29

A returning theme on our beautiful and productive Meta sites is the topic on downvoting without leaving a comment. Yes, that is a bit cheeky and admit it: you've done that as well.

In a recent post it was claimed that 90% of the downvotes weren't accompanied by a comment. That percentage seemed rather high.

Given we have some public statics in the Data Explorer, how many posts (count / percentage) are there that received a downvote, but no comment? Per site would be a good start, but if any one fancies a network-wide query I'll be super excited.

I realize the downvoter and the commentor might not be the same user, but the user that claimed that 90% figure had that same restriction so for now it is safe to ignore that point.

If a CM can run one of the queries in the answers to come on their private SEDE instance to include stats over deleted posts, it would be awesome.

I've found one question that is related, but its scope is rather narrow for the purpose of this question.

  • 6
    It would also be interesting to see statistics for downvoted questions that are neither closed nor have any comments, since the “closed” description acts like a comment. – user289905 Dec 5 '19 at 10:28
  • 3
    Comments get purged a lot though. It's also unlikely a user will comment if they already see a comment addressing the issue (In regards to the 90% figure) – dustytrash Dec 5 '19 at 14:26
  • 1
    @dustytrash I'm looking for a ball-park figure that ideally gets backed with some confirmation from the public and private SEDE instance. I have to start somewhere.... – rene Dec 5 '19 at 14:32
  • How hard would it be to take into account more than 1 downvote on a post? A post scoring -5 with no comments will probably get your percentages higher if you count 5 downvotes vs 1 downvoted post? – Tinkeringbell Dec 5 '19 at 14:34
  • @Tinkeringbell not hard. That would be a having and group by on the votes table. You can easily fork my query and see how that pans out. Do know I expect the percentage to vary a lot between sites. IPS has probably a different outcome. – rene Dec 5 '19 at 14:48
  • 2
    You're not going to do it for me? :( Then today might be the day that I actually have to use SEDE ;) – Tinkeringbell Dec 5 '19 at 14:50
  • 1
    A short comment, as a lengthy answer aren't appreciated: this specific information isn't meaningful. A quantitative approach doesn't help users who get comments, but no comments that help them to understand the downvotes they received. This approach is similar to a "code quality" tool that derives "you got great documentation" by counting comment blocks somehow. As said: meaningless. – GhostCat salutes Monica C. Dec 5 '19 at 15:02
  • 'Rather high'? IMHO, it's 90% of what it should be. – Martin James Dec 6 '19 at 0:46
  • 3
    -1 downvoted without comment. – Mark Dec 6 '19 at 1:34
  • 2
    Why doesn't anyone talk about upvotes without comment? An upvote can mean many different things, including "this post has a lot of upvotes. Thus it must be right. Upvoted!". – P.Mort. - forgot Clay Shirky_q Dec 6 '19 at 13:16
25

For Stack Overflow, this SEDE query returns that 26% of the posts with a downvote don't have a comment. 11% of the downvoted posts that are scoring less than zero don't have any comment.

Here is the source query

select sum(case when p.commentcount = 0 then 1 else 0 end) nocomment
     , sum(case when p.commentcount > 0 then 1 else 0 end) comments
     , sum(case when p.commentcount = 0 and score < 0 then 1 else 0 end) [low score]
     , count(*) [downvoted posts]
     , sum(case when p.commentcount = 0 then 1 else 0 end) / convert(decimal, count(*)) * 100 [% no comment]
     , sum(case when p.commentcount > 0 then 1 else 0 end) / convert(decimal, count(*)) * 100 [% comments]
     , sum(case when p.commentcount = 0 and score < 0 then 1 else 0 end) / convert(decimal, count(*)) * 100 [% low score]
from posts p
where p.id in (select postid 
               from votes 
               where votetypeid = 3) -- downmod

and here are the results:

nocomment comments low score downvoted posts 
--------- -------- --------- --------------- 
1156543   3245022  499404    4401565         

% no comment           % comments             % low score   
---------------------- ---------------------- ---------------------- 
26.2757223851062065400 73.7242776148937934500 11.3460553235042536000 

Keep in mind SEDE is updated once a week on Sunday.
Give a big shout-out to Monica Cellio for the awesome SEDE Tutorial she wrote.
Say "Hi" in SEDE chat.

  • 8
    Are you sure the percentage values are precise enough? – honk Dec 5 '19 at 8:30
  • 2
    @honk I aim high and on that distance a few digits do matter ... – rene Dec 5 '19 at 8:31
  • 1
    Ghostcat posted an answer claiming: "... how many of the comments provide helpful feedback that enables the OP to improve content. And unfortunately, a SEDE query can't tell us anything about that.". --- Couldn't this query be improved by: #1 Look for one or more downvotes before comments, #2 look for comments (any content), #3 Look for edit, #4 Look for upvotes and undownvotes - comment or not, if the OP edited and afterwards received a number of upvotes the comment (or lack thereof) was 'helpful' (enough) to prompt an improvement. – Rob Dec 5 '19 at 13:38
  • 1
    Wouldn’t it be easier to compare the number of downvotes a user has with the number of posts they’ve commented on and aggregate it across users? If there is a problem with people leaving a lot of downvotes without explanation, that would be a rough estimate of the magnitude of the problem, and whether it’s a lot of people doing it a few times or a few users doing it a lot. We should leverage the fact that users can only vote once per post. Don’t ask me to write the query for it though; I can barely remember how to join. ;) – ColleenV parted ways Dec 5 '19 at 15:03
  • @ColleenVpartedways that is an interesting approach. Downvotes is a field in the users table, Comments has a userid but, yes, that need a join to get result ;) – rene Dec 5 '19 at 15:06
  • @ColleenVpartedways There are some instances where a comment on downvotes is really not necessary, though. E.g. when flagging as "not an answer" I often downvote but don't comment because the comment will usually be added by the review queue (canned comments). Leaving my own comment would be cluttering. There might be more cases where it makes sense to not comment. – user204841 Dec 5 '19 at 15:23
  • @ModusTollens Absolutely. However, it’s best to get a rough idea of what the patterns are before trying to refine the question. The danger lies in people drawing conclusions the data doesn’t support, but there’s really nothing that can be done about that. If downvotes aren’t that different from the number of posts commented on, then we have to ask why are there so many complaints. (Or are there a lot of complaints, statistically speaking?) – ColleenV parted ways Dec 5 '19 at 15:26
  • 1
    @ColleenVpartedways There's also the case where someone might upvote someone else's existing comment instead of adding their own. Or comment without downvoting (e.g. asking clarifying questions, suggesting partial answers). If your query removes the downvoted post association by looking at all comments, I think the results will be drowned out by a bunch of noise. – Troyen Dec 5 '19 at 23:31

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .