“Home” in 2010: A Report on the Feasibility of Enumerating People in
Prison at their Home Addresses in the Next Census
Brennan Center for Justice at NYU School of Law
Kirsten D. Levingston & Christopher Muller
Submitted to the U.S. Census Bureau
February 15, 2006

“Home” in 2010
Page 2 of 22

About the Brennan Center
The Brennan Center for Justice at NYU School of Law unites thinkers and advocates in
pursuit of a vision of inclusive and effective democracy. Our mission is to develop and
implement an innovative, nonpartisan agenda of scholarship, public education, and legal
action that promotes equality and human dignity while safeguarding fundamental
freedoms. The Center operates in the areas of Democracy, Poverty, Liberty and National
Security, and Criminal Justice.

About the Authors
Kirsten D. Levingston directs the Criminal Justice Program at the Brennan Center for
Justice, whose mission is to reduce the number of people in prison and the nation’s
unnecessary over-reliance on incarceration.
Christopher Muller is a research associate in the Criminal Justice Program.

Acknowledgements
This paper was made possible by the generous support of the JEHT Foundation and an
anonymous donor, for which the Center is grateful. The authors wish to thank Lynn Lu
for her research and input. The authors also wish to thank Patricia Allard, William
Cooper and David McMillen for their review and comments. We are indebted also to the
insights of Kenneth Prewitt. Special thanks are due to the criminal justice officials and
data experts interviewed for the report. Their comments comprise much of the content of
this report. Of course, all opinions and any errors are the authors’ alone.

Brennan Center for Justice at NYU School of Law
161 Avenue of the Americas, 12th Floor
New York, NY 10013
(212) 998-6730
www.brennancenter.org

“Home” in 2010
Page 3 of 22

Executive Summary
In November 2005, Congress directed the Census Bureau to take a first step in the
direction of changing the way it counts people in prison. In a conference report
accompanying H.R. 2862, last year’s Science, State, Justice, Commerce and Related
Agencies Appropriations Act, it directed the Bureau to study the feasibility of using
“home of record” information to enumerate people in prison. Congress gave the Bureau
90 days, until February 2006, to report its study findings to Congress.
The Congressional directive raises three questions for consideration. First, it requires the
Bureau to define “home” for the purpose of enumerating people in prison. Second, it
requires the agency to consider how best to count the majority of people in prison, those
with accurate and verifiable home addresses. Third, it necessitates consideration of how
to enumerate exceptional cases, those in which a person in prison cannot be counted at
his or her home address.
People in prison are temporarily absent from their home communities. Yet at census time
the Bureau enumerates them as if prison were their “usual residence.” In the past the
Census Bureau has looked beyond physical presence in deciding how to define usual
residence, arguing in the U.S. Supreme Court that a person’s home can be based on an
“enduring tie” to a community. Because most people in prison have a home address they
can report on census day, the Bureau should treat them as it treats the majority of the
people it counts, by providing a form to complete or conducting an interview if
necessary. As this report discusses, the exceptional cases, those in which a home address
is unavailable, can be enumerated in other ways.
Congress’s November 2005 directive was a critically important step en route to making
the 2010 prison count fairer and more accurate. Given the quick 90-day turnaround,
spanning the holiday-filled months of November, December and January, and the lack of
funds appropriated to conduct a full study, the February 2006 report should not be the last
word from the Bureau. Congress should provide the Census Bureau with more time and
funds to conduct a thorough test of this issue, one that will be the basis for an improved
2010 prison count.

