Wikipedia talk:Requests for arbitration/Statistics

Thanks
Thanks heaps for the work and time taken in producing this report - I'd love to read any analysis from one smarter than I :-) Privatemusings (talk) 02:45, 6 February 2009 (UTC)
 * Thanks - me too. Paul August &#9742; 17:52, 6 February 2009 (UTC)
 * Kudos for this, Paul. You've done some useful work here. AGK 12:29, 19 May 2009 (UTC)

Changes 12 Mar 09

 * (Copied from Wikipedia talk:Requests for arbitration)

FYI: I've made some changes to Requests for arbitration/Statistics:
 * 1) All the "Arb activity" tables are now sortable (for example see: Arb activity (2009) — requests).
 * 2) The "Cases" section now tracks each case's drafter (see: Cases involving 2009 arbs only) and Arb activity (2009) — cases).
 * 3) The "Proposals" section now tracks the time order of each action (i.e. a support, oppose, or abstain) on each proposal. This is represented in the tables by appending to each "S", "O" and "A", a number indicating the order that each action occurred, (i.e 1 = first, 2 = second etc.). This allows for computing two new statistics for each arb, "firsts", which is the number of first actions -- generally indicative of being the drafter of the proposal -- and "AVR" (average vote rank), which is the average of the rank orders of an arb's actions on a proposal, following the first action (i.e. the average of the ranks > 1) -- giving an indicator of earlier versus later voting. So, for example for the five cases closed so far this year, encompassing 105 proposals, Coren with 38 "firsts" has apparently drafted about 36% of those 105 proposals, while Rlevse is on average the earliest voter with an AVR of 4.1 (see: Arb activity (2009) — case proposals).

Paul August &#9742; 19:40, 12 March 2009 (UTC)

Changes 19 Mar 09
I've decided to eliminate the vote rank for case proposals. Instead, for each arb's action on a proposal (i.e. a support, oppose or abstain), the page now tracks the length of time between that action and the first action on that proposal. I've restructured the proposals data tables to include a second set of arb columns. The first set, as before contains the votes and status data, but without the vote rank data, similar to how the data table looked originally, except that a new bit of data is recorded, an "S1" denotes the first support, which is used as an indicator of proposal drafting (instead of first action). The second set of arb columns, tracks the time of vote data. Paul August &#9742; 19:08, 19 March 2009 (UTC)

2008 data
I've completed collecting and tabulating arbitration statistics for 2008. The data is posted here:

Requests for arbitration/Statistics 2008

Some highlights follow.

The 2008 Arbitration Committee considered 255 requests, voted on 53 56 motions, publicly heard 38 cases, and drafted and voted on 841 case proposals.

There were 152 case requests, open on average for 4.6 days, with 67% declined, 23% accepted, 8% withdrawn, and two disposed of by motion. There were 103 clarifications and other requests, open on average for two weeks, with the longest duration being nearly three months. The motions were open on average for 12 days, with 47% passing. The cases were open on average for 39 days, with the longest being open for just over four months.

Here is a smattering of individual arb statistics. On case requests, Flo has the honor of having the highest voting percentage of 85, followed closely by Sam with 82 and Brad with 78, against an average of 41 and a low of 6. Kirill drafted almost half of the cases, Brad though missing for a quarter of the year, drafted nearly a quarter of the cases. Kirill, Sam and Flo each acted on about 95% of case proposals, compared with an average of 74 and a low of 42. Kirill, Brad and Flo were on average the quickest to act on case proposals. (On a personal note I find that on most measures I myself was solidly mediocre. There was one area however in which I did excel -- declining case requests -- which I did a remarkable 93% of the time. This might be attributed either to judicial conservatism or laziness, take your pick.)

Having the 2008 data now allows comparison between years. For example the average case request duration is surprisingly the same for both years at 4.6 days. There are differences. For example, so far this year there have been 24 motions offered and voted on, versus only 53 56 for all of 2008.

Paul August &#9742; 16:02, 3 April 2009 (UTC)
 * Errata: I've just discovered three "missing" 2008 motions, so I've updated the above remark. Paul August &#9742; 15:46, 23 July 2009 (UTC)

Request/suggestion
One thing which might be interesting is if some analysis can be done of case voting to see if there were patterns between arbitrators in tending to support the same proposals - particularly where the committee was divided. I certainly had the feeling that there were some other members of the committee whom I always tended to be agreeing with, though it never rose to the level of a US Supreme Court-style divide between the Liberal and Conservative wings. Would this be an easy thing for someone to do? Sam Blacketer (talk) 22:40, 18 June 2009 (UTC)

2009 half-year summary
Through the first half of 2009, the Arbitration Committee considered 111 requests, voted on 41 motions, publicly heard 19 cases, and drafted and voted on 928 case proposals.

There were 59 case requests, open on average for five days, with 63% declined, 29% accepted, 7% disposed by motion and one withdrawn. There were 52 clarifications and other requests, open on average for 11 days. Motions were open on average for four days, with 61% passing. Cases were open on average for 60 days, with the longest being open for nearly five and a half months.

With respect to case requests, Wizardman had the highest voting percentage of 93, followed closely by Casliber with 89, against an average of 61, and a low of zero. Coren and Stephen Bain shared the highest accept percentage of 53, and Vassyana had the highest decline percentage of 80. With respect to clarifications and other requests, Vassyana commented on the most (32). Overall, Carcharoth could be considered to have been the most active, with the lowest DNA (did not act) percentage of 18, against an average of 47 and a high of 86. With respect to motions, John Vandenberg had the highest voting percentage of 79, with many arbs following closely behind, against an average of 59 and a low of 18.

Wizardman drafted over a quarter of the cases (five) with Newyorkbrad drafting three, while John Vandenberg drafted the largest case, as measured by number of proposals (186, 100 passing), and Rlevse drafted the second largest (132, 98 passing). Six arbs were virtually tied for the highest case proposal voting percentage: Carcharoth and Sam Blacketer, (with 94) and FloNight, Kirill Lokshin, Rlevse, and Wizardman (with 93). Arguably, Rlevse was on average the quickest to act on case proposals, with Casliber second.

Comparing Apr-Jun with Jan-Mar, the number of case requests declined 36% and the percentage of cases accepted declined 15%, thus the number of cases accepted was almost halved. The number of clarifications and other requests increased 74%, the number of cases closed more than doubled, and the number of case proposals considered increased by more than five times.

Comparing the first half of 2009, with the first half of 2008, the number of case requests dropped dramatically from 105 to 59, continuing an apparent trend, and the number cases heard dropped from 27 to 19, while the average case duration nearly doubled from 32 days to 60.

Paul August &#9742; 19:35, 2 July 2009 (UTC)

Excellent Resource
I have been editing Wikipedia for years and have been taking an interest in WP-space issues for months, and I am just now discovering this excellent analytic tool. It helps us better appreciate the hard work of all our volunteer arbitrators. Thanks! — James Kalmar  07:34, 21 January 2010 (UTC)