The team wants to do a skin-specific analysis for the A/B test for Talk pages project/Usability. This means we have to know whether someone is using Vector 2022 vs Vector 2010. (We don't necessarily expect to have enough users to do a specific analysis for MonoBook, Timeless, or other skins.)
Description
Description
Details
Details
| Subject | Repo | Branch | Lines +/- | |
|---|---|---|---|---|
| Add new fragment for editattemptstep | schemas/event/secondary | master | +603 -3 |
Related Objects
Related Objects
Event Timeline
Comment Actions
This turns out to be doable without needing any modifications -- the existing logging to the DesktopWebUIActionsTracking schema includes the skin data, and it can be joined to the EditAttemptStep logging via the shared pageToken/page_token keys.
@MNeisler if you do need it added to the EditAttemptStep logging directly I could do so.
Comment Actions
@MNeisler if you do need it added to the EditAttemptStep logging directly I could do so.
@DLynch If possible, it would still be helpful to get a skin field added to EditAttemptStep directly to help make querying easier and not always limit us just to the events sampled in DesktopWebUIActions for the planned editing-related metrics. (Note: I will still need to join with DesktopWebUIActionsTracking for engagement metrics (i.e.proportion of talk page views that lead to clicks)).
I also think this would be a useful field to have in EditAttemptStep for other future product analyses where we want to calculate editing metrics by skin type.
Comment Actions
Change 919157 had a related patch set uploaded (by Jdlrobson; author: Kimberly Sarabia):
[schemas/event/secondary@master] Add new fragment for editattemptstep
Comment Actions
Change 919157 merged by jenkins-bot:
[schemas/event/secondary@master] Add new fragment for editattemptstep
Comment Actions
I was checking data logged for the Usability Improvements AB test and noticed that the skin field added to EditAttemptStep is currently only returning NULL values. It looks like a task (T337270) has already been filed to address this. Pending a fix to this issue, I should be able join AB test data in EditattemptStep to data logged in DesktopWebUIActionsTracking to obtain the desktop skin being used by the editor.