“Home” in 2010
Page 4 of 22

Introduction
On November 22, 2005, President Bush signed into law H.R. 2862, last year’s Science,
State, Justice, Commerce and Related Agencies Appropriations Act.1 The attendant
conference report included the following language:
The conferees direct the [Census] Bureau to undertake a study on using
prisoners' permanent homes of record, as opposed to their incarceration
sites, when determining their residences. The Bureau should report back
to the Committees on Appropriations on its findings within 90 days of
enactment of this Act.2
This directive is not the first of its kind: The Census Bureau has undertaken enumeration
studies in the past, most recently in 2004, when Congress directed it to test the feasibility
of counting Americans overseas. What distinguishes this study from the most recent is a
matter of funds and sufficient time to allow for necessary testing. In the 2004 study, the
Bureau spent $7.8 million preparing for, implementing, and evaluating its test, only to
conclude that counting Americans overseas “would not be cost-effective.”3 The test alone
took 5 months to complete.
While a critically essential first step, the current directive, by contrast, appropriates no
funding and affords the Bureau only 90 days in which to report back to Congress.
Consequently, Congress should not expect this study to yield such conclusive results.
What the Bureau can accomplish in 90 days, however, is to (1) survey the state of prison
and jail administrative records containing home address information and (2) oversee an
internal consideration of best practices and methods for implementing a rule change
absent sufficiently reliable jail and prison administrative records. That is, the agency can
determine what home address information currently exists and come up with options for
how to enumerate people whose home address information is unavailable.
In the months of December 2005 and January 2006, the Brennan Center for Justice at
NYU School of Law conducted interviews with criminal justice officials and data users
of wide-ranging expertise.4 This document is informed by their insights. In it, we address
the following three topics:
1

While this Act names an annual Congressional appropriation, it is the first to be titled thus. In early 2005
the purview of the corresponding House subcommittee on appropriations was renamed and expanded. See
Reorganization Prompts House Dems to Make Changes, NATIONAL JOURNAL’S CONGRESS DAILY,
February 15, 2005.
2
H.R. CONF. REP. NO. 109-272, at 140 (2005).
3
$7.8 million divided by a total of 5,390 responses equals approximately $1,450 per response. United
States Government Accountability Office, Counting Americans Overseas as Part of the Decennial Census
Would Not Be Cost-Effective, GAO-04-898, August 11, 2004.
4
We interviewed Jim Austin, JFA Institute; Allen Beck, Bureau of Justice Statistics; Jim Beck, U.S. Parole
Commission; Eric Cadora, Justice Mapping Center; William Cooper, FairData 2000; Ryan King, The
Sentencing Project; Jeremy Travis, John Jay College; Bruce Western, Princeton University; and Reggie
Wilkinson, Ohio Department of Rehabilitation and Correction.

“Home” in 2010
Page 5 of 22

1. How to define “home” for the purpose of enumerating people in prison,
2. How best to count the majority of people in prison, those with accurate and
verifiable home addresses, and
3. How best to deal with exceptional cases in which a person in prison cannot be
counted at his or her home address.
Each interview we conducted brought a fresh perspective to bear on the prison
enumeration, but over time a common theme emerged: While the Census Bureau faces
obstacles to changing its method of enumerating people in prison, this change is entirely
possible. As one interviewee put it, “it is certainly no easy task, but not an
insurmountable one”—a statement that could well describe the taking of the census itself.
Still, the Census Bureau does not act in isolation. In order to properly test and adopt a
new way of counting people in prison, it will need additional support from Congress. It is
with this in mind that we encourage the Bureau to report to Congress, at the conclusion of
its 90-day deadline, with a request for additional time and funds to prepare for a rule
change in the 2010 decennial census.5
This report does not purport to offer all the answers. Indeed, the Bureau—with a proper
supply of time and resources—is in the best, and perhaps only, position to do that. It
does, however, synthesize the long-term expertise and knowledge of many of the top
experts in the field to evaluate the fairness, accuracy, and feasibility concerns of a set of
alternative enumeration options, including the Bureau’s current approach.6

5

One of the Bureau’s selected dress rehearsal sites provides the perfect grounds for continued testing. As
the Bureau notes in its January 19, 2006 press release: “San Joaquin County, just south of Sacramento, was
selected because it presented an urban location with a multilingual population and an assortment of group
quarters housing such as hospitals, college residence halls, nursing homes, prisons and facilities for the
homeless.” See Press Release, United States Census Bureau, Census Bureau Selects Sites for Census Dress
Rehearsal, Local Communities to Gain Temporary Jobs, (January 19, 2006), available at
http://www.census.gov/Press-Release/www/releases/archives/census_2010/006345.html.
6
For further discussion, see Patricia Allard, Molly K. Biklen & Kirsten D. Levingston, One Size Does Not
Fit All: Why the Census Bureau Should Change the Way It Counts Prisoners (Brennan Center for Justice,
New York, N.Y.) (2004); Patricia Allard & Kirsten D. Levingston, Accuracy Counts: Incarcerated People
& the Census (Brennan Center for Justice, New York, N.Y.) (2004); Patricia Allard & Christopher Muller,
Incarcerated People and the Census: Painting a Distorted Picture of Virginia (Brennan Center for Justice,
New York, N.Y.) (2005).

