19

I'm curious to know what percentage of users whose first question was an ID question:

  • accepted answers to their questions
  • went on to become members of the site (if they were unregistered)
  • went on to become regular members of the site (asking non-ID questions)
  • were already members of another stack when they did so

Related stats:

  • How many regular members have asked ID questions?
  • How many ID questions do we see in a day on average?
  • How has the IDs/per day stat changed over time?
  • How often are ID questions the landing page for random net visitors to our site?

How can we get these stats? I'm not sure if data.SE is available for beta sites.

17

First, I kinda wish this data were available to the public. Second, I'm really interested in figuring out whether these questions are worth the effort. So I'm going to try to pull up stats that I think get at the underlying question, which I read as:

Do identification questions bring useful new users to the site?

Paging through the site's Greatest Hits, you won't find many identification questions. Since that page is designed to be "the most frequently found questions" on the site, there's a good chance that the questions that bring the most visitors are not identification questions. To verify, here is a sample of the search terms that people use to get to the site:

These are a touch dodgy, but it seems clear that people land on some of the better questions on the site. So identification questions aren't really honey for page views by any means.

Do identification questions help new users become contributors?

On the other hand, ID questions are surprisingly common first posts for new users. The top five tags of first posts that new users (N) ask or answer are:

N   Tag
--- ---
353 identify-this-movie
139 plot-explanation
69  identify-this-tv-show
49  science-fiction
43  analysis

To put it another way, the first time 353 people decided to type a post was because they were interested in identifying a movie and 69 people were interested in identifying a TV show. (There might be some overlap.) So the identification questions are substantially more interesting to new users than you might expect from the popular tags.

Moving on the users who eventually register:

N   Tag
--- ---
249 identify-this-movie
122 plot-explanation
46  identify-this-tv-show
38  analysis
28  science-fiction

And go on to post again:

N   Tag
--- ---
84  identify-this-movie
58  plot-explanation
22  analysis
17  identify-this-tv-show
10  ending

Looking at it this way, something like 101 contributing users started off with an identification post. That's a pretty good chunk of the user base. Other tags are "stickier" in the sense that people stick around to post again. (Only a quarter of people post again after starting with identification as opposed to half of those who start with .)

But are those contributors adding value to the site? We can measure that with user reputation:

Avg  Max   Sum     Name
---  ----  ------  ----
117  4982  41,651  identify-this-movie
184  2556  25,581  plot-explanation
 91  1824   6,295  identify-this-tv-show
 95  1824   4,660  science-fiction
527  8761  22,679  analysis

(By the way, if your first question was about , you can expect to get 8,761 reputation and become a moderator. ;-)

So if a new user writes an question or answer, you can look forward to quality content from them in the future. But the tag that introduces the most user reputation to the site (by far) is . As you might expect there are an order of magnitude more identification first posts from users who do not have an association bonus than those who do. Presumably, Stack Exchange veterans have a natural aversion to such questions.

What's going on here? My hunch: users find the site's best content (non-ID questions) and want to participate. One of the easiest questions to ask is the "I remember this movie..." type because that's what you naturally think of when you are trying to come up with a question about movies. After they get their feet wet with a (frankly trivial) question, some users gain confidence to ask or answer better questions.

While it's certainly possible to argue that ID questions are not worth the trouble, the data does not support the stronger claim that such questions never produce good users.

Do identification questions get accepted answers?

Thankfully, this data is public: of the 545 [identify-this-movie] questions, 304 have accepted answers. That's 56%, which is actually better than the site average (35%). But that also means that 44% of the people who asked these questions never got an answer that seemed right to them or failed to come back to the site to see their answers.

52 of those questions are unanswered, which means that they don't have at least one answer with a score > 0. 90% of the ID questions are answered which is less than the site average of 93%.

