Community health initiative/Metrics kit

The Community Health Metrics Kit is a project being investigated by the Trust and Safety team, in collaboration with the Community health initiative, at the Wikimedia Foundation. The ultimate goal is a public suite of statistics and data documenting the relative health of Wikimedia communities on a per-project basis. This project was researched in the 2018–19 financial year, and will inform further development of community metrics in the future.

Background

As a movement, Wikimedians have always measured aspects of their communities. Data points such as editor activity levels, new users, and editor retention have been regularly collected. While these metrics provide some useful indications about the health of a project, they do not give major insights into challenges and specific areas needing improvement. The Community Health Initiative wants to build on the metrics work already done by individual contributors, affiliate groups, academics and researchers, and the Wikimedia Foundation.

This project has two primary goals:

1. to have regularly updated quantitative statistics that provide useful insights into aspects of a community’s health, and
2. to provide better qualitative options for finding insights that can’t effectively be measured through quantitative approaches.

This kit and the data it will contain will be targeted towards two major audiences:

1. The local and global communities, in order to ascertain the relative health of their projects and to more easily identify the early signs of an unhealthy community; and
2. The Wikimedia Foundation, who may make use of the data (now in one centralised place) to better identify trends in community health and direct development efforts towards solving problems.

Where we are now: Consultation

Process

Date	Stage
July–August 2018	Metrics definitions; brainstorming; internal publicising
September 2018	Design involvement; determining location; community consultation on metrics
October–December 2018	Further work on design; community discussions to continue
January–March 2019	Technical implementation, prototyping
Before end of June 2019	Launch of Metrics Kit

This project is currently at the design and consultation phase. At this point, we have a rough idea of who this project is for (local community members interested in judging the relative health of their communities/projects, and Wikimedia Foundation staff interested in monitoring this health for things like tool or policy development). We also have a rough shortlist of metrics we'd like the kit to include. That is where we need your help.

How to give feedback

While we have set up the framework for this and put down our ideas, this project will suffer without the community's knowledge and expertise. Here, we are asking you to give us feedback on our ideas and to offer your own.

Please use the discussion page of this project for your feedback. To make it easier, we have set up a number of sections there for the aspects of this project about which we are most excited to hear your opinions. One of those sections is titled Other feedback, because we might miss things otherwise. Please feel free to leave your comments and suggestions in one, several or all of the sections! Thank you in advance for your thoughtful insights!

How to engage deeper

In addition to asking for feedback on what we do, we are also looking for volunteers who like to work more closely with us on the project. There are several ways you can engage, if you are interested:

• You can work with us to refine the way we gather the data in one or more fields for all the Wiki communities
• You can help find bugs and mistakes we may make
• You can get our support to understand your own community better through community health metrics
• You can help spread the knowledge about those new metrics throughout the wikiverse

If you are interested in investing time into this project on one or more of those issues on a mid to longterm basis, please sign up under the corresponding section of the discussion page!

How we will use the results

Your feedback at this stage will directly affect our thinking with regards to this project. As one of the major audiences for this work, the community's insight is naturally invaluable. As such, it will all be taken aboard and used to guide the future of this project as we focus on its design and usecases.

Metrics

We're currently looking at metrics that reflect community health. These will be informed by comments and suggestions made during this consultation, as well as by individual user interviews and other design research methods. Right now, we're looking at metrics reflecting things like:

• Active administrator and user statistics;
• Backlog statistics (compared with the number of people working on them);
• Statistics already used to compare Wikimedia projects with each other (such as Wikipedia article depth);
• Rates of vandalism, blocks, and other administrative tasks;
• Other metrics we haven't yet come up with :)

Design

Two screenshots from the prototype as it presently exists.

We've spent a little time thinking about how this kit might ultimately look and be used by members of the community. Our requirements for this project are that it is:

• Findable: It should be easy for community members to find and make use of in their work.
• Readable/Accessible: It should be simple to read and absorb the information it is attempting to impart.
• Up-to-date: It should be updated live (or at least regularly), automatically and with minimal maintenance.
• Translatable: It should be available to users in as many languages as possible (ideally beginning with the UN standard languages), and be available for users to supply translations where none exist.