“Home” in 2010
Page 6 of 22

I. Defining “Home” for People in Prison
A. The Census Bureau’s definition of “home”
Developing a reliable repository of the home of record of every incarcerated person in the
United States is as much a question of definition as it is of thorough record keeping and
data systems automation.
Today the Census Bureau relies on the concept of “usual residence” to determine the
homes of people in prison. In a letter to Rep. William Lacy Clay (D-MO) and Rep. Adam
Putnam (R-FL), Census Bureau Director Charles Louis Kincannon traced the history of
this concept:
[U]sual residence is the modern version of “usual place of abode,” as stated in
the Census Act of 1790. We found that the Census Act of 1790 has been
interpreted with reasonable consistency over time, albeit with some finetuning along the way. One can trace the interpretation of “usual place of
abode” and “usual residence” in census enumerator manuals and other
documentation. In 1910, usual place of abode was described as “commonly
the place where one regularly sleeps.” In 1920, 1930, and 1940, usual place
of abode was the “permanent home” or “regular lodging place.” Not until
1950 did “usual place of abode” become “usual residence,” that is, the place
where the person lives or sleeps most of the time. Thus, for the 1950 census,
usual residence returned to a more readily defined and precise definition,
“permanent home” was not mentioned.7
As Director Kincannon’s letter demonstrates, over time the Census Bureau’s
interpretation of the concept of “usual residence” has evolved. Congress’s directive gives
the Bureau yet another opportunity to revisit and fine-tune this concept.
In Franklin v. Massachusetts, a 1992 U.S. Supreme Court case, the state of
Massachusetts challenged the Bureau’s decision to treat federal personnel deployed
overseas as residents of their “home of record” during the 1990 census. This approach led
to over 900,000 overseas federal employees being counted at their “home of record” and
ultimately resulted in Massachusetts’ loss of a Congressional seat. In challenging the
Bureau’s approach, Massachusetts argued that using “home of record” information
maintained in employee personnel files to apportion Congressional seats was arbitrary
and capricious under the Administrative Procedure Act. The federal trial court agreed and
ordered the President to change the Congressional allocation. In appealing that decision
to the U.S. Supreme Court, the Bureau argued, “[i]t is far too late in the Nation’s history
to suggest that enumeration of the population of the States must be based on a rigid rule

7

Letter from Charles Louis Kincannon, Director of the United States Census Bureau, to Representative
William Lacy Clay and Representative Adam Putnam, (April 9, 2004) (on file with the Brennan Center for
Justice).

“Home” in 2010
Page 7 of 22
of physical presence on the census date—a rule that has never been applied and that is
especially out of place in an age of ever-increasing mobility.”8
The Bureau’s defense of a more nuanced and historically sensitive definition of residence
ultimately succeeded in the Supreme Court. The Franklin Court stated that using “home
of record” information was in keeping with the Census Bureau’s historic standard in that
it “reflected the more enduring tie of usual residence.”9 The Court went on to explain that
usual residence
can mean more than mere physical presence, and has been used broadly
enough to include some element of allegiance or enduring tie to a place.
The first enumeration Act itself provided that “every person occasionally
absent at the time of the enumeration [shall be counted] as belonging to
that place in which he usually resides in the United States.” The Act
placed no limit on the duration of the absence, which, considering the
modes of transportation available at the time, may have been quite
lengthy. For example, during the 36-week enumeration period of the 1790
census, President George Washington spent 16 weeks traveling through
the States, 15 weeks at the seat of Government, and only 10 weeks at his
home in Mount Vernon. He was, however, counted as a resident of
Virginia.10
Today the Bureau continues to treat temporarily absent populations—including military
and federal employees stationed abroad, members of Congress, and boarding school
students—as it treated the first President in the first census. These populations are
considered residents of the places to which they maintain an “enduring tie” even though
they are temporarily removed from that place on census day.
B. Applying the Bureau’s definition of “home” to people in prison
When the first census was taken in 1790, counting people in prison as residents of the
correctional facilities in which they were held made sense: imprisonment as we know it
today was still in its infancy.11 Prisons, where they existed, were located close to the
homes of those locked away in them.12
8

