Re: [kde-community] Please participate in our survey for input on KDE's Mission
- Date: Wed, 29 Jun 2016 13:03:34 +0200
- From: Thomas Pfeiffer <thomas.pfeiffer@xxxxxxx>
- Subject: Re: [kde-community] Please participate in our survey for input on KDE's Mission
On 28.06.2016 23:50, Sven Brauch wrote:
On 28/06/16 23:16, Thomas Pfeiffer wrote:
It looks like especially WikiToLearn, but also Applications are a bit
underrepresented compared to how many people are contributing to them overall.
I'm now also sending this to the wikitolearn mailing list (yes, I should have
done that before) and hope that we'll get more community members to participate
in the survey, especially (but not only) applications and WtL contributors.
Not exactly what a statistican would call a "randomly sampled poll"
I trust you are aware of the biasing the amount of participants from the
different areas introduces and aim to provide the results of the poll in
a form which is transparent with respect to this bias.
Actually, the goal in this case would not be a random sample, but a full census,
which unfortunately won't be possible, though.
A truly random sample is not possible either, however, because the overall
is too small for random sampling to work properly and the participation was
Neither is a stratified sample, because we don't even know the size of either
population nor the sub-groups.
Note that "Invite everyone and see who participates" is not random sampling, either.
It is biased by who decides to participate, but also by who actually receives
For WikiToLearn, for example, the problem is that many of them are probably not
subscribed to any of the lists I sent the original invitation to (although it
would be great
if all of them could subscribe to kde-community), which introduces a bias by itself.
The only true random sampling would be if I had a list of _everyone_ in the
would select a sample at random from that, and then _forced_ everyone in the sample
to participate and not allowed anyone else to do so.
Therefore, a statistically valid survey is not possible in this case. Which is
not too big a
problem, however, because we're not doing market or scientific research here.
The only purpose of this survey was to get input from more than the handful of
who had been discussing on the mailing list get a feeling for what the majority
The survey is not binding in any way. In the end, it will be the same as always
in KDE: Who does
the work, decides. I just wanted us to decide based on more than our own personal
Of course I will offer an analysis split by the contribution areas. However, 11
a very small sample, and we could do much better there given the size of the WtL
The sub-sample for Applications is clearly big enough to do statistics on them,
it just felt to me that there must be far more people contributing to KDE
took the survey so far, so I wanted to invite them specifically as well.
I hope this clarified things a bit.