New subject: proposal for regular surveys of community opinion

26 Feb 2017

I'm keen on surveys, used to work in that line a few years ago, and the first we did
was I think at least in part a response to a proposal I made on the 2009 Strategy wiki. In
hindsight the big mistakes of that survey were that we didn't repeat it annually, and
the lack of community input in setting and analysing the questions.

I'm not convinced that we need to move to a monthly survey, I could live with
quarterly but still prefer annual as the ideal interval - long enough to avoid survey
fatigue, short enough that we can plan around it and use it to answer questions worth
addressing. As for recruiting people, make it annual and I'd hope we could get
consensus for a site notice. I'd like that site notice to be tailored to ask different
and relevant questions based on people's number of edits. - not much point asking
someone with less than a 1000 edits if they are an admin.

The place to set the questions is on meta, not on some external site.

There are of course biases in self reported surveys, there could even be a seasonal bias,
but biases tend to even out as your sample size grows, and an annual survey of the editing
community could get a very high turnout. Also biases don't necessarily hide trends,
provided the biases are consistent. If we were doing an annual survey of the editing
community I suspect we wouldn't need many years before we knew whether our gender skew
was stable, growing or improving. 

As well as the gender skew, it would be good to have an updated age profile of the
community. We still sometimes see people referring to teenage admins without realising
that the adolescents who were our youngest crats and admins ten years ago are now mostly
graduates. I suspect that a new survey would confirm the theory of the greying of the
pedia - our growing number of silver surfers combined with our near total failure to
recruit very active editors from tablet/smartphone only users means that the average age
of our most active editors is going up by more than a year a year.

I'm happy with most of Will's suggestions re questions, but instead of date people
started editing you really want month or quarter to keep the survey anonymous. On smaller
wikis that would need to be year.

It would also be good to survey former editors and particularly those who left after only
a brief period of activity. We have a long tail of people who probably don't consider
themselves Wikipedians but who have fixed one or two things while they are reading
Wikipedia. But we also have a huge attrition rate among editors who have started out and
done 50 or 500 edits. Many will have gone because sourcing edits is too much like hard
work, their view on notability was different to ours or because they couldn't work out
how to deal with an edit conflict. But it would be good to get an idea of the ratio
between those main reasons, and also to find out if there are other significant reasons
for losing goodfaith newbies.

Regards

WereSpielChequers

...
  Message: 4
 Date: Fri, 24 Feb 2017 19:18:47 -0700
 From: Bill Takatoshi &lt;billtakatoshi(a)gmail.com&gt;
 To: wikimedia-l(a)lists.wikimedia.org
 Subject: [Wikimedia-l] proposal for regular surveys of community
   opinion
 Message-ID:
   &lt;CAETpP4uGv3O-=cjW8shgw5s6ax84O7+qn0rs1EJMozubf7nkNQ(a)mail.gmail.com&gt;
 Content-Type: text/plain; charset=UTF-8

 Over the past few weeks I have been discussing how to correct the lack
 of information about community opinion and the disadvantages of
 relying on opt-in (RFCs or less formal "speak up and stick your neck
 out") methods for addressing the problem with Foundation staff, other
 community members, and outside researchers experienced with surveying
 wikipedians. A number of themes are apparent, most prominently that I
 should, "collectively propose and work to develop additional systems,"
 as one Foundation staffer put it.

 So to get that ball rolling, I propose a monthly survey of editing
 community members as follows:

 (1) Anyone may suggest a topic or subject area to be included, for
 each of the top 20 largest language editions of Wikipedia by number of
 active editors, by sending email to an independent, outside firm
 experienced with surveying community members. All such emails will
 have their sender and other identifying information removed and then
 will be posted in a public location on the web for review by anyone
 interested.

 (2) Each month, the independent firm will pick the top five most
 popular topics to be included in each language's Wikipedia community
 survey, and will compose two to five opinion questions on each of
 those topics, with the goal of producing a neutral opinion
 questionnaire with about twenty likert and multiple choice tally
 questions. Every question will have an "other" option when
 appropriate, enabling a fill-in-the-blank opportunity when selected.

 (3) All questions will be clearly indicated as entirely optional. Each
 survey will conclude with demographic questions asking the
 respondents' age, sex, education, household income, and household
 composition, in compliance with the instructions at
 https://meta.wikimedia.org/wiki/Survey_best_practices along with
 state-level geographic location, estimated hours spent editing over
 the past month, and the date each respondent started editing.

 (4) When each month's survey is ready, the independent firm will use
 the Recent Changes history for one day randomly selected from the past
 two weeks to select 1,000 users with contribution histories of at
 least 100 edits and going back at least one year, and who have email
 enabled, and send a link to a Qualtrics survey questionnaire to each
 of those 20,000 users. I believe this step can be efficiently
 automated, but bot approval will be necessary at least for the final
 step of sending the survey email text and links.

 (5) The email will indicate that the survey will be open for two
 weeks. At the end of the two week period, the raw Qualtrics results,
 expected margins or error, and any significant cross-tabulations
 information apparent in the data will be made public at a new web page
 for each language each month, all linked from a static URL where
 highlights from the results will also be summarized in paragraph form.

 I would be thrilled to learn what you think of this proposal. I hope
 the Foundation will consider funding such a regular opinion survey,
 and I certainly hope they will help with implementing the technical
 aspects, but if not, I am willing to pass the hat in the form of a
 GoFundMe or similar.

 Finally, it seems to me that more than a few of the nagging
 controversial questions concerning the Draft Code of Conduct for
 Technical Spaces, a subject of ongoing apparent acrimony on this list
 recently, could easily benefit from such a facility, were it
 available.

 -Will

Re: [Wikimedia-l] proposal for regular surveys of community opinion