User:Mr.Z-man/newusers

I ran a script that looked at all the user accounts created in February 2010. It looked at whether the user's first edit was to create a page, whether their first edit was deleted, whether their first edit was to a mainspace page, and whether or not the user was still active (that is, made any edits during the 30 days before the script ran on 22 September 2010). The results are below:

Summary

 * Users who create an article are much more likely to leave the project if their article is deleted. 1 in ~160 will stay if their article is deleted while 1 in ~22 will stay otherwise
 * Our main source of regular users is not users who start by creating an article.
 * Users who start by editing an existing page outnumber article creators by 3:1
 * Users who start by editing an existing article are far less likely to have their first edits deleted (and therefore are far more likely to stay)
 * From the users analyzed, we got fewer than 200 regular users from those who created articles (1.3% retention), we got more than 900 from users who edited existing pages (2.5% retention)
 * A significant number of regular users (24%) get their start outside of mainspace.

The script
For performance reasons, the user accounts were determined from the new user log, rather than the user table directly. The script then gathered basic information about each account - editcount, user groups, block status. Users with a bot flag, 0 edits, or those that were indefinitely blocked were excluded. Users with indef blocks were excluded, as these are likely to be mainly vandals and sockpuppets rather than legitimate new users; a user with a shorter block may still be a regular user. There may be a small amount of error in the determination of whether or not a user created a deleted page (false negatives). If a user created a deleted page that was created by someone else and deleted, they won't have the earliest deleted edit for that title, so they won't appear to be the creator, even if they did create a version of it. Whether a user is still active was determined based on whether or not they had made any action in recentchanges.

The results

 * Total number of accounts: 226099
 * No edits: 155820
 * Indef blocked: 5473
 * Bots: 3
 * Errors (likely due to an account being renamed + rename log oversight): 1