A/B testing for major interface changes isn't such a good idea, I
think. It's great for small things like login fields that people
aren't likely to notice, but once you get into major interface
elements you're going to get quite a few people confused or
frustrated.
"Why is your Wikipedia different to mine?", "Why does this not look
like the help page says?", etc.
I'm not sure the side-effects would be worth the data.
Andrew.
On 16 July 2014 13:01, Erik Zachte <ezachte(a)wikimedia.org> wrote:
Disclaimer: I haven't been following the
discussion
Full disclosure: 1) I personally like the new feature 2) I value community
opinion a lot
Just my 2 cents on methodology:
It seems to me that if we could present half of a target population with old
and half with new settings from the outset (e.g. by focussing on new users
only), then the outcome would be more convincing.
In the current proposal if 40% would choose to switch , is that because 60%
like the new settings better, or because they can't be bothered at that very
moment to explore and make up their mind, or for any other reason cling to
what they are offered initially.
Erik
From: analytics-bounces(a)lists.wikimedia.org
[mailto:analytics-bounces@lists.wikimedia.org] On Behalf Of Gilles Dubuc
Sent: Wednesday, July 16, 2014 14:00
To: A mailing list for the Analytics Team at WMF and everybody who has an
interest in Wikipedia and analytics.
Subject: Re: [Analytics] Media Viewer User Preference Data
Could we sample only logged-in users?
I think the idea is to measure both separately, because the default might
end up being different for logged-in users and for anons, depending on the
preference ratio for each.
On Wed, Jul 16, 2014 at 7:29 AM, Nuria Ruiz <nuria(a)wikimedia.org> wrote:
a) aim to track total users who enable/disable
Media Viewer, rather than
just events
b) switch to a 3-state preference setting: enabled
/ disabled / default
c) try to measure the total number of users in
each group (instead of daily
events)
I assume we are talking about logging stuff for logged in users and not
logged in users.
If that is the case this strategy will dwarf editor data with reader data,
in the RFC most concerns seem to be about editor workflow. Could we sample
only logged-in users? That way we will get a sample of data that better
represents editor's interaction with media viewer and it will be a lot
easier to quantify our results.
Once we figure out a practical way to implement
this request together, we
will present it to community members to >discuss its impact on the RfC, and
agree on acceptance criteria (e.g. keep Media Viewer enabled by default
unless >a 51% majority of users disables the feature on Enwiki by September
30?).
Perhaps if we want the analytics to impact the RFC we should work with the
community to come up with a survey strategy they find aceptable?
On Tue, Jul 15, 2014 at 11:59 PM, Fabrice Florin <fflorin(a)wikimedia.org>
wrote:
Dear Analytics Team,
As discussed with Toby, Dario and Aaron, we would be grateful for your
guidance to collect user preference data for Media Viewer (1), which is
being challenged in RfCs on English (2) and German (3) Wikipedias, as well
as on Wikimedia Commons (4).
The English RfC closed last week, requesting that we disable Media Viewer
right away. The multimedia team, in consultation with community and legal
teams, is not comfortable disabling the feature on the basis of this small
RfC, which we believe does not accurately represent the views of the
millions of users whom we serve.
To address this issue, we would like to conduct a wider outreach to collect
more accurate data about user preferences than either this RfC or our
optional surveys can provide. We propose to develop a more prominent viewing
options panel (5) that would make it very easy to switch quickly to your
preferred viewing mode, as shown in this prototype (6). Right after launch
of this new feature, all users would be shown this prominent panel and asked
to select their favorite viewing option. Once we have collected and analyzed
that data over a period of a month or two, we would be able to make more
informed decisions with our community on whether or not to keep the feature
enabled -- based on actual responses from all users, rather than
speculation.
With your help, we have already developed a basic dashboard that tracks
Media Viewer opt-in/out events (8), based on clicks on the less prominent
‘Disable’ feature at the bottom of the metadata panel. We propose to modify
this dashboard to support this new initiative, as outlined in this ticket
(7), with these changes:
a) aim to track total users who enable/disable Media Viewer, rather than
just events
b) switch to a 3-state preference setting: enabled / disabled / default
c) try to measure the total number of users in each group (instead of daily
events)
Also note that most users would start in the default state, showing them
Media Viewer, since it is enabled by default — but after 10 or so image
views, the viewing options panel would appear automatically, asking them to
either enable or disable the feature. After they have made their selection,
the panel would remain accessible (but much less prominent) and they can
still switch state (but can't go back to default).
We could use your guidance on the right way of logging total logged-out
users, so we can use them as a basis for percentages: should we log
enabled/disabled state on some specified event/action: page load? thumbnail
click? first site visit of the day? — or do you recommend another method? We
would want to collect this data for a month or two, so we don’t only capture
the immediate responses from the most active users, but also those of less
frequent visitors.
We would also appreciate your comments on the specific tickets (5) and (6),
with any recommendations for improvement. Keep in mind that we would like to
respond quickly to our community’s concerns and have limited resources, so
we would prefer to get this work done in the next week or two. And it will
take a couple months to develop, release and collect enough data -- so even
if we start tomorrow, we may not have conclusive data to share until
September. So time is of the essence. :)
Gergo is spearheading this project, and may have more technical questions
for you. But I wanted to give an overview from a product perspective, before
we dive in to the implementation details. Once we figure out a practical way
to implement this request together, we will present it to community members
to discuss its impact on the RfC, and agree on acceptance criteria (e.g.
keep Media Viewer enabled by default unless a 51% majority of users disables
the feature on Enwiki by September 30?).
Thanks in advance for your advice. We look forward to working with you soon
on this important project, which could impact other features now in
development.
All the best,
Fabrice
(1)
https://www.mediawiki.org/wiki/Multimedia/About_Media_Viewer
(2)
https://en.wikipedia.org/wiki/Wikipedia:Media_Viewer/June_2014_RfC
(3)
https://de.wikipedia.org/wiki/Wikipedia:Meinungsbilder/Medienbetrachter
(4)
https://commons.wikimedia.org/wiki/Commons:Requests_for_comment/Media_Viewe…
(5)
https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/787
(6)
http://pauginer.github.io/prototypes/media-viewer/desk/disabling-settings/i…
(7)
https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/793
_______________________________
Fabrice Florin
Product Manager, Multimedia
Wikimedia Foundation
https://www.mediawiki.org/wiki/User:Fabrice_Florin_(WMF)
_______________________________________________
Analytics mailing list
Analytics(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/analytics
_______________________________________________
Analytics mailing list
Analytics(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/analytics
_______________________________________________
Analytics mailing list
Analytics(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/analytics