December 30, 2014

Am Nat Switches to Double Blind Review in the New Year




Over the next few months, the American Naturalist will be making the transition to a form of double blind reviewing--reviewers will no longer be informed of the authors' identities in an effort to minimize explicit or implicit biases triggered by author names or institutions. As a speaker on the problem said, "Implicit bias produces 'micro-inequities'" that are not always easy to detect.

This transition began at Evolution 2014 in Raleigh, North Carolina, where workshops run by About Women in Science explored issues facing early career women. A presentation for the post-graduate workshop is here:
http://www.clfs.umd.edu/biology/dudashlab/Evol2014%20Post-Tenure%20Workshop%20ImplicitBias%20Final.pdf

And another on implicit bias here:
http://www.clfs.umd.edu/biology/dudashlab/Resources%20to%20learn%20more%202July2014.pdf

The editorial board of the American Naturalist met during the annual meeting and discussed ways we might address the problem of implicit bias. At that time, it was proposed that we try a form of double-blind review. We would blind the authors' identities in the system but not attempt to anonymize the paper itself.

Many people have commented that double-blind review won't work in the sciences, at least in its most exacting form. I tended to agree because I started out in publishing at a true double-blind social science journal and that's how I understood its workings. To anonymize a paper, authors were required to write their papers in such a way that their connection to their previous work was disguised. This produced some very unclear and often turgid writing since it required writing in a passive voice and in the third person to avoid indicating who had done the author's previous work in the same area. Reviewers would criticize the writing. Then after acceptance, the papers had to be thoroughly rewritten. This rewriting and re-rewriting is a labor-intensive process that puts the burden on the authors and requires extensive investments in journal office staff  (to police every page of 900 submissions and/or rewrite every successful manuscript after acceptance). However, the field itself sought anonymity and took the process of double blinding very seriously, in spite of the extra work. I'm using the past tense because even the humanities and social sciences are questioning whether double blind is "a fiction" in the "age of Google." https://www.insidehighered.com/news/2011/05/31/american_economic_association_abandons_double_blind_journal_reviewing

In science, there has always been the problem of how to disguise the origins of a paper from a particular lab or research group. The study organism, the methods, and the references to prior publications can make it all too easy to guess the principal investigator no matter how much a paper is rewritten. Research that has already been presented at conferences is readily searchable and might even have been heard by a reviewer. If authors post preprints, then a quick Google will reveal all. It's not enough to say that reviewers should be too ethical to Google the topic. Checking on the state of the topic or looking up particular references cited are natural parts of the reviewing process. Though not all papers would get "unmasked" in these ways, it does make rewriting and re-rewriting papers to achieve double blind purity a bad investment of everyone's time--and it would require considerable disruption of regular professional activity. If, however, the goal is to minimize unconscious bias, simply removing author names, email addresses, and institutions from title pages and correspondence may make a difference in leveling the playing field at very little cost. There is a claim that just increasing the uncertainty of authorship has a positive effect in leveling the playing field.
http://www.cs.utexas.edu/users/mckinley/notes/snodgrass-tods-2007.pdf

Increasing the use of forms of double-blind review in the sciences seems to be an idea that's spreading:



And so, the American Naturalist will experiment with redacting author identities throughout the system and in reviewer correspondence to minimize bias. Starting in January 2015, authors will be asked to provide title pages with no author names, affiliations, or email addresses. Authors will instead provide that information inside the submission system. They will also be asked to upload their acknowledgments separately during review.

It's taken awhile to figure out how to convert the system. It's just going to have to be a bit messy for a few months! It will be an interesting New Year.


Trish Morse
Managing Editor

2 comments: