On Wed, Jun 16, 2010 at 8:26 PM, phoebe ayers <phoebe.wiki(a)gmail.com> wrote:
There's been discussion of the gender gap among
Wikimedia editors on
and off for many years now, and it's a focus of the strategic planning
process. This is a part of a larger issue of how to get members of
underrepresented groups to edit more, to combat system bias on all
fronts. (Or, simply how to get more people to edit regardless).
You may find it interesting that these kind of large imbalances can
arise out of a simple but surprising mathematical truth:
If you have a mixed population with a skill, say skateboarding, that
follows the typical normal distribution and one sub-population (e.g.
people with red hair) have an average performance only slight higher
than another sub-population (blondes), and you were to select the
best skateboarders out of the group you would end up with a
surprisingly high concentration of the red-hair subgroup, so high that
it doesn't at all seem justified by the small difference in average
This is is because in normal distributions the concentration of people
with a particular skill falls off exponentially away from the average,
so if you take the two distributions (amount of skateboarding skill
for red-hairs and blondes) and shift one a very small amount the ratio
between the two becomes increasingly large as you select for higher
and higher skill levels.
The same kind of results happen when, instead of a difference in
average performance, there is simply a difference in the variation. If
red-hairs have the same average skate-boarding skill but are less
consistent— more klutzes _and_ more superstars this has an even larger
impact than differences in the average, again biasing towards the
These effects can be combined, and if there are multiple supporting
skills for a task they combine multiplicatively.[*]
The applicability here is clear: There is a strong biological argument
justifying greater variance in genetically linked traits in men (due
to the decrease in genetic redundancy) which is supported by many
studies which show greater variance in males. So all things equal any
time you select for extremes (high or low performing) you will tend to
tend to end up with a male biased group. (There are small also
differences in measured averages between men and women in many
And many of the 'skills' that are reasonable predictions of someone's
likelihood of being a Wikipedian, if we're even to call them 'skills'
as many aren't all that flattering, are obviously male super-abundant
in the greater world. How many female obsessive stamp collectors do
you know? Male? The kind of obsessive collecting trait is almost so
exclusively male that it's a cliché, and it's pretty obvious why that
kind of person would find a calling in Wikipedia.
One piece of insight that comes out of is that general approaches
which make Wikipedia more palatable to "average people", as opposed to
uber-obsessive techobibilo walking-fact-machines, may have a greater
impact at reducing gender imbalance than female centric improvements.
(and may also reduce other non-gender related imbalances, such as our
age imbalance). So this gives you an extra reason why "more people to
edit regardless" is an especially useful approach.
Though are limits to the amount of main-streaming you can do of an
academic activity such as encyclopaedia writing. :-)
In any case, I don't mean to suggest that your work isn't important or
can't be worthwhile. Only that I think you're fighting an uphill
battle against a number of _natural_ (not human originated) biases,
and I wish you luck!
[*] A while back I wrote up a longer and highly technical version of
this explanation as part of an argument on gender imbalances in
computer science with a mathematician. Anyone into math-wankery may
find it interesting: