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 [email protected] To: [email protected] Subject: [Wikimedia-l] proposal for regular surveys of community opinion Message-ID: CAETpP4uGv3O-=cjW8shgw5s6ax84O7+qn0rs1EJMozubf7nkNQ@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
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. Cheers, Peter
-----Original Message----- From: Wikimedia-l [mailto:[email protected]] On Behalf Of Jonathan Cardy Sent: Sunday, 26 February 2017 2:44 PM To: [email protected] 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 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 [email protected] To: [email protected] Subject: [Wikimedia-l] proposal for regular surveys of community opinion Message-ID: CAETpP4uGv3O-=cjW8shgw5s6ax84O7+qn0rs1EJMozubf7nkNQ@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 https://meta.wikimedia.org/wiki/Wikimedia-l New messages to: [email protected] Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, mailto:[email protected]?subject=unsubscribe
----- 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
Yes surveys are useful if set up properly. Having a group of volunteers interested in doing this work would be amazing. Not seeing why we could not manage this in house. Surveys could be developed collaboratively on meta.
James
On Sun, Feb 26, 2017 at 8:59 AM, Peter Southwood < [email protected]> wrote:
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. Cheers, Peter
-----Original Message----- From: Wikimedia-l [mailto:[email protected]] On Behalf Of Jonathan Cardy Sent: Sunday, 26 February 2017 2:44 PM To: [email protected] 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 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 [email protected] To: [email protected] Subject: [Wikimedia-l] proposal for regular surveys of community opinion Message-ID: CAETpP4uGv3O-=cjW8shgw5s6ax84O7+qn0rs1EJMozubf7nkNQ@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 https://meta.wikimedia.org/ wiki/Wikimedia-l New messages to: [email protected] Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, mailto:[email protected]?subject=unsubscribe
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 https://meta.wikimedia.org/ wiki/Wikimedia-l New messages to: [email protected] Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, mailto:[email protected]?subject=unsubscribe