The question Is there hard data on users' cliques for close votes? made me think about whether there is a difference in people's closing and reopening vote patterns.

Do we have any data, for example from SEDE, that would give any indication to users' voting patterns in regards to reopen votes?

Basically I would love to know whether we can see that users normally follow a "Close everything and reopen nothing!" pattern or a "Close fast, reopen fast!" pattern. The latter would be the ideal constellation in my eyes. Maybe there are some other interesting patterns to see in the data.

The information for this question and the linked one may be used so that we as a community can think about which patterns we would love to see continued or changed.

  • 1
    Speaking only for myself, I try to use the same standard to reopen that I do to close. I rarely find a question that has been modified sufficiently to justify the reopen. A stat I wish Kingledion had included was the number of closed Q's that were nominated for reopening AND edited. I suspect the majority of closed questions are never edited and never nominated. I'm surprised Skeptics' reopen rate is as high as it is and I'm not surprised at all by its deletion rate. That's a hard audience. – JBH Dec 10 '17 at 23:19
  • @JBH That would be a nice stat, though it would probably also include small typo-edits by other users, which send the questions to the review queue, too. There are a lot of things to consider with this whole topic and getting the data and the context for interpreting the data together is not an easy task. If you feel it would be a valuable addition you could ask king whether he would do some further analysis by commenting under his answer. More data is definitely good. I don't know if there is data available that would answer your request, but king probably knows a lot more about that. – Secespitus Dec 10 '17 at 23:25
up vote 11 down vote accepted

Close voting by user

There have been 15403 close votes since the site started, and 1416 reopen votes; or about 10 times as many close votes as open votes. Note, this does not count the deleted questions. I measured each voter's close to re-open stance using decibels, so a score of 10 means you close 10 times more than you reopen, 20 means 100 times more, -10 means you reopen 10 times more than you close, etc.

The average for the site is 10.4 decibels. I use a system to account for zeros, so the highest user is 26.8 decibels (48 close votes, no re-opens) while the lowest user is -21.8 decibels (0 close votes, 15 reopens).

Breaking the users down into categories by decibel, 0.37% of all close votes are given by people who's scores are negative (more reopen than close), 22% of close votes are given by users whose scores are positive but below average, while the remaining 78% percent of users have scores above average. 18 % of close votes come from users with a score of 20 or higher; that is, they have cast at least 100 times as many close votes as re-open votes.

User changes in 2017

If you look at data just since 2017, the average score has increased to 11.1 decibels, while the close votes from users with a score of 20 or higher is up to 23% with 84% of votes coming from users who score above 10.

For all time, the weighted score of the top 10 closers is 16.3, while for 2017, the weighted score is 17.8.

Overall question outcome

Edit: More changes to these numbers, I'm learning how to use the database better, sorry

For all time, there are 14166 questions on this site, and 3258 that have been closed but remain, and 2115 that have been deleted; 38% of all questions have been closed.

For 2017, there are 5546 questions posted, and 1522 that have been closed and remain, and 847 that have been deleted; 43 % of questions posted in 2017 have been closed.

Cross Site Comparisons

              Questions Closed Reopened Deleted CloseRate ReopenRate DeleteRate
Worldbuilding     14166   5788      415    2115     40.1%       7.2%      36.5%
Role Playing      27205   6990     1539    2663     25.7%      22.0%      38.1%  
SciFi & Fant      50628  12044     1282    6442     23.8%      10.6%      53.5%
Workplace         26121  16847      668    9013     64.5%       4.0%      53.5%
Travel            34798  14423      845    7816     41.4%       5.9%      54.2%
IT Security       61690  25488      276   18108     41.3%       1.1%      71.0%
Physics          155167  62415      790   45388     40.2%       1.3%      72.7%
Skeptics          12575   5828      602    4378     46.3%      10.3%      75.1%
Cross Valid      168852  61992     1431   53719     36.7%       2.3%      86.7%

Allow me to draw a conclusion. We have the lowest 'delete' rate, that is, the number of questions that get closed before anyone answers or comments on, so that they get auto-deleted. The only other site that has such a low delete rate (that I found, so far) is Role Playing. Yet that site re-opens nearly 1 in 4 questions that get closed. Note there is a general increasing trend in reopen rate with decreasing delete rate.

