Using GitLab CI to check for inclusive language in job family pages
Highly gendered language may be found and merged into our job families and job vacancy posts. Gendered language may deter applicants from groups that are current underrepresented at GitLab so we'd like to find a solution that identifies gendered wording and notifies authors of times they add gendered wording when authoring a MR in a job family or job vacancy post.
Tl;DR - research shows that masculine-coded language in adverts is less appealing to women. For men, feminine-coded adverts were only slightly less appealing and there was no effect on how much the men felt they belonged in those roles. For our team to be representative of society, our inclusion data shows we need to hire more non-male colleagues so we should ensure our content appeals to non-male candidates.
From Kat Madfield's summary of Danielle Gaucher, Justin Friesen, and Aaron C. Kay: Evidence That Gendered Wording in Job Advertisements Exists and Sustains Gender Inequality (Journal of Personality and Social Psychology, July 2011, Vol 101(1), p109-28): "In this paper the researchers showed job adverts which included different kinds of gender-coded language to men and women and recorded how appealing the jobs seemed and how much the participants felt that they 'belonged' in that occupation. No non-binary people were included in this research, and the research didn't touch on non-binary-coded words.
Their results showed that women felt that job adverts with masculine-coded language were less appealing and that they belonged less in those occupations. For men, feminine-coded adverts were only slightly less appealing and there was no effect on how much the men felt they belonged in those roles."
First step: identify whether there is a Ruby-built version of the gender decoder tool
Second step: implement for new job families or edits to job families
- someone makes an MR to add a new or edit an existing job family
- the CI job checks only that specific job family
- in the CI job it will be outputted what should be improved or changed
- the MR author is responsible to fix this
Third step: check the existing ones
For all the existing ones what we could do is run a manual script and dump the output into a GitLab epic and for every job family that needs to be fixed, we create an issue with the link to the job family and what needs to be fixed.
I know we love iteration, so I would suggest, we get started with the first step. But open to your (and others) ideas. At this point it would probably be best we discuss a solution in an issue on the People Ops Engineering project. But feel free to copy paste my comment here in that issue when you create it."