“Urban America likely to gain clout after Census”

Reid Wilson, in The Hill, discusses the population shift to cities

For those following census issues closely, one of the impacts of the implementation of differential privacy may blunt that apparent move a bit on the margins.  As a really brief (and inadequate) primer: the Census Bureau plans to change the way in which it safeguards the confidentiality of responses by adding privacy protections that are far more resistant to reverse engineering than past practices.  It involves adding noise to the actual counts (the way the algorithm works, small geographies end up noisier than large ones: state level total counts reflect the actual count exactly, but census-block information will be a bit different). 

At the moment, the Bureau is leaning against releasing numbers that look fake (so, no negative counts).  What that means is that big populations get both bigger and smaller — but tiny populations get larger than they get smaller (because a block with 2 people may be reported with 5 people, but won’t be reported with -1).  And that potentially means a fairly small, but systemic, bias toward reported overall growth in rural populations that isn’t there.

Share this: