User:Mike Christie/FACstats monthly data
This page links to pages showing the individual review credits for each month's FACs, starting with November 2023.
Review statistics pages
[edit]Column headings
[edit]- Article: The name of the article. If it's been moved use the new name.
- Archive: The FAC archive number.
- P/A: P for promotion, A for archive.
- Editor: Name of the relevant editor. Always the actual underlying username, never any sig name. The bot does this. For renames, use whatever the bot uses with an exception or two noted below.
- Role: The next three columns are the ones that need judgement. Role can be one of N for nominator or C for coord. Anyone else should be blank.
- Type: Blank for a content review, S for source review, I for image/media review, A for acccessibility review.
- Declaration:
- Blank if no declaration was ever made
- S for support
- OS for oppose converted to support
- SS for a struck support
- SO for a struck oppose
- Supports converted to oppose are not tracked.
- Restart (only on older tables). This appears if the nomination was restarted, and is an integer (always 1 or 2). A "1" means the review was done before the restart; a "2" means the review was done after the restart.
Instructions for correcting FAC data
[edit]The monthly files are initialized by a bot that guesses what the contribution of each editor in the edit history. The bot attempts to identify coords, image and source reviews, and supports and opposes. It always considers the creator of each FAC archive to be the nominator. There are always changes needed to the file to correct it. The easiest way to do this is to have two monitors available, and open the featured log and then the archive log one screen, and the monthly file from the table above on the other screen. (I strongly recommend using the visual editor for this; editing tables without VE is much more cumbersome.) With two screens, one can scroll each page, using the logs to check that the monthly file is correct, and fix it where necessary.
Here is a list of possible corrections that should be checked for. See the section above for the definition of the abbreviations.
- The bot assumes the person opening the FAC is the nominator so this is very rarely wrong. However, check for a conominator -- they get an "N" in the role column. The other columns for a conominator should be left blank for clarity; they will be ignored if the role is "N".
- Editors only get one row per FAC except when they have two or more roles -- S and blank (source review and content review), or S, blank, and I, for example. If they're a nominator blank out or overwrite any C or review records (this should never happen for the initial nominator, but it could happen for a conominator). When a coordinator nominates an article it should correctly flag it as an N role rather than a C, but correct it if it gets that wrong.
- C role records are for coords; these are automatically marked by the bot though it tries to detect if they have recused. If they recuse and the bot gets it wrong this is a source of possible errors. The threshold for counting a coord comment as a recused review is a judgement call -- don't bother if it's just a couple of MoS questions at the end as they're considering promotion, but after, say four or five comments I would usually credit a review. The bot will ignore C records; they're only there for convenience if they need to be converted to an S type or blank (content) or I (image) or A (accessibility), and in those cases make sure to get rid of the C on that row.
- Put an S in the role (indicating a source review) if:
- The reviewer declares or creates a "source review" or "spotchecks" or similar section name
- The review does a significant follow up to help out in that section. There is some nuance here. E.g. Editor A might post half-a-dozen comments in this section so create an S record for them, but further down it is apparent they did a full content review. In those case it's a judgement call; delete the source review record (or just delete the S to change it to a blank so it's a content review) or add an additional review row for the content review. There is usually no need to credit with a full S review in most such cases. If a reviewer who has not otherwise contributed to the FAC puts in such a comment, keep the record but use an S only if it is substantial. If more than one person does a source review, even partial, they each get a credit.
- Some reviewers will include source formatting in a content review. This is a judgement call again; credit the reviewer with both a content review and source review if the source review is sufficiently substantial.
- Image reviews get an I, and are subject to the same comments as source reviews regarding additional commenters on image topics.
- A row in the table should be deleted completely if the comment is not a review. Anyone whose *only* comment is not intended to help with promotion/archiving decisions can be deleted -- e.g. "can we move this to the talk page" or "please don't use tq". This may require checking the history of the FAC to find the edits, particularly if it was just a formatting edit. If I'm having trouble figuring out what an editor contributed, I usually open the edit history on another tab and search for their name, and use popups to see what they did. Note there are some editors such as Bruce1ee and Jonesey95 who never review but who fix linter errors in signatures, and their names will show up as a result. Make sure to delete these; that is, for every name you see on the monthly page, check that there is actually a review of some kind, or else delete that row.
- Review comments on talk page. There *should* always be a note and a link, so I often forget to do this, but fortunately these are now rare (and are not usually germane to the review where they do exist, so would not count anyway).
- Collapsed sections. Always expand them if there's any doubt about whether reviewer X was an S or I. Occasionally a collapsed section will have a comment by an editor who did not comment elsewhere.
- Declarations. Enter any bolded declarations. If it's clear that there was a prior oppose or support make it OS (oppose converted to support) or SS (struck support) as appropriate. There's no special notation for a support converted to an oppose; it's just an O. If a user says something like "you can count this as a support/oppose" rather than bolding, include those declarations. If in doubt try to determine if a coord would consider it a declaration. Any call to withdraw expressed in a way a coord would pay attention to is counted as an oppose. I have not been consistent about putting in "support" for source and image reviews; I used to include them but have not generally done so for the last three or four years at least. That is, even if a source reviewer enters a bolded support, I usually don't put an S in the declarations column. It's OK to do so, though, as nothing depeneds on this; it's just a habit derived from some source/image reviewers preference for not making a declaration and just leaving the information for other reviewers to see. However, an oppose or struck oppose (for source and image reviews) should be entered, just as for content reviews. An oppose converted to a support for source/image/accessibility reviews could be entered as OS or SO, as you prefer.
TL;DR: Overall this sounds complicated but in fact the common sense rule "Make each row reflect what that reviewer contributed to the review" covers almost every case.