Right now, we are thinking of hosting this data on Toolforge, as that option would cover most of the above requirements. We are aware that many of the metrics points we'd like this kit to hit are already collected by Wikistats, and we imagine the finished product would likely turn out quite similar to this in look, feel, and target demographic.

Original plans

Building on existing work

Some top-level community insights are part of the WikiStats portal. This platform provides data on:

• Total unique devices
• Top-viewed articles
• Newly registered users
• New pages
• Edits
• Editors
• Edited pages
• Net bytes difference
• Absolute bytes difference

The Community Engagement Insights survey provides a window into a number of community health aspects, and provides useful demographic information as well.

Individual projects and contributors have used various approaches and API queries to gain specific insights about specific workflows and issues. Examples include the English Wikipedia’s AdminStats and this analysis of Adminstrator numbers and ratios on the different Wikipedias.

There is a large body of past research projects that deal primarily with topics of community health, or look at related aspects and issues.  We have collated some of that here in this table; if we have missed something, please add to this subpage!

Research name	Date begun	Date published	Qual. or Quant.?	Topic	Contact/Author	Relevance
How much do Wikipedians value editing Wikipedia?	2017-12	2018-05	Qualitative	Editor satisfaction	Jana Gallus; Avinash Gannamaneni; Aaron Halfaker	High
Growth and diversity of Technology team audiences	2017-11	2018-05	Qualitative	Technical contributors satisfaction	Jonathan Morgan	High
Holder Administration in anderen Sprachversionen AdminCon 2018	N/A	2018-03	Quantitative	Administrators	User:Holder	High
The Keilana Effect: Visualizing the closing coverage gaps with ORES	N/A	2017-08	Quantitative	Gender gap	Aaron Halfaker; Emily Temple-Woods	High
Wikimedia CH Community Survey 2016	2016-12	2017-02	Qualitative	Community Survey	Wikimedia CH	High
New editor support strategies	2016-05	2016-10	Qualitative	New Editors	Jonathan Morgan; Joe Matazzoni; Pau Giner	High
Becoming Wikidatians: evolution of participation in a collaborative structured knowledge base	2016-05	2016-10	Qualitative	New Editors	Alessandro Piscopo, et al.	High
Impact of The Wikipedia Adventure on new editor retention	2014-11	2016-10	Qualitative	New Editors	Sneha Narayan, et al.	Medium
Online Community Conduct Policies	2016-03	2016-06	Qualitative	Conduct policies in other online communities	Patrick Earley; Karen Brown	Medium
Teahouse long term new editor retention	2015-11	2016-05	Qualitative	Editor satisfaction	Rebecca O’Neill	High
Survey of user in Ireland	2015-09	2016-05	Both	Editor retention	Jonathan Morgan; Aaron Halfaker	High
Wikimedia Nederlands survey (2015)	N/A	2016-04	Qualitative	Community Survey	Wikimedia Nederland	High
Wikimedia CH Community Survey 2015	2016-01	2016-02	Qualitative	Community Survey	Verena Linder	High
Wikimedia Deutschland Editor Survey 2016	2015-06	2016-02	Qualitative	Community Survey	Wikimedia CH	High
Monthly wikimedia editor activity dataset	2015-04	2015-08	Quantitative	Editor activity	Aaron Halfaker	Medium
Active editor spike 2015 (July update)	N/A	2015-07	Qualitative	General - Community Health	Haitham Shamma; María Cruz	High
Community Health learning campaign	2015-07	2015-07	Quantitative	Editor activity	Aaron Halfaker	Medium
VisualEditor's effect on newly registered editors (May 2015 study)	2015-06	2015-05	Quantitative	New editors	Aaron Halfaker	Medium
