Sauropod Vertebra Picture of the Week

Regulars will remember that nearly two years ago, I reviewed a paper for the Royal Society’s journal Biology Letters, recommended acceptance with only trivial changes (as did both other reviewers) and was astonished to see that it was rejected outright. There was an invitation to resubmit, with wording that made it clear that the resubmission would be treated as a brand new manuscript; but when the “resubmission” was made, it was accepted almost immediately without being sent to reviewers at all — proving that it was in fact a minor revision.

What’s worse, the published version gives the dates “Received August 21, 2012.
Accepted September 13, 2012”, for a submission-to-acceptance time of just 23 days. But my review was done before August 21. This is a clear falsifying of the true time taken to process the manuscript, a misrepresentation unworthy of the Royal Society, and which provoked Matt and me to declare that we would no longer provide peer-review for the Society until they fix this.

By the way, we should be clear that the Royal Society is not the only publisher that does this. For example, one commenter had had the same experience with Molecular Ecology. Misreporting the submission/revision cycle like this works to publishers’ benefit in two ways: it makes them look faster than they really are, and makes the rejection rate look higher (which a lot of people still use as a proxy for prestige).

To the Society’s credit, they were quick to get in touch, and I had what at time seemed like a fruitful conversation with Dr Stuart Taylor, their Commercial Director. The result was that they made some changes:

• Editors now have the additional decision option of ‘revise’. This provides a middle way between ‘reject and resubmit’ and ‘accept with minor revisions’. [It’s hard to believe this didn’t exist before, but I guess it’s so.]
• The Society now publicises ‘first decision’ times rather than ‘first acceptance’ times on their website.

As I noted at the time, while this is definitely progress, it doesn’t (yet) fix the problem.

A few days ago, I checked whether things have improved by looking at a recent article, and was disappointed to see that they had not. I posted two tweets:

Again, I want to acknowledge that the Royal Society is taking this seriously: less than a week later I heard from Phil Hurst at the Society:

I was rather surprised to read your recent tweets about us not fixing this bug. I thought it was resolved to your satisfaction.

I replied:

Because newly published articles still only have two dates (submitted and accepted) it’s impossible to tell whether the “submitted” date is that of the original submission (which would be honest) or that of the revision, styled “a new submission” even though it’s not, that follows a “reject and resubmit” verdict.

Also: if the journals are still issuing “reject and resubmit” and then accepting the supposed new submissions without sending them out for peer-review (I can’t tell whether this is the case) then that is also wrong.

Sorry to be so hard to satisfy :-) I hope you will see and agree that it comes from a desire to have the world’s oldest scientific society also be one that leads the way in transparency and honesty.

And Phil’s response (which I quote with his kind permission):

I feel the changes we have made provide transparency.

Now that the Editors have the ‘revise’ option, this revision time is now incorporated in the published acceptance times. If on the other hand the ‘reject and resubmit’ option is selected, the paper has clearly been rejected and the author may or may not re-submit. Clearly if a paper had been rejected from another journal and then submitted to us, we would not include the time spent at that journal, so I feel our position is logical.

We only advertise the average ‘receipt to first decision’ time. As stated previously, we feel this is more meaningful as it gives prospective authors an indication of the time, irrespective of decision.

After all that recapitulation, I am finally in a position to lay out what the problems are, as I perceive them, in how things currently stand.

1. Even in recently published articles, only two dates are given: “Received May 13, 2014. Accepted July 8, 2014”. It’s impossible to tell whether the first of those dates is that of the original submission, or the “new submission” that is really a minor revision following a reject-and-resubmit verdict.
2. It’s also impossible to tell what “receipt to first decision” time is in the journal’s statistics. Is “receipt” the date of the revision?
3. We don’t know what the journals’ rejection rates mean. Do they include the rejections of articles that are in fact published a couple of weeks later?

So we have editorials like this one from 2012 that trumpet a rejection rate of 78% (as though wasting the time of 78% of their authors is something to be proud of), but we have no idea what that number represents. Maybe they reject all articles initially, then accept 44% of them immediately on resubmission, and call that a 22% acceptance rate. We just can’t tell.

All of this uncertainly comes from the same root cause: the use of “reject and resubmit” to mean “accept with minor revisions”.

What can the Royal Society do to fix this? Here is one approach:

1. Each article should report three dates instead of two. The date of initial submission, the date of resubmission, and the date of acceptance. Omitting the date of initial submission is actively misleading.
2. For each of the statistics they report, add prose that is completely clean on what is being measured. In particular, be clear about what “receipt” means.

But a much better and simpler and more honest approach is just to stop issuing “reject and resubmit” verdicts for minor revisions. All the problems just go away then.

“Minor revisions” should mean “we expect the editor to be able to make a final decision based on the changes you make”.

“Major revisions” should mean “we expect to send the revised manuscript back out to the reviewers, so they can judge whether you’ve made the necessary changes”.

And “reject and resubmit” should mean “this paper is rejected. If you want to completely retool it and resubmit, feel free”. It is completely inappropriate to accept a resubmitted paper without sending it out to peer review: doing so unambiguously gives the lie to the claim in the decision letter that “The resubmission will be treated as a new manuscript”.

Come on, Royal Society. You’ve been publishing science since 1665. Three hundred and forty-nine years should be long enough to figure out what “reject” means. You’re better than this.

And once the Royal Society gets this fixed, it will become much easily to persuade other publishers who’ve been indulging in this shady practice to mend their ways, too.