Brief for the Appellants at 17, Franklin v. Massachusetts, 505 U.S. 788 (1992) (No. 91-1502) (emphasis
added).
9
Franklin v. Massachusetts, 505 U.S. 788, 795 (1992).
10
Id. at 804 (citation omitted).
11
Besides which, intrastate population disparities due to incarceration did not matter in 1790. “Article 1,
Section 2, of the U.S. Constitution established [only] that the apportionment of the U.S. House of
Representatives shall be based upon a national census.” See U.S. Census Bureau, Strength in Numbers,
Your Guide to Census 2000 Redistricting Data From the U.S. Census Bureau, July 2, 2000, available at
http://www.census.gov/clo/www/strenghth2.pdf.
12
The year 1790 holds the double distinction of producing both the first census and the first use of
incarceration as punishment for people convicted of crimes. The nation’s first functioning “prison”
emerged when sixteen cells at the Walnut Street Jail in Philadelphia were converted into housing for people
convicted of felonies. Prisons in the United States had previously been used strictly to hold people awaiting

“Home” in 2010
Page 8 of 22

Since 1790, and most markedly in the thirty-year period following the mid-1970s,
incarceration has become an inescapable feature of our national landscape. Nationally,
the number of people in state and federal prison has grown more than fourfold—from
329,821 to 1,496,629—since 1980.13 Including local jails, that number jumps to
2,267,787 people.14

Sentenced Prisoners in State and Federal Institutions
1,600,000
1,400,000
1,200,000
1,000,000
800,000
600,000
400,000
200,000

19
25
19
29
19
33
19
37
19
41
19
45
19
49
19
53
19
57
19
61
19
65
19
69
19
73
19
77
19
81
19
85
19
89
19
93
19
97
20
01

0

Source: Bureau of Justice Statistics, Sourcebook of Criminal Justice Statistics, Table 6.28.2004

Each year, nearly 650,000 people are released from prison.15 With an average time served
of 34 months in state prison, most people counted in prison in the 2000 census will have
been released before the next census in 2010.16 Of those, many will return to the
communities to which they have enduring ties.
In explicitly referencing “prisoner’s permanent homes of record” in its directive,
Congress has offered one possible interpretation of place of enduring ties. Indeed, “home
trial. The first walled penitentiary in the U.S. was not built until 1829. MARC MAUER, RACE TO
INCARCERATE 3 (New Press, 1999).
13
George Hill & Paige Harrison, Prisoners under State or Federal Jurisdiction, Federal and State-byState, 1977-98 (Bureau of Justice Statistics, 2001) available at http://www.ojp.usdoj.gov/bjs/prisons.htm;
Paige M. Harrison & Allen J. Beck, Prisoners in 2004, 3 (Bureau of Justice Statistics, October 2005).
14
Harrison & Beck, Prisoners in 2004, supra note 12, at 1.
15
Department of Justice, Office of Justice Programs, http://www.ojp.usdoj.gov/reentry/learn.html.
According to one estimate, this figure balloons to 7 million if we include releases from jails. See Theodore
M. Hammett, Cheryl Roberts & Sofia Kennedy, Health-Related Issues in Prisoner Reentry, 47:3 CRIME &
DELINQUENCY 390, 391 (2001). The Council on State Governments reports that “[v]irtually every person
incarcerated in a jail in this country—and approximately 97 percent of those incarcerated in prisons—will
eventually be released.” Report of the Re-Entry Policy Council, xviii (Council on State Governments).
16
“Mean sentence length and mean time served for first releases from State prison,” BUREAU OF JUSTICE
STATISTICS, SOURCEBOOK OF CRIMINAL JUSTICE STATISTICS 2003 511, table 6.44 (2003).

