What Merits Correction?
A rather interesting comment chain on the website of a social scientist got GG thinking about corrections. (The blog post and comments deal with how to address published mistakes, with comments ranging from “never contact the authors” to “of course you contact the authors”). If fact, GG has gotten into lukewarm water with a couple of folks for pointing out things in their published papers in this blog. Anyways, what merits a correction? And what merits making a stink when there is no correction?
Take, for instance, a map. GG can identify several instances where maps in papers were simply in error. In one case, the author had misaligned a published map he was copying from and put a bunch of isopachs far away from where they belonged. In another, an author was rather cavalier about his location map, which placed a sampling locality far away from its actual location. Now in each case the problem could be recognized (in the first, by looking at the original source that was cited in the caption, and in the second from data tables with coordinates). Neither of these errors have been corrected (and in one case, I know the author is aware of the problem). As in neither case does the error influence the interpretation in the paper, is this worthy of correction? Of a comment?
Let’s rise up a level. GG complained here about a paper by Manny Gabet where one figure accompanying his article seemed off. Now Gabet was very upset GG didn’t contact him directly about this [it was this response that resonated with that post mentioned up top] but never was able to argue that the figure in the paper was correct. In this case, as GG had hoped to make clear in the original post, the graphical error which could make a reader skeptical did not in fact mean that Gabet’s analysis was wrong–it was in fact correct–merely that something was off in the map pane of the figure and fixing it would improve a reader’s confidence in the analysis. This figure too remains uncorrected; in this case, it would seem to the author’s advantage to fix it (perhaps Gabet tried and failed, GG doesn’t know–it would be interesting to hear. GG screwed up a figure in a paper and then had a correction issued with the updated figure, so at least some journals will do this). Correct? Comment?
One more step up. GG was examining a paper by Mix et al. on oxygen isotopes in the Sierran Auriferous Gravels and found that one particularly important site was assigned an incorrect coordinate. In this case, GG did contact the lead author directly and he investigated and kindly acknowledged that that site’s coordinate (the distance upriver from the Eocene shoreline) was incorrect (it was inherited from an earlier paper). Unlike the previous examples, where the mistakes were essentially blunders irrelevant to the point being made, this one did have an impact. Fixing this mistake changes an R2 value from 0.50 to 0.26, which does alter one’s view of the significance of the findings, though the 0.50 value before was hardly overwhelming proof in the first place. There has not been a correction made to this paper (or, for that matter, to the paper that originated the error). Should this be taken up with the editors? Does it (separate from GG’s other issues with the analysis) merit a comment?
These examples here are well below the typical problems that make up the bulk of material at RetractionWatch (whose site pointed out that first blog mentioned up top). But this illustrates the spectrum of problems that can exist in the literature. Given the anger that can be generated by pointing out these problems, it reinforces GG’s belief that good peer review is highly valuable for weeding these out before they cause embarrassment but also suggests that those of us who happen upon these blunders usually shrug and work around them, the work of fixing them being a greater effort (and risking alienating colleagues) more that it is worth.