Active editor spike 2015	2015-03	2015-04	Quantitative	Editor activity	Aaron Halfaker; Dario Taraborelli	Medium
Global South User Survey 2014	2014-09	2015-02	Qualitative	Community Survey	Haitham Shammaa	High
Talk Sentiment Analysis: Editor Retention and Editing Prediction	2014-10	2014-12	Qualitative	Editor retention	Sergio Martinez-Ortuno; Lars Roemheld; Deepak Menghani; Leila Zia	Medium
The Role of Wikipedia Mentorship Programs in the Newcomer Experience	2014-07	2014-12	Qualitative	New Users	Gabriel Mugar with Chris Schilling and Aaron Halfaker	Medium
Wikipedia Gender Inequality Index	2014-06	2014-12	Quantitative	Gender Gap	Piotr Konieczny	Medium
Communicating on Wikipedia while female	2014-11	2014-11	Qualitative	Gender gap	Laura Hale	High
Framing Support for Newcomers	2014-07	2014-10	Qualitative	New Users	Yla Tausczik; Jonathan Morgan	Low
Asking anonymous editors to register	2014-03	2014-10	Qualitative	New Users	Aaron Halfaker	Medium
Non-finite Processes in Human Social Phenomena	2012-10	2014-10	Qualitative	Cooperation	Simon DeDeo	High
Women and Wikipedia	2014-01	2014-09	Qualitative	Gender Gap	Amanda Menking; Jonathan Morgan	High
Women and Wikipedia: Contributions in a Collaborative Online Space	2012-03	2014-05	Quantitative	Gendergap	Melanie Kill	Low
Wikimedia Community Visualization	N/A	2013-12	Quantitative	Editor activity	Haitham Shammaa	High
Newcomer survival models	2013-09	2013-09	Qualitative	Editor retention/New editors	Aaron Halfaker	High
Onboarding new Wikipedians/OB6	2013-09	2013-09	Quantitative	New Editors	Aaron Halfaker	Low
Exploring editing dynamics in different language Wikipedias – Towards a substantive grounded theory	2013-05	2013-08	Qualitative	Social dynamics	Pasko Bilic	High
Getting to know the grassroots	N/A	2013-07	Qualitative	Community Survey	Wikimedia Nederland	High
VisualEditor's effect on newly registered editors/June 2013 study	2013-06	2013-07	Quantitative	New Editors	Aaron Halfaker	Low
Gender micro-survey	2013-06	2013-07	Quantitative	Gender	Howie Fung	Low
WikiProjects as virtual Teams	2013-01	2013-07	Quantitative	Social dynamics	Jonathan Morgan	High
Notifications	2013-05	2013-05	Quantitative	Impact of Technical Feature	Aaron Halfaker	Medium
Anonymity and conformity over the net	2012-01	2013-03	Quantitative	Effects of anonymity	Aaron Halfaker	Medium
Post-registration editor survey	2012-11	2012-12	Quantitative	Editor motivation	Steven Walling	Medium
Wikipedia Editors Survey 2012	2012-06	2012-11	Both	Community Survey	Tilman Bayer	High
Wikipedia Editor Satisfaction Survey	2009-09	2012-10	Both	Editor satisfaction	Denis Barthel; Manuel Merz	High
MoodBar/Email confirmation	2012-07	2012-08	Quantitative	Communication	Dario Taraborelli	Low
Editor Lifecycles	2012-06	2012-07	Quantitative	Editor retention	Shilad Sen	High
Master Thesis Cultural differences in motivations to contribute to Wikipedia	2012-05	2012-07	Quantitative	Editor Motivation	Sjarlot Stal; Nick Geurts	Low
Editor milestones	2012-05	2012-07	Quantitative	New Editors	Shilad Sen; Aaron Halfaker	Low
Shallow Entry: A problem-centric approach to new editor orientation	2012-04	2012-07	Quantitative	Editor Motivation	Maryana Pinchuk	Medium
[1]	2012-04	2012-06	Quantitative	Editor Motivation	Aaron Halfaker	High
Motivations to Contribute to Wikipedia	2012-03	2012-06	Unclear	Editor Motivation	Audrey Abeyta	Low
Necromancy	2012-04	2012-05	Quantitative	Editor Retention	Steven Walling	Medium
Wikipedia Editors Survey 2011 November	2011-04	2011-11	Quantitative	Community Survey	Barry Newstead	High
