>As Erik says, we don't have unique-reader numbers. We used to collect
>data via ComScore but this was dropped as it was felt to be inaccurate
>(though I'm not familiar with the details of exactly what was wrong
>with it) 
The biggest inaccuracy of comscore data was that it did not measure mobile usage, we know from our own metrics that about half of our pageviews come from mobile devices and, in many instances, for some wikis, mobile access represents >80% of access,  so comscore was grossly underestimating unique users. 

Reader team sends a newsletter to wikitech-l@ with this kind of information that might be helpful. See newsletter from Mayl: https://lists.wikimedia.org/pipermail/wikitech-l/2016-May/085465.html


On Thu, May 19, 2016 at 10:06 AM, Andrew Gray <andrew.gray@dunelm.org.uk> wrote:
Hi Nathan,

As Erik says, we don't have unique-reader numbers. We used to collect
data via ComScore but this was dropped as it was felt to be inaccurate
(though I'm not familiar with the details of exactly what was wrong
with it) - see https://meta.wikimedia.org/wiki/User:Stu/comScore_data_on_Wikimedia
& https://meta.wikimedia.org/wiki/ComScore/Announcement

For example, as of Jan 2010, estimates for some languages were:

* English, 190m/month
* Japanese, 35m/month
* Spanish, 32m/month
* French, 26m/month
* German, 25m/month

You might be able to use this, coupled with contemporary editor
numbers, to get a rough approximation for activity rates *at that
time*.

Incidentally, the old stats interface has a very rough "active editors
per million *speakers*", which isn't at all the same thing (not all
speakers use the internet, & not all internet users read WP) but may
be interesting - https://stats.wikimedia.org/EN/Sitemap.htm?sortcol=9D

Numbers more that 50/million (or 0.05% of speakers) are rare, and most
are statistical anomalies driven by very small languages. However,
there's a few with five to ten million speakers where we have fifty to
a hundred editors per million speakers (eg Czech, Catalan, Swedish,
Hebrew, Norwegian, Finnish). Note that these figures are for active
editors (5+ edits/month) rather than "any editors", so the true
participation rate will be higher.

I used to use a figure of "one in five thousand" for English Wikipedia
in talks, but it looks like this was "active editors per speakers",
and so definitely an overestimate. At a wild ballpark guess, maybe one
in a thousand readers is an editor? Very curious to see what you come
up with - please do let us know!

Andrew.

On 19 May 2016 at 00:06, Nathan Marwell <nbmarwell@gmail.com> wrote:
> Hello,
>
> I am an economist working on a research project analyzing contribution
> behavior on Wikipedia, and am interested in computing the fraction of
> individuals who use the site and make contributions.  I have found the data
> on daily contributions, and am now looking for data on the number of
> individuals who use Wikipedia. I believe I have found the data I am looking
> for here:
>
> https://stats.wikimedia.org/EN/TablesUsageVisits.htm
>
> Unfortunately, the data contained in that file only covers the period from
> August 2002 through October 2004. Does a similar database exist for later
> time periods?  Any information you can provide would be greatly appreciated.
> Thank you for you time and I look forward to hearing from you.
>
> Sincerely,
> Nathan Marwell
>
> _______________________________________________
> Analytics mailing list
> Analytics@lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/analytics
>



--
- Andrew Gray
  andrew.gray@dunelm.org.uk

_______________________________________________
Analytics mailing list
Analytics@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/analytics