An interesting discussion on the ethical implications of reviewing a manuscript that one has previously reviewed for a different journal has arisen in another venue. (Hat tip: various participants; not sure which of them want credit but they know who I'm talking about.) Since you, DearReaders, have considerable paper reviewing and submitting experience I thought I'd solicit some viewpoints.
What do you do when receiving a scientific paper to review that you have already reviewed for another journal?
Do you accept it for review?
I think yes. I have done so as a reviewer and as an author, cannot come up with any consistent grounds on which I should object. My score is: no problem.
Does it matter whether you were positive or negative on the prior review?
It shouldn't make any difference. And in fact if your do use this criterion either way, then you are introducing a bias that should be avoided. As a reviewer, re-reviewing can be a burden but if one felt strongly in the "accept" or "reject" camps, one is motivated to do so. It is the neutral / bored cases that one would be motivated to avoid reviewing again. As an author I would of course prefer the obviously demented negative reviewers did not re-review my work and that the positive ones returned! That is no grounds for a policy however: again, no problem.
Do you have an ethical obligation to inform the editor that you have already reviewed the paper for Journal of Bunny Hopping?
This seems to be a point of some contention in the early going. As far as I am aware, I haven't seen any explicit instructions to reviewers that they must do so from any journal for which I review. I have, however, attended on-scientific-publishing sessions in which journal editors have come down strongly that this was an obligation of the reviewer. My feeling at this point is that we have no explicit and consistent field guidelines on which to rely and thus it is worth discussing how people feel about the issue.
As an author, I think that on the whole I would prefer that the editor be informed when a reviewer has seen the paper before. As a reviewer all I can say is that I do generally indicate this to the editor. I'm not entirely sure why but I think this is the most above-board and ethical thing to do. I am one that is at least a little concerned about my own biases and feel quite strongly that the most general prescription for combating the bias that is inherent in any situation in which human judgment is rendered is 1) sunlight and 2) competition of biases. The Editor is the one that is supposed to be accounting for biases when making the final decision and s/he should be aware of potential landmines. (You may recall that this is one reason I prefer Editors who are active scientists over the professional-editor variety.) Note that I do not think that we are at a point where there is an effective and general policy on this and thus there is no stench of ethical impropriety if anyone feels differently. Just because I vote in one direction at present does not mean that I'm ready to etch anything in stone.
I do not think that there is any obligation to tell the editor for which journal one has previously reviewed the manuscript, however. In fact, I think that indicating which journal may be a violation of the confidentiality of review. In a way that merely indicating that one has previously reviewed the paper is not. The reason being that the politics of impact factor and journal reputation are so fraught with issues which are irrelevant to the science at hand. Say a society level journal receives a paper which has been sent out for review (but rejected) at a CNS journal and a paper which has been rejected from another society journal of equivalent status or (gasp) a supposed dump journal? Are they going to be treated equivalently? One thinks not. Suppose an author gets a reputation for constantly submitting stuff to a GlamourMag which is marginal and then working their way down the chain- is this going to given them a bad reputation in the eyes of the editor of a society level journal? It might.
Does the nature of the critique matter?
Here we might focus on the two biggies for which a reviewer might recommend rejection of the manuscript. First because there is something scientifically flawed about it- missing controls, weak data, data not supporting the argument, etc. Second if the manuscript is fine, but not viewed as sufficiently exciting, hot, etc for the journal to which it has been submitted. As a reviewer, of course, the case in which a major flaw has not been fixed but the paper has merely been re-submitted unchanged is a BigDeal. Nothing pisses me off as a reviewer quite so much (or as a reader of a paper that finally appears unchanged in another journal). As an author, such criticism is completely off-base, biased and of course it is a fantastic outcome if one gets one's unchanged manuscript past another set of reviewers! (Ahem.) Actually in full disclosure I have never pulled this. I always try to revise the manuscript (almost) as if I were resubmitting it for two reasons. First, I believe strongly that this process of responding to even the most idiotic reviews as best one might improves the paper- remember that the reviewers stand in proxy for your eventual reader population. Not that you have to take every idiotic suggestion, far from it. Merely that you should chew over each critique as objectively as you can and see where you might improve the manuscript. Second, I operate under the assumption that I may get the same reviewer again!
With respect to the "fine but not hot enough for this journal" critique....hard to say because I'm not a fan of that critique anyway. What I strive to do is to point out why I think a paper is exciting or not exciting but not say anything about not being hot enough for a given journal. The closest I come is criticizing a paper for being a bit too lean on data, not going far enough or something. It has to be really lean for me to say something, in most cases, I only say this when I think it shouldn't be published anywhere. There is one journal for which I review on occasion that insists on a percentile estimate for overall scientific impact and quality- as in "top 25% of papers" or "top 10% of papers", something like that. In this case, I do my best to make the estimate. So I suppose if I saw one of these again at a "lower" journal I might have to think a little harder on this topic.
As an author, the question really becomes a question of reviewers taking a "set" against your paper from reviewing it for a higher-reputation journal that they cannot overcome for the lower journal. That's a tricky one. Suppose you have two reviewer mindsets: one for your society journal and one for your favorite GlamourMag. Let us further suppose that you can apply these standards appropriately under most circumstances. Will using GlamourMag standards on one manuscript bias your ability to use your society journal standards when you see it again? Could be. As an author, score me as "rather not" have a repeat review under these circumstances.
No rousing conclusion here from my present perspective. Obviously I have more uncertainty on this issue than certainty and feel that this is an issue that has not been hashed out to the point of consistent journal policy. Luckily I have a well-experienced and opinionated commentariat to help! Go to it, Dear Readers.