Trust in Online Collaborative IS - Community-Built Databases: Research and Development - page 363

anonymous users is 49% [ 55 ]. Furthermore, 70% of the reverted revisions (vandal-
istic content) are associated with anonymous users. Together these results suggest
that user registration has a positive effect on the quality of Wikipedia.

14.4.1.2 Featured Vs. Nonfeatured Articles

By comparing content evolution in both featured and nonfeatured articles, we
aim to find what parameters affect content quality and how the open editing
model of Wikipedia lets featured articles attain high quality. In [ 20 , 44 ], authors
have compared featured and nonfeatured articles and concluded that, on average,
featured articles benefit from higher number of edits and distinct editors. In order to
have a more detailed comparison between featured and nonfeatured articles, we
examined the evolution of content in these articles and extracted the statistics
shown in Table 14.5 .
Although 39.1% of the total inserted tokens in nonfeatured articles are contrib-
uted by anonymous authors, this figure drops to 15.2% when only the last revisions
are considered. In the case of featured articles, the total percentage of inserted
tokens by anonymous authors is 56.3%, with this figure dropping to 7.8% in last
revisions. According to these statistics, most of the remaining content in both
featured and nonfeatured articles belong to registered users, but this percentage is
higher in featured articles. This observation, together with the result in Sect. 14.4.1.1 ,
provides strong evidence for why featured articles contain higher quality content
throughout their lifetime.
Furthermore, the token survival ratio presented in Table 14.5 shows a much
higher turnover of the content in featured articles over time; a higher ratio of tokens
is deleted in featured articles compared to other articles. This might be counterintui-
tive as one might expect that the content inserted in the featured articles should be of
higher quality and therefore more stable. However, it can be interpreted as higher

Table 14.5 Statistical comparison of featured and nonfeatured articles
Featured articles Nonfeatured

articles

Total inserted tokens
Registered 43.7% (before: 50.9%, after: 36.7%) 60.9%

Anonymous 56.3% (before: 49.1%, after: 63.3%) 39.1%

Tokens in the last revisions
Registered 92.2% (before: 93.1%, after: 88.2%) 83.9%

Anonymous 1.5% (before: 6.9%, after: 11.8%) 16.1%

Token survival
Registered 23.1% (before: 33.8%, after: 8.7%) 50.9%
Anonymous 1.5% (before: 2.6%, after: 0.7%) 15.2%
Ratio of reverted revisions 17.8% (before: 11.4%, after: 25.4%) 9.9%
Ratio of revisions submitted by admins 17.4% (before: 20.7%, after: 14.2%) 10.9%
The statistics for featured articles consider revisions submitted before and after the articles were
marked as featured