This question has been in the back of my mind since it happened many years ago. A paper was submitted to a not-very-good journal, and the process was:

  • Paper shown to editorial board member, who says she is too busy to review it, but recommends her postdoc instead (she possibly confused the role of the editor and reviewer).
  • Paper shown to postdoc. Postdoc recommends revision.
  • Paper revised.
  • Postdoc recommends acceptance, paper accepted.
  • During copyediting, the copyeditor comes to you (a senior editor in the publisher) with the following issue. One of the plots has four data points, through which the authors have drawn a line. Problem is, with only four data points, the authors could have drawn any of several lines through the data. They would fit less well, but they would still superficially fit the data (see image below; the blue line is the line the authors drew and the red one is the one pointed out by the copyeditor). Given that we know almost nothing about the subject of the paper, what do we do now?

enter image description here

The obvious possibilities are:

  1. Ask the editorial board member. Problem is she didn't make the decision.
  2. Ask the postdoc. He has already recommended acceptance.
  3. Ask the authors directly to revise the paper. It doesn't seem like a particularly difficult issue to address - just add more data points. However, we have no topic expertise to judge the revision (or to even tell if "add more data points" is the right way to handle the issue). If the authors say "we do not need more data points because ...", we are stumped too.
  4. Do nothing and proceed as normal. This avoids the issue of potentially rejecting an already-accepted paper, which is unpleasant for everyone. On the other hand, if the paper actually should be rejected, then this seems irresponsible.

We wound up doing #3, the authors added a couple of sentences explaining why there are only four data points, and we didn't encounter any problems after publication. Still, I'm wondering if any of the other options are better.

  • For number 3, perhaps you should have asked the authors why they draw that line, and why not one of the other possible ones. For example, is it linear regression?
    – gib
    Jun 7 at 8:37
  • @gib I think so yeah. It's presumably some kind of least squares fit. The original line is the best-fitting one, it's just that the other lines seem at least plausible (or not conclusively rejected).
    – Allure
    Jun 7 at 9:06
  • 1
    There are always many plausible lines one could draw. I would expect every reader to know that, so I don't see the problem with showing only the best fit. "Just add more data points" seems to me very easy to say if you don't have to pay for those extra data points (in time, resources and/or money). Assuming it is even possible. I think the authors did the best they could have done with that somewhat unreasonable request. Jun 7 at 9:25


You must log in to answer this question.

Browse other questions tagged .