giving readers both a date of the revision and an explanation of what was changed. With such explanations, access to the original version of the paper with the error may not be essential. The explanation could be given in a footnote in the revised paper, or as a note in a separate online document listing error corrections. Whether authors could also add to such a separate document later comments and notes that are not actually error corrections, is another question.
Still another question: When errors are corrected in a paper that includes discussant comments, and the error is potentially relevant to the discussion, should the discussants be given the opportunity to revise their comments? And how should this decision be made?
I like the idea of publishing all code and data sets to the extent the latter is permitted. Many of us use proprietary data (especially when you have person-level health data or spatial data on humans), with confidentiality issues a major concern. In many such cases data simulated to have a similar structure could be provided along with results of analysis of those data.