Editor Growth and Contribution Program/Contribution portal - Phase 0

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search

Phase 0 of the Contribution Portal project was piloted on Arabic Wikipedia during the second half of June and all of July 2012. The project was planned in May 2012 as part of the Editor Growth and Contribution Program, and we launched and measured the outcomes of this phase over 4 weeks between 18th June and 17th July 2012.

Some key findings of the phase 0 pilot, discussed in more detail below, include:

  • Running banners using the local site notice that invite readers to contribute can be a powerful tool for attracting new editors
  • Providing new users with simple, easy to understand, visual tutorials can encourage more new users to contribute to Wikipedia
  • The newly introduced metric, Contribution Ratio, appears to be an effective tool for measuring activities of new registered users on Wikipedia
  • The contribution ratio on Arabic Wikipedia increased from 4.5% to become 5.7% as a result of running motivational banners landing on simple tutorial help pages

Concept[edit]

The key idea of phase 0 of the contribution portal is to create a page with a simple design in order to test the overall concept of a portal, while monitoring the ability of new users to perform various simple Wikipedia tasks relying merely on reading visualized tutorial pages.

Name[edit]

Bawabatu Almusharakah (Arabic : بوابة المشاركة), which is literally translated as “Contribution Portal” or “The Participation Gateway” in English, is meant to evoke a space which provides help as well as an being an entrance point for newcomers to Wikipedia, enabling them to contribute to its contents and share the knowledge they possess, while participating and communicating with existing Wikipedia community members.

Design[edit]

Scheme of the Contribution Portal for Phase 0.

Phase 0 of the contribution portal consisted simply of a main portal page leading to visual help tutorials. The main page listed six buttons, each linking to a step-by-step tutorial page similar to the scheme on the right.

Main Page[edit]

Snapshot of the Contribution Portal main page (Arabic).

The main page of the portal starts with a question at its top saying “What would you like to do in Wikipedia?” followed by six buttons each listing an answer to this question, and linking to the corresponding tutorial page. We tried to keep the page design as simple as possible in order to avoid any visual distraction for the first test of the concept. We also aimed to used as little text as possible in order to quickly guide visitors to their target page without spending much time reading. In addition, we added an icon next to each of the tutorials links resembling the subject of the tutorial in order to make the link more informative and visually appealing. Finally, at the bottom of the main page, we added a link to add a new comment in the portal discussion page, in hopes of gathering early feedback from visitors. While drafting the design of the main page, we implemented some ideas used in the design of New contributors' help page on English Wikipedia.

Tutorial Pages[edit]

Tutorial showing first step in creating a new article starting from the searchbox.

We created six tutorial pages with topics covering the most frequently asked help topics in the Arabic Wikipedia help desk. The tutorial activities cover both creating new content and improving the existing content of Wikipedia. These six pages are :

  1. How to create a new article
  2. How to fix a typographical error
  3. How to create a user page
  4. How to post a message on a user discussion page
  5. How to create a redirect
  6. How to add a reference

Each tutorial page contains a series of snapshots of Wikipedia along with comments pointing out the call for action (e.g. Click “Edit”) and a simple explanation. The figure on the left shows the first step of creating an article starting from the search box, and then creating a missing topic if the article doesn’t exit.

Users[edit]

Scheme of user access through phase 0 of the contribution portal.

To invite users to the portal, we ran banners using the local site notice on Arabic Wikipedia. During the first week of the pilot the banners targeted only logged-in users. This brought a small number of visitors to the portal every day. After the first week, the banners were switched to target all readers, logged-in or not, which significantly boosted the number of visitors. The overall user flow through the contribution portal is described in the figure to the right. Users who click on the banner land on the main portal page, and then click through to the tutorials.

User registration[edit]

Initially, in order to make it easier to measure the impact of the portal on those who visited it, we created a simple registration process using Javascript which activates on loading the main portal page. The registration process followed this algorithm:

  1. Open the contribution portal (on a banner ... etc)
  2. Open the registration page in edit mode (small window)
  3. Show a confirmation message "Would you like to register your username in the Contribution portal? Yes, or No?" (go to 4 if yes, go to 5 if no)
  4. Add current user name to the registration page (Skip if user name is already on the page)
  5. Save the registration page
  6. Close the registration page

Registered names were stored in the registration page‏, and they were intended to be used for data analysis. However, in the first week of launch we quickly encountered some issues concerning this registration process, including:

  • Many browsers are set by default to prevent popup windows, so most users didn't view the registration process and got a warning from their browser instead
  • Realizing that some users could be confused, as they might think that the registration itself is everything they need to do for the moment to start contributing to Wikipedia

As a result of the above issues, we decided to remove the registration step and instead think about more robust methods for data analysis. These methods are listed in the data analysis section below.

Community Discussion[edit]

The community discussion started on June 18 at the Arabic Wikipedia village pump. The largest concern was over the registration process. One user stated his concern of using a gadget to get user names without their consensus, as that might infringe on their privacy, and requested to make the registration process more clear to everyone. We added a confirmation message which says "Would you like to register your username in the Contribution portal? Yes, No?", and the user is still able to see the portal even their name is not registered. Once this was added, the community reached consensus to launch the contribution portal and begin running a site notice.

Data Analysis[edit]

Page Views[edit]

We monitored the number of page views derived from the Wikipedia article traffic stat reports which are updated daily. Although these reports do not reflect the number of unique page visitors, they are a widely accepted indicator for page views, particularly for high-traffic pages.

