How Should the Census Count Prisoners?

In today’s New York Times, NYU law professor Anthony Thompson takes on the interesting issue of how those in prison should be counted in next year’s United States Census. For most of the article he’s on the right track, but he finishes by suggesting a “third way” when, in fact, it’s pretty clear that one side of the debate is right.
The fundamental question is this: Should prisoners be counted where they are incarcerated, or at their last non-prison address? Currently, prisoners get counted in their place of incarceration. Mayors in urban areas generally want prisoners counted as city residents, while those in rural areas where the prisons are want to keep the current system.

The urban mayors are right. Prisoners aren’t, in any real sense, residents of the places where they are incarcerated. With a few tiny exceptions, the principle of “least eligibility” means that prisoners can’t receive the government services that free citizens do. Only a few states let them vote while in prison, and felons — like children and the mentally disabled — often can’t vote upon release either. Their victims and families very likely live in the places where they came from. In short, the costs that prisoners impose on society fall on the municipalities where they lived and will probably return after release.

Doing what Thompson suggests — counting the last known address for prisoners due to be released within ten years, and the prison address for others — seems fraught with problems. While many states have abolished traditional parole, all states without parole maintain “time off” systems that let well-behaved prisoners qualify for shorter sentences. Since no two states have exactly the same system, figuring out the meaning of “within ten years” would present a bureaucratic nightmare for the census and, very likely, would end up encouraging some states legislatures to change sentencing rules and definitions to manipulate the census.

Furthermore, although Thompson says differently, adopting a rule like this would set a bad precedent for debates over counting college students, or retirees who evenly split their time between two locations. While the current system for counting these groups aren’t perfect, in that members of both groups get sometimes get double-counted, trying to establish a similar “halfway” rule for either group would present another nightmare.

In short, a simple solution will work best.