“Home” in 2010
Page 9 of 22
of record,” as reported in state and federal administrative records, is widely considered
the most robust extant predictor of the place people in prison will return upon release. For
example, Eric Cadora, director of the Justice Mapping Center,17 told us:
We’ve found that collectively there isn’t much difference when we map a
parole population and home address of a prison population. We’re pretty
confident about the collective accuracy of self-reported prison addresses.
In other words, when we map a parole population and map the home
addresses of prisoners, we’ve found that statistically speaking, the
concentrations of population are on the whole in the same
neighborhoods.18
Jim Beck, former director of the United States Parole Commission,19 explained the link
between sentencing and release locales in the following way:
If the person is released under supervision, they must be released to the
district where they were sentenced. They must have a strong rationale for
requesting release to a different area, and must explicitly request this.
There are exceptions, for example if an inmate arrested in DC wants to be
released to his mom’s house in North Carolina, but for the most part, they
must be released to the district where they were sentenced – assuming
they’re under supervision. They’d have to have a pretty good reason to
justify moving from the district where they are sentenced.20
State parole operates similarly.21
17

For more information on the Justice Mapping Center see http://www.justicemapping.org.
Telephone interview with Eric Cadora, Director, Justice Mapping Center, (December 19, 2005).
19
http://www.usdoj.gov/uspc/.
20
Telephone interview with Jim Beck, Research Administrator, United States Parole Commission
(December 16, 2005). The statute governing “[r]elease of a prisoner,” 18 U.S.C. § 3624, sets forth
requirements regarding a prisoner’s date of release, calculation of good-time credits, pre-release custody,
and “[a]llotment of clothing, funds, and transportation” upon “expiration of the prisoner’s term of
imprisonment,” including “transportation to the place of the prisoner’s conviction, to the prisoner’s bona
fide residence within the United States, or to such other place within the United States as may be authorized
by the Director.” 18 U.S.C. § 3624(d)(3) (1996). See also 28 C.F.R. § 571.22 (2005) (providing for
transportation of formerly incarcerated person to his/her “place of conviction or to his/her legal residence
within the United States or its territories.”). Federal regulations state that prisoners granted parole
“[g]enerally . . . will be released only to the place of their legal residence unless the [Parole] Commission is
satisfied that another place of residence will serve the public interest more effectively or will improve the
probability of the applicant’s readjustment.” 28 C.F.R. § 2.33(b) (2003). For those who are not paroleeligible, some prisoners will be “released, as if on parole,” under supervision. 28 C.F.R. § 2.35 (2003).
Where the released prisoner remains subject to conditions of supervised release imposed at the time of
sentencing, the prisoner supervising district is generally the one in which the sentence was imposed.
21
Release practices at the 2008 dress rehearsal sites are instructive. In California, pursuant to California
Penal Code § 3003(a), “an inmate who is released on parole shall be returned to the county that was the last
legal residence of the inmate prior to his or her incarceration.” An inmate may be released to a different
county “if that would be in the best interests of the public,” upon consideration of several factors, “giving
the greatest weight to the protection of the victim and the safety of the community.” CAL. PENAL CODE §
3003(b) (1999). North Carolina does not appear to require a particular place of release for inmates.
18

“Home” in 2010
Page 10 of 22

Taken together, the above comments support the claim that for the majority of addresses,
“home of record” is a reliable measure of return address. On a collective level, most
people in prison return to their home of record and report for post-release supervision in
their county of commitment.22
One such person wrote the Brennan Center this month after reading about the census
issue in a Charleston newspaper. He grew up in Minnesota, lived there for 37 years, and
is now incarcerated at the United States Penitentiary – Hazelton, in Bruceton Mills, West
Virginia. His letter expressed frustration that he was not treated as a Minnesota resident
even though he will be under post-release supervision there when his sentence ends.23
As we conducted our interviews, we began to detect a secondary trend in the responses: if
individual exceptions constituted the major objections to the rule change, aggregate
effects formed the main supports. If Congress is concerned about the distribution of large
swaths of population, the options laid out in the following two sections take strides
toward correcting distortions caused by the current prison enumeration and producing a
more accurate picture of our nation’s communities.

