Research:Metrics/threshold(t,n)

Rationale

This metric measures whether the editor performed at least $n$ edits within $t$ days. It can be particularly useful in order to measure early engagement (when $t$ is measured from account creation) or during a certain window of interest (for example in A/B test or a usability test for an editing gadget/feature).

Values of parameters $t$ and $n$

In general the values of the parameters $t$ and $n$ will depend on the requirements of the analysis. For example, an A/B test might set a certain objective in terms of volumes of edits and an observation window and thus both parameters would have a clear definition.

If the objective is to study early engagement, instead, the values of the parameters might be chosen based on the editor class of interest. For example, previous analysis on editor activity showed that extremely active editors with a lifetime average activity between 1 edit every 160 minutes and 16 minutes (which is equivalent to a monthly edit count between 250 and 2,500 edits) reached their peak activity rate of 20 edits per day in the first 200 days since registration.

Pros/Cons

Pros:

• Boolean variable, thus easy to understand and report (e.g. “70% editors in the control group performed at least 10 edits in the two days following registration”).

Cons:

• Depends on two parameters

Technical dependencies

Relevant tables to query:

• Revision
• Article feedback

Also (contains data from mid-2011 onwards):

• UserDailyContribs (already grouped by day)