In my travels around the country, I get a lot of questions about unusual housing situations and how the 2010 Census will enumerate people who live there. A common one concerns prisons.

The decennial census has the goal of counting everyone living in the country in the “right” place. That is, it is insufficient for us to have a perfectly accurate count of the total population if we cannot place each person enumerated in some location. (This burden flows from the need to reapportion the House across states and then to redistrict the states based on counts down to the block level.)

Since the Census Act of March 1790, we have placed each enumerated person at their “usual residence.” With some persons, where they are living now is not where they might think of themselves usually living. For example, I’ve blogged earlier about how many of the victims of Hurricane Katrina think of themselves as living on the Gulf Coast even though, since the event, they’ve lived far inland. I talked about how some college students in dormitories think of themselves as really living in their parental home, even though they only spend summers there. So too, many prisoners may think of themselves as living somewhere other than prison.

For some prisoners, however, they are only temporarily incarcerated. For persons in short-term jails, awaiting a hearing, we direct that the person should be included at their residence that they usually occupied before the jail incident. For those in hospitals for a short-term stay we direct that they be counted among their usual household.

Thus, conceptually there needs to be some time-based cutoff. The Census Bureau rules note that if in the last year the majority of the months were spent at a residence, then the person should be included in that residence’s count. Could such a rule be used for prisoners?

One can easily see that the problem is a logistical one. There are logistical issues and conceptual issues:

Enumerating in prisons. We seek to have each prisoner fill out an individual census form, but we don’t succeed in all institutions. In some prisons with very large security issues, prison management directs us to use the administrative records of the institution to enumerate.

Defining “usual residence” outside the prison. There are several optional definitions that could be used:

• Where the prisoner lived immediately prior to the arrest.
• Where the prisoner lived at the time of the arrest.
• Where the prisoner lived at the time of the sentencing.
• Where the prisoner’s former household now lives.
• Where the prisoner wants to live after exiting the institution.

Administrative records for state prisons vary widely in content and how they’re updated. When we have used records from institutions we have missing data rates that approach 50 percent. So changing the residence rule for prisoners would probably demand a greatly increased level of use of individual census forms in prisons; this requires more cooperation from prison officials.

Choosing which definition of a non-prison residence is best is not obvious; logical arguments can be made for each of the five above.

Some have argued that different rules could be used for different prisoners. For example, those who serving long term sentences, have been at the prison for some time and will be there for years more, might be counted as residents of the prison. Others, the argument goes, could be counted as residents of one of the five ideas above. Such a design would also require linking to some administrative rules to apply the sentence length rule.

Some users of census data care about this for redistricting purposes within states. They observe that prisoners often resided in areas far removed from the location of the prison and should be counted where they’re from. They note that the former homes of the prisoners are “cheated” of the benefits derived from the census counts. They argue that the locales of the prisons unfairly benefit from the counted prisoners, even though the prisoners do not enjoy the benefits that the census counts provide to the area.

This decade we are releasing early counts of prisoners (and counts of other group quarters), so that states can leave the prisoners counted where the prisons are, delete them from the redistricting formulas, or assign them to some other locale.

As a nonpartisan scientific organization, the Census Bureau is not involved in redistricting. We collect the information under uniform rules that offer the promise of accurate counts. We provide this early release to allow users more information in doing their jobs.

Counting members of all group quarters is complicated; we re-evaluate our “residence rules” after each census, to keep pace with changes in the society. We’ll do that again after the 2010 Census.

Please submit any questions pertaining to this post to ask.census.gov.

Director Robert Groves