The majority of program leaders reported about content production. Reported events averaged almost 24,000 characters in total, with individual participants averaging almost 3 pages of characters added each.
File:EditathonMap.JPGParticipants at the reported edit-a-thons averaged almost 3 pages of content each.
The majority of program leaders were able to provide us with the amount of new content that got added to Wikipedia's article namespace during the event.[2] Events produced an average 23,993 characters with the most productive event reporting 157,586 characters of text produced during the event.[3]
In order to make the metrics easier to understand, we converted "characters" into "printed pages," assuming that one printed page equals 1,500 characters. Printed pages of text produced during edit-a-thons ranged from .1 to 8.6 pages an hour with an average of 2.9[4] and from 0.1 to 3.8 pages of text per participant with an average of nearly three-quarters a page of text (0.7 pages) per participant. [5]
Reported events averaged three media uploads each. Half reported had no or unknown numbers of media uploads.
Regarding image uploads, the average was 3 uploads per event, with 11 edit-a-thons reporting with 0 or no known uploads. The largest reported number of uploads was 85.[6]
Content produced at events varies, and even smaller events—participant or budget wise—can produce lots of content. Events with more participants generally produce more content, but the cost of the event doesn't necessarily effect participation rate or content production rates.
We were able to analyze the dollars to content produced for only 5 edit-a-thons. This is because only those 5 reported both the budget and the amount of content added during the event. We looked at the budgets and characters added at those five events and were able to determine how much one "printed page" added costs. The average cost for one printed page of content for these five events was $17.15 US (see Graphs 3 and 4).[7]
Graph 3: Dollars to pages This bubble graph depicts how much content gets produced based on budget and participation rates. The size of the bubble, and corresponding number they are labeled with, show how many total printed pages were produced at the event. It shows that the amount of content created varies, meaning it's not clear if participation or budget makes a difference in how much content is produced. Even the events that had small or zero budgets were productive regarding content production. The graph does show that the more participants, the more content is created, but the amount of money it costs to produce the event does not relate to the amount of participants or content produced.
Graph 4: Dollars to pages This box plot depicts the distribution of how many dollars were invested for each printed page added during the event. As illustrated, by the long vertical line running from the the low of $7.08 to high of $153.44, results were highly variable. However, the 50% of reports occurring around the median of $17.15 ranged from $8.86 to $23.28 per page of bytes added, a somewhat smaller range for what is most typical for these types of events.
During edit-a-thons, it takes little over one hour to produce one page of content.
We also wanted to know how many hours it took to produce one page of content based on the hours that were put into implementing an event. Reports of staff and volunteer hours input into edit-a-thons was available for 11 of the events reviewed.[8] Using those reports, we were able to take those hours and the amount of characters added during those 11 events and calculate that the average was 1.11 hours to produce one page of content. The smallest amount of time was 0.25 hours and 4.5 hours was the largest amount of time to produce a page worth of content[9] (see Graphs 5 and 6).
Graph 5: Hours to pages As illustrated in the bubble chart, higher number of text pages added was not always associated with higher number of participants and more hours input to implementation did not necessarily related to more participants or more content added for the 11 edit-a-thons for which data are available.
Graph 6: Hours to pages This box plot depicts the distribution of how many hours were invested into edit-a-thon events per page created (by bytes added), reports were highly varied and ranged from 0.25 hours to 4.5 hours invested per page of text added. However, the 50% of reported values occurring around the median of 1.1 hours ranged from 0.63 to 2.25, a much closer approximation of the most typical range for these events.
Hourly productivity at edit-a-thons is all over the place, regardless of size and length. Edit-a-thons with lots of new editors can also be just as productive as those with lots of experienced editors.
We also looked into how productive new editors at the events were, to see if the budget and hours put into implementing the edit-a-thons were supporting new editors to be productive at the event (see Graph 7).
File:Productivity per hour.pngGraph 7: Hourly productivity by participants This bubble graph shows how the number of participants and proportion of new account creators relate to the number of pages of text produced each edit-a-thon hour. As illustrated by bubble size and label, in which larger bubbles represent more pages of text each hour, there was a lot of variation in the number of pages of text produced per hour. The two largest bubbles demonstrate an interesting thing, that high proportions of new editors in an edit-a-thon group does not necessarily lead to less productivity, new user groups can be highly productive in terms of adding content to the article namespace.
The most commonly reported data about production at edit-a-thons was about article creation/improvement. This is followed by media and article quality, respectively.
The majority (72%) edit-a-thons reported included how many pages were created or improved on wiki during the event. In total, 620 pages were created or improved during the 46 edit-a-thons! It was also reported that 334 images or media were added, and 81 of those were added to project pages. 7 of the edit-a-thons reported producing Good Articles, totaling 51. Two of the edit-a-thons reported 4 Featured Articles coming out of the event (see Graph 8).