However, the North Carolina Department of Correction’s Office of Research and Planning has published
Standard Operating Procedures (“SOP”) (revised August 2004) to implement the “Going Home Initiative.”
Under a memorandum of agreement/understanding, several government agencies are to coordinate their
efforts to support offender reentry. Eligible inmates meet with a case manager to “assess the status of the
inmate’s home plan, employment, vocational, and treatment needs.” SOP at 7. An agent of the Division of
Community Corrections then “will conduct a home visit to assess suitability of the proposed home plan, if
one exists.” Id. A call to the North Carolina Department of Correction Office of Victim Services indicated
that no default “home plan” exists; most often, an inmate’s home plan will involve living near supportive
family members, which is frequently also the place where the crime occurred.
22
Nearly 80 percent of all people in state prison will be released to post-release supervision. See Bureau of
Justice Statistics, Reentry Trends in the U.S., available at
http://www.ojp.usdoj.gov/bjs/reentry/releases.htm.
23
Letter to Brennan Center for Justice (February 1, 2006) (on file with the Brennan Center for Justice).

“Home” in 2010
Page 11 of 22

II. Enumerating the Majority: Counting 80% of People in Prison
According to Bureau of Justice Statistics statistician Allen Beck, “you have perhaps 1020% of inmates where home of record information isn’t meaningful, but for 80% it is
24
quite meaningful.” In this section we consider the two major operational options for
enumerating the majority of people in prison: using extant administrative records and
interviewing and distributing forms in prison.
Option 1: Using extant administrative records
The strictest interpretation of Congress’s directive focuses on the “home of record”
addresses maintained by the Federal Bureau of Prisons, state departments of corrections
(“DOCs”), and local jails. The clearest benefit to using these records is that, in most
cases, they already exist; the Bureau would not have to canvass most correctional
facilities as it canvasses rural areas because correctional officials have already effectively
so done. The clearest pitfall to using these records is that they vary in degrees of
accuracy, completion, and automation. According to a 1998 Department of Justice study,
29 state departments of corrections maintain electronic home address information for
“more than 75% of offenders.” Four state departments of corrections maintain the records
in paper format and 9 do not collect the information at all.25
The uncertain status of state DOC records is a serious impediment to an enumeration that
meets the Bureau’s usual standards of precision. Utilizing these data as they currently
exist would require significant additional legwork on the Bureau’s part.
This assumes, however, that the Bureau alone would be responsible for ensuring accurate
and complete home of record information. A number of our interviewees recommended
that the burden instead be shifted onto state departments of corrections, which are wellpositioned to collect and report accurate data needed by the Bureau.26 According to Ryan
King of The Sentencing Project,27 all departments collect some information at intake;
24

Telephone Interview with Allen Beck, Director, JFA Institute, (December 19, 2005). Similarly, Reggie
Wilkinson, the Director of the Ohio Department of Rehabilitation and Correction, told us that in “80% of
the cases where the person is from is going to be the county where they were convicted.” Telephone
Interview with Reggie Wilkinson, Director, Ohio Department of Rehabilitation and Correction, (December
16, 2005).
25
See Bureau of Justice Statistics, State and Federal Corrections Information Systems, August 18, 1998.
26
Each state and the District of Columbia participate, with the Census Bureau, in the Federal State
Cooperative Program for Population Projections, whose by-laws identify the following objectives: “To
promote the improvement and advancement in the techniques, methodologies, and testing of population
projections and to exchange such information among the federal and state governments; [t]o encourage and
promote the collection of reliable and valid statistical series from which population projections can be
developed; and [t]o provide a forum for interaction between federal and state governments with regard to
policies on the development and uses of population projections.” U.S. Census Bureau, About the Federal
State Cooperative Program for Population Projections, available at http://www.census.gov/population/
www/fscpp/coophist.html. Under this program, both state agencies and the Census Bureau stand to benefit
from the sharing of corrections data for use in accurately assessing and addressing federal, state, and local
needs.
27
For more information, see http://www.sentencingproject.org.

