You may remember the long-running story of my letters to the Office of National Statistics, and the more-concentrated effort by another blogger, in regard to the automatic “correction” of supposedly-”erroneous” data in the 2011 census, like somebody having multiple partners or identifying as neither gender. You don’t? Well here’s a reminder: part one, part two, part three, part four.
Well: we’ve finally had some success. A response has been received from the ONS, including – at last – segments of business logic from their “correction” code.
It’s hard to tell for certain what the result of the correction will be, but one thing’s for sure – Ruth, JTA and I’s census data won’t have passed their validation! Their relationship validations BP2, BP2a, and BP2b state that it is logically-impossible for a person to have a spouse and a partner living with them in the same household.
I should invite them around for dinner sometime, and they can see for themselves that this isn’t true.
I also note that they consider it invalid for anybody to tick both or neither of the (two) gender option boxes, although again, it’s not clear from the data they’ve provided how the automatic correction occurs. Increasingly, I’m coming to suspect that this might actually be a manual process, in which case I’m wondering what guidelines there are for their operators?
One good piece of news from this FoI request, though: the ONS has confirmed that the original census data – the filled-in paper forms, which unlike the online version doesn’t enforce its validation upon you – is not adjusted. So in a hundred years time, people will be able to look back at the actual forms filled in by poly, trans, and other non-standard households around the UK, and generate actual statistics on the frequency with which these occur. It’s not much, but it’s something.