Research talk:Ideas/How does the availability of VisualEditor affect the rate of edits and editor acquisition in different languages?

Query[edit]

I used the following query to find how VE affects new editors with the following definiton: VE editor - more than 10 edits (you can use wikitext/ve-edits ratio instead if you want). Notice: number of edits is all edits (not just ve edits) since we expect that a newcomer that gets into editing using VE learn to use wikitext after sometime.

select 
VEEditor,count(*) Editors, avg(revs) AvgEdits, avg(Age) AvgSpan
from
   ( select
             user_name, min(if(ct_tag ='visualeditor',rev_timestamp-user_registration,9999999999)) AgeForFirstVeEdit,
             max(rev_timestamp)-user_registration Age, sum(if(ct_tag ='visualeditor', 1, 0))/count(*) VeEditRate, count(*) as revs,
              if(sum(if(ct_tag ='visualeditor', 1, 0))>10,1,0) VEEditor 
             from user 
             inner join revision 
                 on rev_user=user_id 
             left join change_tag
                on ct_rev_id=rev_id where user_registration>20140101000000 group by user_id)  veEditTable 
    group by VEEditor;

hewiki result:

VEEditor	Editors	AvgEdits	AvgSpan
0	5434	10.1108	39846084.43393449
1	138	82.7681	186499149.5652174

frwiki:

VEEditor	Editors	AvgEdits	AvgSpan
0	39316	10.9991	27405708.73588361
1	904	65.5077	150631463.78539824

eranroz (talk) 22:30, 31 July 2014 (UTC)[reply]

Additional figures:

Additional data as of 2014-12-15
Wiki	VEEditor	Editors	AvgEdits	AvgSpan
itwiki	0	28922	17.4515	56912064.03682318
itwiki	1	853	88.7644	279884956.05861664
plwiki	0	10975	10.8118	58964831.483371295
plwiki	1	289	123.0104	303089996.9757785
ptwiki	0	32660	8.7297	37167646.98772199
ptwiki	1	699	89.2933	236829436.07296136
By way of contrast, a non-default wiki:
enwiki	0	613350	9.2777	45204917.99928263
enwiki	1	810	177.8963	290263299.2209877

Is this sound data, however?

Jdforrester (WMF) (talk) 19:40, 15 December 2014 (UTC)[reply]

Jdforrester (WMF), there is some bias in the query I gave above, as it count also users who did less than 10 edits in the regular edits but not in VE editors. Maybe adding "having count(*)>10" would be somewhat less bias:

select 
VEEditor,count(*) Editors, avg(revs) AvgEdits, avg(Age) AvgSpan
from
   ( select
             user_name, min(if(ct_tag ='visualeditor',rev_timestamp-user_registration,9999999999)) AgeForFirstVeEdit,
             max(rev_timestamp)-user_registration Age, sum(if(ct_tag ='visualeditor', 1, 0))/count(*) VeEditRate, count(*) as revs,
              if(sum(if(ct_tag ='visualeditor', 1, 0))>10,1,0) VEEditor 
             from user 
             inner join revision 
                 on rev_user=user_id 
             left join change_tag
                on ct_rev_id=rev_id where user_registration>20140101000000 group by user_id
                having count(*)>10
)  veEditTable 
    group by VEEditor;

eranroz (talk) 20:15, 15 December 2014 (UTC)[reply]

… using tweaked script
Wiki	VEEditor	Editors	AvgEdits	AvgSpan
itwiki	0	3652	117.5088	190036235.92141291
itwiki	1	853	88.7784	279998008.077374
plwiki	0	1202	76.6381	189069045.58319467
plwiki	1	289	123.0415	303090212.0380623
ptwiki	0	3022	67.5242	148893591.62210456
ptwiki	1	700	89.2400	236790051.04714286

Jdforrester (WMF) (talk) 20:32, 15 December 2014 (UTC)[reply]

Too early?[edit]

Someone has suggested that it's too early to assess VisualEditor's impact on such metrics yet. Example (if I understand correctly): «[...] not even to the point where we're really talking about usablility problems. These are tablestakes problems (i.e., load time)». --Nemo 13:48, 15 December 2014 (UTC)[reply]

Why would it be too early? Hundreds, if not thousands, of edits are made using VE every day in the French Wikipedia alone, and thousands more in other languages. I'd say that this raw data is pretty useful.

Some people may also ignore that unlike in the English Wikipedia, in many languages VE is enabled by default, even for anons, and it's much more accessible. --Amir E. Aharoni (talk) 14:37, 15 December 2014 (UTC)[reply]

Not too early :-) Research:VisualEditor's effect on newly registered editors/Results was in june 2013 and should be repeated. For reference [1], French Wikipedia VE metrics dashboard, Hebrew Wikipedia VE metrics dashboard, Italian Wikipedia VE metrics dashboard etc. --Atlasowa (talk) 18:04, 15 December 2014 (UTC)[reply]

Also: Research:Visual editor usage (January, 2014) --Atlasowa (talk) 18:30, 15 December 2014 (UTC)[reply]

It's never too early to gather data, but the title of this page asks an interpretation. --Nemo 21:27, 8 January 2015 (UTC)[reply]