Some data first. As of today, there have been 27636 suggested edits:

  • 23490 have been approved,
  • 4146 have been rejected,
  • 6873 were improved.

The top 20 reviewers have the following stats:

enter image description here

What do you people think of these statistics? are we too lenient when it comes to suggested edits? or the other way around? what would you consider a healthy ratio of accept-to-reject?

The last column seems an important metric: do you think the numbers are good? should they be higher? do you think it is common for reviewers to accept edits even if there is a lot of room for improvement? what could we do to encourage reviewers to improve the edits when it is necessary?

  • $\begingroup$ It would be nice to be able to compare these numbers to other communities, but unfortunately I don't have enough reputation. Also, I don't want to make this post about me, but it bears mentioning that my ratio of reject-to-approve is rather high. It seems that I reject an edit for every two that I approve, which in retrospective seems rather harsh (and even then, some bad edits slip by). I've been trying to be more indulgent lately. I wanted to know what others have to say about these numbers though. $\endgroup$ Oct 6, 2017 at 19:02
  • $\begingroup$ The mods might be able to get corresponding stats for other communities to compare to. But I think we shouldn't feel too obligated to match other sites in this respect. Our ratio should depend on the kinds of edits people submit, and it seems quite plausible that the kinds of suggested edits we get may differ in their appropriateness from suggested edits on other sites. $\endgroup$
    – David Z
    Oct 6, 2017 at 19:05
  • $\begingroup$ Isn't that data available on SEDE to begin with? $\endgroup$ Oct 6, 2017 at 19:41
  • $\begingroup$ @EmilioPisanty the data from other communities? I don't know. As D.Z said, it is not really that important though. $\endgroup$ Oct 6, 2017 at 19:42
  • $\begingroup$ Yeah, SE-wide. The SuggestedEditVotes table has everything you need. $\endgroup$ Oct 6, 2017 at 19:44
  • $\begingroup$ Also, without more qualitative data, this one is extremely hard to answer ─ but I'm going to go out on a limb and claim that the ideal reject-to-approve ratio is exactly 24.53% ;-). More seriously, though - maybe it's worth figuring out a way to sample 20 edits from the most rejective and the most acceptive editors out of the pool of folks with 250+ reviews under their belt, and see whether there's some qualitative difference? Otherwise, I'm not very sure what to make of those numbers. $\endgroup$ Oct 6, 2017 at 19:50
  • $\begingroup$ I'm also not sure whether the correct metric is reject/approve, as opposed to reject/(approve+improve), which would change the reject ratio for folks with high improve ratios. $\endgroup$ Oct 6, 2017 at 19:52
  • $\begingroup$ Personally, I think my 50% approval is being generous. $\endgroup$
    – Kyle Kanos
    Oct 7, 2017 at 2:32
  • $\begingroup$ Comments to the post (v1): 1. If the reviewer sees the suggested edit directly from the front page (rather than from the review queue), it is reasonable to be more lenient towards minor edits, since the post is already on the front page. I.e. reviewing is really a two-tier situation, which the stats may not properly display. 2. The improve stats does not include cases where the reviewer improve independently right after the review without clicking the "Improve Edit" or "Reject and Edit" buttons. $\endgroup$
    – Qmechanic Mod
    Oct 7, 2017 at 10:54
  • 2
    $\begingroup$ Speaking for myself, I have found such variability in the quality of the suggested edits that I would hesitate to draw general conclusions based on statistics. $\endgroup$ Oct 7, 2017 at 15:52
  • 2
    $\begingroup$ Wait up, all y'all. Before we get all worked up and go off on some crusade: is this actually a problem? If so, how, why, and at what scale? Are harmful edits getting through? Or is it merely a case of too-minor edits getting approved? If the latter, are they actually swamping out other content to a significant degree? $\endgroup$ Oct 8, 2017 at 16:37


You must log in to answer this question.