Member-only story
Lies, damn lies, and the Sullivan Review
All data is ideological. Simply asking to record one set of information over another is spurred by an ideological framing, as is the options presented, and the analysis carried out. There is no such thing as objective data science because every part of the data collection and curation process relies on decisions made by subjective humans. This is data science 101, and all data scientists worth their name will do everything in their power to minimise the amount of subjectivity in their work. They will fail to make it purely objective, but that is not the point. The moment you want to ask a question is the moment you are accepting that the question itself is subjective.
Thus, when you have a report such as the Sullivan Review in the UK which purports to have found massive failings in the way “biological sex” is recorded because robust and accurate data has been lost due to changes in methodologies alarm bells ought to be rung. Not for the fact that methodologies have been changed, but rather that “biological sex” is being used to frame sex in data. As any data scientist working with personal characteristics will tell you a request for information can lay out questions and answers but it cannot force someone to tick the box you wish. This is the most basic reason why EDI surveys are broad brush strokes at best and should never be used to frame absolutes. Even if you legally…