Women and Wikimedia Survey 2011	2011-05	2011-10	Qualitative	Gender gap	Sarah Stierch	High
Summer of research 2011	N/A	2011-08	Both	mostly New editors, but also Editor retention		Medium
Wikipedia Editors Survey 2011 April	N/A	2011-04	Quantitative	Community Survey	Mani Pande	High
Expert participation survey	2010-12	2011-02	Quantitative	Expert participation	Dario Taraborelli	High
Hebrew Wikipedia satisfaction survey	2007	2007-07	Qualitative	Community Survey	Wikimedia HE	High
100 Questions for Wikipedians	2005	2005-02	Qualitative	Community Survey	Aphaia	Medium
Teahouse group dynamics	2017-05	Ongoing	Quantitative	Impact of harassment	Yiqing Hua; Dario Taraborelli; Leila Zia; Lucas Dixon; Nithum Thain; Cristian Danescu-Niculescu-Mizil	Medium
Detox	2017-01	Ongoing	Quantitative	New Editors	Ellery Wulczyn	High
Study of harassment and its impact	2016-05	Ongoing	Qualitative	Social dynamics	Jonathan Morgan	Medium
Good Faith Newcomer Prediction	2016-05	Ongoing	Qualitative	Social Dynamics	Anna Filippova, et al.	High
Understanding the Dynamics of Hackathons for Science	2016-04	Ongoing	Qualitative	Time commitment and workflow	Kevin Schiroo	Medium
Measuring editor time commitment and workflow	2016-01	Ongoing	Quantitative	Toxic language	Ellery Wulczyn	Medium
Beyond the Gender Gap: Understanding Women's Participation in Wikipedia	2015-10	Ongoing	Qualitative	Gender gap	Danielle McDonald Corple	Low
Online harassment resource guide	2015-07	Ongoing	Qualitative	harassment	J. Nathan Matias, et al.	Medium
Editor Behaviour Analysis & Graphs	2015-06	Ongoing	Quantitative	Editor statistics	Jeph Paul	High
Wikipedia Equality Sweden	2014-10	Ongoing	Quantitative	New Users	Aaron Halfaker	Low
HHVM newcomer engagement experiment	2014-08	Ongoing	Qualitative	Gendergap	Björn Helgeson; Jonathan Morgan	Medium
The sudden decline of Italian Wikipedia	2014-05	Ongoing	Quantitative	Editor Activity Decline	Nemo with Aaron Halfaker	Medium
Dynamics of Online Interactions and Behavior	2012-01	Ongoing	Both	Roles and Functions	Ofer Arazy; Dario Taraborelli	High
Wikipedian and Internet addiction	2010-09	Ongoing	Both	Social dynamics	Dario Taraborelli	Medium
Measuring editor labor hours	2016-04	N/A	Qualitative	Addiction?	OppidumNissenae; Geoide; Alexmar983	Low
Roles and Functions within Online Production Communities	2015-08	N/A	Quantitative	Editor activity	Aaron Halfaker	Low

Focus areas

We began this project by looking at what is already collected and what isn’t, and identifying which data points and areas should be prioritised for this project. This will involve both collecting new metrics and using existing metrics differently (e.g. providing new ratios and comparisons of already-collected information). Some of the broad areas we looked at to include in the Metrics Kit:

• Demographics
• Administrator statistics
• User statistics
• New accounts
• Activity levels
• Retention rates
• Prevalence of vandalism, trolling, harassment
• Abuse filter, vandalism reports
• Cultural, ethnic, gender barriers
• Content bias
• Language barriers
• Access to tools
• Internet access
• Libraries and knowledge resources access

Future plans: More and easier qualitative surveying

Building on the Community Health initiative's analysis of the English Wikipedia's Administrator's Noticeboard/Incidents. This project surveyed users on their experiences with an important noticeboard. We want to do better survey assistance - helping communities survey themselves through easily adaptable survey forms, good techniques for getting respondents, and help analysing the results.