We are a close and reopen fast type of site, that neglects to reopen.

  • Made changes to overall question outcome based on better understanding of the dataset. – kingledion Dec 7 '17 at 19:31
  • 3
    I wonder how we compare to other sites. On the sites where I'm active, most closed questions never get edited and reopened. I wonder what the network-wide patterns are. – Monica Cellio Dec 7 '17 at 19:39
  • @MonicaCellio I can run this on a few other sites. Are there any in particular you are interested in? – kingledion Dec 7 '17 at 19:40
  • Thanks. I don't have specific sites in mind, but we probably want to look at some other creative/subjective sites. (SF&F? RPG? Workplace? Lifehacks?) The patterns might be different on more objective sites like Physics than they are here. – Monica Cellio Dec 7 '17 at 19:54
  • The review queue for closing questions is a lot more active than the one for reopening. – Vincent Dec 7 '17 at 20:16
  • 1
    I'm with @MonicaCellio, if we are going to do any comparative analysis, it should be against sites dealing with similar types of subjects, particularly with regards to subjectivity (no pun intended). Workplace is probably good. Travel might not be a bad candidate. Information Security? (Deals in facts, but also often opinionated.) The religious sites maybe; Buddhism, Christianity, Islam, Judaism/Mi Yodeya? Parenting might not be a bad idea either. Note: many of these are sites I don't particularly frequent, but I can easily imagine that they battle many of the same issues we do on some level. – Michael Kjörling Dec 7 '17 at 20:28
  • "Note there is a general increasing trend in reopen rate with decreasing delete rate." Is that a general trend, one on our site or a trend you observed on other sites? – Secespitus Dec 7 '17 at 21:30
  • @Secespitus Between all the sites, if the site deletes more questions, it re-opens fewer. We are a big outlier in that case, as is Skeptics, but they are an outlier that reopens more than expected. – kingledion Dec 7 '17 at 21:35
  • Ah, okay. While reading the text I thought this was a trend as in "a trend over time". Thanks for the clarification and the numbers. – Secespitus Dec 7 '17 at 21:45
  • @kingledion can you add Interpersonal Skills.SE? – Vylix Dec 8 '17 at 1:44
  • 1
    @Vylix IPS is still in Beta. They are much smaller and younger, and they are still in a phase where they are supposed to sort out things that work and things that don't work. I am not sure if that is a good comparison to a graduated site like WorldBuilding. – Secespitus Dec 8 '17 at 8:05
  • @Vylix I skipped all the Beta sites on purpose. I basically have nothing to do at work today, so I'm working on some nice graphs to illustrate the differences between sites. – kingledion Dec 8 '17 at 15:50
  • 2
    Regarding your conclusion "We are a close and reopen fast type of site, that neglects to reopen.", I'm not sure I agree, at least in light of the data you show to support that conclusion. A bundle of the sites listed do close about the same percentage of questions (which I use as in "40%-ish" here) but the only site out of that set which reopens more of them is Skeptics which reopens a few percentage points more than does Worldbuilding. Out of the ones you list, Worldbuilding is #4 out of nine by reopen rate. That hardly feels like a terrible record given some of the questions that we get. – Michael Kjörling Dec 10 '17 at 21:10
  • @kingledion, is it possible to gather data on either (a) the number of closed questions that are never nominated for reopening or (b) the number of closed questions that are nominated AND edited (in any way)? – JBH Dec 11 '17 at 0:49
  • @MichaelKjörling I am gathering some more data on that. After my last finals on Wed, I intend to update both this and the re-open post with more information. I'm in the process of sampling every graduate site (except StackOverflow....too big!) so we'll have a more robust comparison eventually. – kingledion Dec 11 '17 at 1:21

You must log in to answer this question.

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