| Casey,
On the statistics: No two samples drawn from a population, using an identical sampling
method, will come up with the same results. The sample results themselves follow a
probability distribution (in this case, the binomial distribution).
Results will average out over time, i.e. drawing a large number of samples will eventually
yield an average result that will more and more closely reflect the true percentage in the
population, but it is impossible to say whether a particular sample result corresponds to
the true percentage, or whether it is higher or lower than that
percentage.
If you have a box containing 100 red and 900 green balls and blindly draw a sample of ten,
not every sample will contain 1 red and 9 green balls. Many will contain no red balls, others
will contain 2, or 3, or once in a while even 10. The results will only average out over time,
once many samples have been drawn. Confidence intervals can be calculated, based on
sample size, to indicate that with 95% or 99% confidence the true percentage is within a
a given range, but even these are just based on probabilities. The bigger the sample, the
narrower (i.e. more precise) the confidence interval becomes.
In this case, the population may have changed since
the last survey, and the survey did not
even use the same sampling method, adding a further source of variation. Calculating a
confidence interval may be useful though; usually statistical results are given with upper and
lower confidence limits.
Andreas
--- On Sat, 2/7/11, Casey Brown <lists@caseybrown.org> wrote:
From: Casey Brown <lists@caseybrown.org>
Subject: Re: [Gendergap] New Survey: 9% female editorsCc: "Mani Pande" <mpande@wikimedia.org>
Date: Saturday, 2 July, 2011, 22:32
On
Sat, Jul 2, 2011 at 1:44 PM, < carolmooredc@verizon.net> wrote: > I dislike phrase Global South since > needs too much explanation. But "the 2/3 (or whatever percent) of the
> human population which lives in the economically developing world" is a > bit of a mouthful. It also helps to remind people that wikimedias exist > in dozens of languages, but how to add that to one short phrase, I know not!
You're definitely not alone in that dislike. :-) Here's a definition though, for anyone who's not sure what it means: < http://strategy.wikimedia.org/wiki/Global_South>
On Sat, Jul 2, 2011 at 5:24 PM, Javier Bassi < javierbassi@gmail.com> wrote: > In march 2010 WP had 12% and now its on 9%? Am I right or I'm missing
> something? :\ > The percentage isn't necessarily going down. The two percentages were found through surveys with pretty different methodologies. I think the most recent survey was intended to be "more scientific" in how it was
executed, hopefully giving a more accurate snapshot and more specific numbers. So, pretty much, it's most likely been between 9% and 12% all along, we're just getting an either more accurate number or just a
different amount of woman participated -- it doesn't mean that we're doing worse and are losing women we already had. I'm not a statistician or anything, though, so this could all be misguided. ;-) I'm sure someone else would be able to add more here.
-- Casey
Brown Cbrown1023
|