* Posts by pbowl

2 publicly visible posts • joined 11 Feb 2016

The mod firing squad: Stack Exchange embroiled in 'he said, she said, they said' row

pbowl

Re: Is this just an English thing ?

* whom am I kidding

Women devs – want your pull requests accepted? Just don't tell anyone you're a girl

pbowl

Only half the story

The conclusions reported here are for "outsiders", which is only half the story. For "insiders" (owners or collaborators) it goes the other way: men and women have similar acceptance rates when gender is unidentified, but women have significantly more acceptances when identified. (From fig. 5.)

I would encourage both genders not to worry too much though. The effects are quite small - they only become statistically significant because of the huge size of the dataset. It's possible that these differences could be dominated by systemic errors, e.g., (a) the uncertain status of open requests which could be gender-linked in some way (e.g., one gender tending to submit bigger patches, which take longer to evaluate), and (b) possible confounding effect if decision to self-identify gender is correlated with ability/experience (e.g. if "serious" women tend more often to appear ambiguously), (c) possible confounding effect if ability to infer gender (which is only managed 35% of the time) is correlated with gender or ability/experience etc.

(There are some mysterious aspects to the paper. What is "Michael's tool"?)