There were a couple of reasons for my question, but I’ll address only one here which is specific to 1Lib1Ref.
Having mentored some 1Lib1Ref sessions, I saw that 1Lib1Ref edit summaries were sometimes tagged without the # so I was interested to establish the extent of under-reporting when I noticed that every entry in the set of edits extracted for 1Lib1Refcontained a #. I suspected that the query used for the official statistics was written by someone who hadn’t directly engaged with the users and had failed to consider real user behaviours.
https://quarry.wmflabs.org/query/17054 (assuming I have published it correctly)
demonstrates what I suspected – even in the recent changes there do exist 1Lib1Ref edits without the # so the query being used to measure the official statistics is underestimating (not everything that turns up in the query above is a true 1Lib1Ref edit – some of them are meta-edits talking about the program). But I don’t see the point of running a program and then under-report its success because someone decided to assume that users do exactly as they are told. What’s the query I need to search over Jan-Feb 2017 (I presume recent changes doesn’t go back that far) so I can see the extent of under-reporting over the whole 1Lib1Ref due to the lack of a #. I also saw other variants of the tag (spaces embedded).
Apart from using the wrong tage, there is an interesting set of real user behaviours to be studied here. Some 1Lib1Ref edits occurred without any identifying tag in edit summary (when new users were grappling with their early edits on Wikipedia, they tended to forget about the tag in the edit summary). I had a number of people say to me “I’ve done my edit, now where do I put that tag?”. So it might be interesting to find the set of users who appear to be engaging in 1Lib1Ref (because some of their edits are tagged) and see what other edits they did in the time period that involved adding citations (can we get the diff in quarry and work out if a ref tag is being added) but weren’t 1Lib1Ref tagged.
I am also aware that some 1Lib1Ref folks made other changes to Wikipedia articles (not involving the adding of citations) and tagged them as #1lib1ref too (not that I think that’s a great tragedy – if we engage new users in any way in improving articles, it’s a positive thing) but still it would be nice to get as big a picture as we can of what happens in such programs where we do NOT have a convenient list of user names to use in tools like the Herding Sheep. Indeed, just being able to extract the list of users who appeared to participate in 1Lib1Ref in some form and finding out how many were editing prior to the program and how many continued to edit after the program would be interesting too.
Now I also hope that I can get access to the internal spreadsheet maintained by a local library that participated where they asked their staff to track their contributions. Since they did over 1000 1Lib1Ref contributions according to their internal spreadsheet (and obviously there may be under-reporting in that too), it would be interesting to see how what comes up in quarry for the same group of users as the difference may give us additional insights into real user behaviour, so we can get smarter with our reporting and our future programs. I am hoping that from that data I can see which staff took part in the edit training sessions and the mentored group sessions (using the known times of these sessions) and compare their 1Lib1Ref performance against the other staff who did not participate in sessions (as well as whether any of them continue to contribute) to see if these sessions have any positive benefit or not. The reason I do outreach is to create a multiplier effect, but if my multiplier effect is 0, then I might as well not bother.
As an aside, I would say that I believe that my multiplier effect for edit training on the source editor was probably very close to 0; after many sessions, none of the users continued to edit beyond the sessions or stopped after a day or so. However, I am getting much more encouraging results from edit training on the Visual Editor. I am seeing (just from my watchlist) that the VE trainees are much more likely to continue contributing. As a double aside, where do I need to go to persuade people to enable on the VE on project pages? At the moment, I am forced to set up REDIRECTs from GLAM project pages (where the VE is not enabled) to User subpages (where VE is enabled) to allow these people to contribute. It’s an ugly workaround but it’s the only solution I have in my toolkit.
Kerry
From: Aaron Halfaker [mailto:aaron.halfaker@gmail.com]
Sent: Saturday, 4 March 2017 2:13 AM
To: Research into Wikimedia content and communities <wiki-research-l@lists.wikimedia.org>
Cc: Kerry Raymond <kerry.raymond@gmail.com>
Subject: Re: [Wiki-research-l] is there a way to get the list of edits containing specific tags in the summary
Hi Kerry,
Querying the database won't be as fast as using the hashtags tool because it maintains a separate index. But here's the query I'd use to get all the revisions for #1lib1ref from the recentchanges table: https://quarry.wmflabs.org/query/17006
Let me know if you need help adapting the query to, say, all revisions or some other subset. :)
-Aaron
On Fri, Mar 3, 2017 at 1:45 AM, Federico Leva (Nemo) <nemowiki@gmail.com> wrote:
Kerry Raymond, 03/03/2017 04:24:
For example, can I get all the edits with 1Lib1Ref in them or some other
tag to be used at an event next Monday?
Since the question was already answered by Pru, I'll just provide an alternative experience: for in-person events, I prefer to let users follow the wiki customs and then show their contributions in a simple list like Herding sheep:
https://tools.wmflabs.org/magnustools/herding_sheep.php?language=it&project=wikipedia&category=Utenti_progetto_WikiDonne&limit=50&doit=Do+it
Nemo
_______________________________________________
Wiki-research-l mailing list
Wiki-research-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wiki-research-l