Summary

I'm not sure what to make of this in the end, but I think we can safely say that:

  1. ID questions are not the best showcases of what the site has to offer.
  2. Identification questions are not how search engine users find the site.
  3. These questions are really good at getting people to post something on the site for the first time.
  4. Some of those people will go on to contribute more content that the community finds valuable.
  • Thanks for looking into this! Could you please ferret out stats on ID questions as landing pages for search engine referrals? That will provide a basis for gauging how useful they are. – coleopterist Oct 31 '13 at 17:15
  • Great analysis! One question. Do these numbers include deleted questions? Many ID questions get deleted when they don't attract answers in a reasonable amount of time, but they still may have brought users to the site. And when these questions were deleted other tags will probably bubble up as first tags. – Gert Arnold Oct 31 '13 at 21:30
  • @coleopterist: That data is kinda hard to get, actually. But I feel pretty confident that ID questions are not really attracting users so much as allowing users to ask their first question or contribute their first answer. I've updated my answer to reflect that (and include a bit more data). – Jon Ericson Nov 1 '13 at 0:19
  • @Gert Arnold: Yes they do. I don't think that they actually bring uses to the site so much as give lurkers a way to post their first question (or possibly answer). – Jon Ericson Nov 1 '13 at 0:22
  • “Presumably, Stack Exchange veterans have a natural aversion to such questions.” I don't know where you got that, and I'm proof that it's wrong. – Gilles Nov 6 '13 at 21:38
  • Also I don't think the search hits prove anything. The meager statistics we get about search queries focus on highly popular or trendy topics. Attention for identification questions tends to be a lot more distributed. A statistic I'd be interested in is how many “me too” answers identification questions get compared with the average — in my subjective impression as a moderator, I think they got more than average. – Gilles Nov 6 '13 at 21:41
  • @Gilles: Well this might have something to do with it. ;-) I was speaking in generalities, though I see that could have been phrased more clearly. I tend to agree that the search statistics don't prove anything. But they are evidence that people do find this site by searching for specific non-identifications questions. I think that it's probably selling a site short to say that ID questions are good for attracting users rather than just engage users. – Jon Ericson Nov 6 '13 at 23:47
11

So looking at stats...

  • Right now we have 657 total ID question over 1 year and 11 months (approx 701 days) = a little less than one per day.
  • These comprise 19.8% of our total questions (3315).
  • SE shows us as having 4.9K users, but when you go to look at the user stats, only 386 users have over 200 points, thousands have asked/answered no questions at all, and thousands have only 1 point or less (only 3900 of the 4900 are even showing, meaning 1000 have 0 or fewer points). So the vast majority of our users (about 90%) contribute nothing.
  • That leaves us with about 400 people with a reasonable level of activity (enough to garner over 200 points). Some may no longer be active, and some of the under 200s may be new and will garner more points over time.
  • Looking at the first 50 Identify-this-movie questions (about 10% of the total 536), 35 of the questions were asked by someone who came back to ask or answer another question (3 of the users asked more than one identify question in that 50), and 15 questions were asked by someone who never asked or answered another question (though they may have been active in upvoting, and some may still browse the site).
  • We've managed to answer over 90% of the Identify-this-movie questions, compared to 95% of analysis questions, 93% of plot-explanation questions, 91% of character questions, 92% of plot-inconsistency questions, and 93% of identify-this-tv-show questions.

While this does not directly answer your questions, it does point to about 30% of identify-askers perhaps not becoming active users, but those users still make a larger contribution to our site than the 4300 or so users who have never asked or answered anything. It does not tell us how many askers of other types of questions don't return to ask or answer another question - those stats may be similar, but I am not prepared to go through 300 questions to get a 10% sample(!).

Our core group of contributors is not very large!

  • Thanks, Mary Jo. This is interesting stuff! There have been 168 (143 + 25) ID questions since August 1 (81 days) which is roughly 2 per day. The total number of questions in this period appears to be 579 which makes ID questions ~29% of the total. – coleopterist Oct 21 '13 at 4:49
3

Here's a start for you...

https://movies.stackexchange.com/tags/identify-this-movie/topusers

Next to each user's name are two numbers. The grayed number is how many upvotes they received, and the darker number is the total number of questions. So @coleopterist has answered 14 identity-this-movie questions for 72 upvotes. 7 of those questions (43 upvotes) were in the last 30 days. You can only see the top 20 askers and answerers, though, for each time period (30 days/ all time).

  • 1
    So i am on the top of all time askers, surprised. May be that is the reason for my support to ID's. – Ankit Sharma Oct 21 '13 at 5:43
  • We have a bunch of ID badges awarded. Interesting. – Donald.McLean Oct 23 '13 at 21:17

You must log in to answer this question.

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