File:Increasing quality.pngGraph 8: Edit-a-thons bar graph: increasing quality The bars on the graph illustrated the total number of pages created, photos added, photos used on other Wikimedia projects, good articles, and featured articles. The largest impacts in terms of quality include that a total of 620 pages were created or improved, 51 of which were rated as “good articles” and 4 that became "featured articles." During the edit-a-thons, there were also 81 photos incorporated into Wikipedia articles.
Recruitment and retention of new editors
The majority of the 46 edit-a-thons reported were able to pull data about retention. Out of 328 new editors who attended these edit-a-thons, after 6-months, only three were actively editing.
We also wanted to learn about the retention of active editors after they attended an edit-a-thon. In the survey, we asked program leaders to report the retention of their active editors 3 and 6 months after the end of the event. A retained active editor is considered a Wikipedia editor making an average of five or more edits a month. Out of the 46 edit-a-thons reported on, 37 of them (80%) had reached or passed the 3 months after their event end date. 29 of the 46 (63%) passed their 6 month mark, so we were able to gather retention data for those edit-a-thons only.
In total, the edit-a-thons reported attracted 328 new editors (36% of 906 total participants). The number of active editors (5+ edits/month) at 6-months follow-up time were reported for 27 of the edit-a-thons (59%), 15 of the events (33%) had not yet reached the point of 6-month follow-up.[10]Only 18 reports provided a separate count for number of active editors at 6-months follow-up for new editors (39%). Active editor retention rates for new users were most often 0% (83% of the reports).[11] For all edit-a-thons reported, the total number of new user participants was 328, of which, only 3 were active editors at 6 month follow-up, 1.4% of new users who had reached the 6-month follow-up window (see Graph 9).
File:New recruit and notice.pngGraph 9: Recruitment and retention The bars on this graph summarize the total number of edit-a-thon participants who created new user accounts as part of the event and the 6-month retention rates for those new users. For all edit-a-thons reported, the total number of new user participants was 328, of which, only 3 were active editors (i.e. 5+ edits per month) 6 months after the event.
Replication and shared learning
Promotional materials can help not only attract participants, but also help others replicate edit-a-thons successfully.
Edit-a-thon program leaders are pro-active at producing materials, blogs, online resources, and other information related to their event which can help others implement their own events.
Finally, we asked program leaders to share with us how replicable their edit-a-thons could be, and what types of shared learning resources were produced for and after the event. This allows us to learn if the reporting program leaders considered themselves experienced in implementing edit-a-thons, which would allow them to perhaps help others design and implement edit-a-thons of their own. We also are able to learn how program leaders and others (i.e. chapters, press, bloggers, etc) were covering the events, and if resources were available for others to use to produce their own events.
For the 46 edit-a-thons reported by program leaders, we learned that the majority (96%) are experienced at producing edit-a-thons and could help others conduct their own. The majority (62%) reported having blogs or other online information available for others to learn more about the event. A smaller amount of program leaders reported that they developed brochures and/or printed materials (27%) and guides or instructions on how to contribute to Wikipedia for event participants (23%) (see Graph 10).
File:Replication and learning.pngGraph 10: replication and learning The biggest strengths that edit-a-thons demonstrated in terms of potential replication and shared learning were that 96% of the events were run by an experienced program leader who could help to guide others and 62% had blogs or online resources for others to learn from. Additionally, 27% reported developed brochures and printed materials and 23% reported they had guides and instructions to inform others how they could implement a similar program.
References
↑Note: Although "content production" is a direct product of the program event itself and technically a program output rather than outcome most of the program leaders who participated in the logic modeling session felt this direct product was the target outcome for their programming. To honor this community perspective, we include it as an outcome along with quality improvement and retention of "active " editors
↑We transformed the WikiMetrics metric "bytes added" to character count in order improve understandability. In most European languages, one byte equals one character.
↑65% of program leaders reported total number of bytes added during their Edit-a-thon event, which ranged from 991 to 157,586 bytes with an average of 23,992.5 bytes (Mean=29,862, Standard deviation=32,576).
↑67% of of reports also included the number of photos added which ranged from 0 to 85 with an average of 3. (Mean=11, Standard deviation=22)
↑Calculating the Pages of Text Added (in bytes) to the Dollars invested, the cost per Pages of Text Added (in bytes) ranged from $7.08 to $153.44 with an average of $17.15 (Mean= $41.96, Standard deviation=$63)
↑(24% of all reviewed, 42% of those directly reported)
↑the cost per pages of text added (in bytes) ranged from to 0.25 to 4.5 hours with an average of 1.11 hours (Mean= 1.52, Standard deviation= 1)
↑For the 15 edit-a-thons that had not yet reached the 6-month follow-up time, there were 173 existing, and 110 new editors not yet eligible for 6-month retention follow-up
↑At 6-month follow-up retention rates ranged from 0% to 100% with an average of 0% (Mean= 7%, Standard deviation= 24%).
