Banning noise will be a disaster for statistical data products

38 points by teymour


wrs

The answer to "why" is "all of the above". As we've seen from multiple previous fiascos (e.g., labor statistics, climate data, all the way back to 2020's "don't let that ship dock because it'll make my COVID numbers worse"), this administration is not interested in having any objective data published that might contradict their storytelling. This one also directly relates to two of their biggest interests, congressional districting and "DEI". And I'm sure they see no utility whatsoever in "deidentifying" the populace, based on the extremely dubious things they're doing to identify the part of the populace they don't like.

0x2ba22e11

It sounds like the worst case scenario will be one dangerous census followed by subsequent censuses being useless because participants will stop answering honestly. A really dangerous one could result in participants learning the hard way that submitting accurate information will lead to it being used to harm them.

classichasclass

For our statistical data products on public health (jurisdiction of under half a million), we almost exclusively rely on suppression for small cell sizes, though we do make a point of reporting explicit zeroes.