ACS Data Users Group

The Commerce Department issued an order on June 4 prohibiting the use of "noise infusion" as a disclosure avoidance technique for Census Bureau statistical products. The potential implications for the ACS are still unclear, but this is worth watching closely.

The order: https://www.commerce.gov/opog/disclosure-avoidance-statistical-products

Depending on how the Census Bureau interprets the order, several ACS disclosure avoidance methods-including swapping, age perturbation, synthetic data, and limited uses of differential privacy-could potentially be affected, which could leave coarsening (aggregation, rounding, ranges) and suppression as the primary disclosure avoidance tools available to the Bureau.

We're hoping to learn more soon about how the Census Bureau plans to respond. Any insights or information would be greatly appreciated.

The short answer is that we don't know yet.

My reading is that small-area tabulations below the county level could face the biggest challenges if disclosure avoidance methods currently used by the CB are no longer allowed. Geographic aggregation is specifically identified in the order as an acceptable disclosure limitation approach. But relying more heavily on aggregation could come at the cost of some of the small-area detail that makes ACS data so valuable.

At this point, we're still waiting for guidance from the Census Bureau on how it interprets the order and how it plans to implement any resulting changes.

I work with the microdata about as, if not more, often than I work with tabulations. The PUMS microdata have had the "tabooed" methods applied to them, as far as I understand. Does that mean that these data will be pulled? (That's a rhetorical question, I know that the answer is, "we don't know".)