Main page[edit]

Comparison between the number of landing page views linked from the site notice on Arabic Wikipedia.

The graph on the right shows a comparison of the number of page views between three landing pages linked from three looping messages on the site notice of the Arabic Wikipedia. In the first week, we restricted the view of the contribution portal banner on the site notice to logged-in users only, which attracted an average of 65 views/day, in contrast to 2586 views/day and 1840 views/day for banners advertising landing pages of Kiwix and introductory videos respectively.

In the following week, when the banner was visible to all Wikipedia readers, we saw a considerable increase in the number of contribution portal page views, up to an average of 570 views/day, which brought page views of Kiwix and video pages to an average of 1830 and 1320 views/day respectively. This experiment shows that banners not restricted to logged-in users have about 10 times more access compared to banners shown only to logged-in users. And it makes sense that banners which advertise topics of broad interest to readers (i.e. Kiwix) get more clicks than those that invite readers to learn how to contribute to Wikipedia, as the number of Wikipedia visitors with an intention to read articles will naturally be greater than the number of visitors intending to contribute content.

Tutorial pages[edit]

Comparison between the number of page views of tutorial pages linked from the main contribution portal page.
In order to understand visitors' interest in various help topics, we recorded the number of page views of each tutorial page . The graph on the left shows the number of page views of tutorial pages. Most viewed pages were tutorials on how to create a new article, how to create a user page, and how to fix a typo successively. This suggests that brand new Wikipedia users may first be interested in learning about general topics (e.g. adding contents, introducing themselves, or copyediting) rather than more detailed editing topics (e.g. add a reference, create a redirect). The peak (TG mail) refers to the peak in the number of page views of creating a user page which was sent as a link to about 1000 volunteers withig the scope of Taghreedat project (details to be announced in a separate report).
In order to assess whether order on the page influenced a user's choice of topics, we investigated the relationship between position of tutorial links and the number of page views. The left figure shows the layout of links for the six tutorial pages on the main page of contribution portal. Although the link for creating a user page is located in the second row, in a sub-optimal position, the tutorial page has more page views than the link above it (fixing a typo). This implies that users are indeed accessing pages of interest, rather than clicking randomly on the first available tutorials.
Layout of the six tutorial pages on the main page of the Contribution portal.

Contribution Ratio[edit]

Although page views provide information on how the contribution portal page is being accessed, it does not provide any information on the impact of the information being provided to new users. In order to understand whether viewing the portal leads to increased editing by new users, we adapted a new metric called "Contribution Ratio." The contribution ratio is calculated as the proportion of the number of contributors (registered users with at least one edit) to the number of new registered users. Note that we consider the total number of edits which the user made until the time of extracting the data in Aug, 2012. The graph on the right shows the value of the contribution ratio extracted on a daily basis starting from 2007 on the Arabic Wikipedia. The graph shows that the contribution ratio was roughly in the range between 4% ~ 8% over the past 4 years, which means that only 4 ~ 8 users out off each 100 new accounts made at least one edit to the Arabic Wikipedia.

Contribution ratio (average in red) over the past 5 years.

The value of the contribution ratio (orange) is also reflected in the graph on the right along with the number of new registered user accounts (blue) and the number of new contributors (red). Over the 2007 & 2008, the curves of new users and contribution ratio show a high mutual coherency which implies that a high proportion of the registered users were interested in contributing to Arabic Wikipedia. Another interesting phenomena is the existence of periodical changes mostly aligned with the start of the new school semesters in spring and fall of every year (marked by violet and vinous triangles respectively). The latest period on the graph marks the period of running banners linking to the contribution portal, where we see an increase in the contribution ratio compared to previous months.

Number of daily registered accounts (blue), number of users started contributed each day (red), and the contribution ratio (orange) on Arabic Wikipedia.
For the purpose of demonstrating the effect of the contribution portal on contributions of new users, the value of the contribution ratio over the past three month is shown in the graph on the right. The graph shows three main distinctive regions:
  1. Control 1 : The period before running any banners and before launching the contribution portal
  2. Control 2 : The period of running banners landing on the existing Arabic version of [[Help:Editing]] help documentation
  3. Contribution portal : The period of running banners landing on the contribution portal with step-by-step tutorials on how to contribute to Wikipedia.

We can see that running banners inviting site viewers to contribute to Wikipedia has an impact on the contribution ratio, as the average of contribution ratio increased from being 4.5 % during the Control 1 period to become 5.1% during the Control 2. However, by landing the banners on the contribution portal with more appealing visualized help pages, the contribution ratio increased even further to become 5.7%. This appears to be the combined effect of inviting users to contribute and also providing them with an easy-to-use method of learning how to contribute.

Change in contribution ratio over May ~ July 2012, before launching the contribution portal and after launching it on Arabic Wikipedia.

Next steps[edit]

Planned Scheme of Contribution Portal for Phase 1.

We are working now on creating a new design for Phase 1 of the Contribution Portal which we are planning to launch later in August, to determine whether improving the overall look and feel of the portal has any impact on its use. Phase 1 aims to incorporate a section for enhancing communication between new users and the existing community member using the concept of the Teahouse project. In addition, more tutorial help pages will be added, including some more advanced topics such as how to create a category, how to add images to articles, and how to watch pages. We plan to continue using the local site notice to attract new users to the portal, and add more quality assessment metrics for the data analysis such as the total number of daily edits, or the total number of daily created articles by new users.

Footnotes[edit]