I agree with the general concepts raised here, far too many surveys (in 
general, not pointing fingers at anyone specific)are appallingly badly set up, 
with leading questions, irrelevant options, insufficient options etc. Much of 
this could be avoided by extra scrutiny before finalisation.

-----Original Message-----
From: Wikimedia-l [mailto:wikimedia-l-boun...@lists.wikimedia.org] On Behalf Of 
Jonathan Cardy
Sent: Sunday, 26 February 2017 2:44 PM
To: wikimedia-l@lists.wikimedia.org
Subject: Re: [Wikimedia-l] proposal for regular surveys of community opinion

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 

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.



> Message: 4
> Date: Fri, 24 Feb 2017 19:18:47 -0700
> From: Bill Takatoshi <billtakato...@gmail.com>
> To: wikimedia-l@lists.wikimedia.org
> Subject: [Wikimedia-l] proposal for regular surveys of community
>   opinion
> Message-ID:
>   <CAETpP4uGv3O-=cjw8shgw5s6ax84o7+qn0rs1ejmozubf7n...@mail.gmail.com>
> 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

Wikimedia-l mailing list, guidelines at: 
https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and 
New messages to: Wikimedia-l@lists.wikimedia.org
Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, 

No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.7998 / Virus Database: 4756/14023 - Release Date: 02/25/17

Wikimedia-l mailing list, guidelines at: 
https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and 
New messages to: Wikimedia-l@lists.wikimedia.org
Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, 

Reply via email to