“Home” in 2010
Page 12 of 22
updating their records would require only (1) surveying current incarcerated populations
and automating the responses and (2) developing a procedure for collecting all
subsequent home address information at intake. Knowing the information would be
electronically stored would give corrections employees further reason to carefully enter
intake responses. This would help avoid errors of collection, a major hindrance to the
creation of useful and complete data sets, in King’s experience. While this overhaul of
data collection is no doubt a serious undertaking, one of our interviewees reported that
he’d witnessed an entire department with only paper records update and automate its
system at little cost in a matter of two weeks.
In addition, the Association of State Correctional Administrators (ASCA) is presently
seeking to create standards of measurement and data collection for all state departments
of corrections across the country.28 ASCA stands to benefit from a nation-wide data
systems overhaul and could potentially work in collaboration with departments to include
home address record-keeping in its renewed performance indicator system.
Option 2: Interviewing and distributing forms in prison
The most popular procedural option among our interviewees was that which had the
Census Bureau send enumerators to correctional facilities to distribute short forms and
conduct interviews where necessary. This option eliminates the problem of discrepancies
across discordant state data sets and establishes a uniform enumeration.29 And it simply
expands the reach of a method familiar to the Bureau.
The interview method additionally would allow the Bureau to set its own standards for
enumeration and definition of home, rather than let the existing data determine this by
their limitations. Many people in prison, for example, could report a more recent or
complete address than that which was collected at intake.
While the interview method may appear to be more labor-intensive than the other
methods, collecting fresh data should prove less nettlesome than adopting the
administrative records that are complete and verifiable and subsequently filling in the
gaps. As our interviews confirmed, problems of completion riddle many departments’
data sets. Matching missing data to individuals and re-canvassing would likely require as
much if not more Census Bureau resources than simply distributing forms and conducting
interviews at all facilities.

28

See Kevin Wright, Judy Bisbee & Patricia L. Hardyman, Defining and Measuring Corrections
Performance, Association of State Correctional Administrators, February 2003, available at
http://www.asca.net/public/perfmeasfinal.pdf.
29
“Census enumeration methods for people in prison vary from facility to facility. In some prisons officials
complete census forms on behalf of incarcerated people, while in others officials distribute forms to
individuals who complete and return them. The Census Bureau should study the accuracy of the approaches
currently utilized and select one to apply uniformly at all prison facilities, ensuring quality control of the
counting process.” See Patricia Allard & Kirsten D. Levingston, Accuracy Counts: Incarcerated People &
the Census (Brennan Center for Justice, New York, N.Y.) (2004), at 16.

“Home” in 2010
Page 13 of 22
The 2,267,787 people behind bars constitute approximately 0.7% of the population of the
United States. This figure can be interpreted in two consistent ways. First, at nearly a
percent of the nation’s population, the number of people incarcerated is clearly
significant enough to merit reexamination and ultimately to properly count. Second, the
number is not large enough to excessively burden the Census Bureau, particularly where
the population at hand is literally a captive one. People in prison, unlike other hard-tocount populations, are quite easy to locate.

“Home” in 2010
Page 14 of 22

III. Enumerating Exceptional Populations: Counting 20% of
People in Prison
As our interviews indicated, most people in prison have addresses that are meaningful,
accurate, available, and verifiable.30 There are, however, exceptions to this general rule.
And these exceptions constitute the most frequently invoked reasons for approaching a
rule change with caution. The following examples present particularly thorny problems:
Box 1: “Exceptional” populations
•
•
•
•

Someone whose family has moved to another address following his or her
incarceration
Someone who will never leave prison
Someone who had no address at the time of incarceration
A non-citizen whose official home is in another country

As more than one interviewee proffered, the Bureau would surely err in incorrectly
counting 80% of people in prison because of difficulties that emerge in counting the
remaining 20%. With the aim of not ruling on the exception, this section focuses on ways
to account for the 20% of people for whom home address enumeration is an
impossibility.
Anticipating likely operational issues, in this section we evaluate four methods for
handling “exceptional” populations in the prison enumeration. Each method is evaluated
against a set of four criteria:
Box 2: Evaluation criteria for enumeration options
availability of information – using this method, how easy or necessary is it for the
Census Bureau to collect home address information for exceptional populations?
verifiability – using this method, is there a need to verify the addresses the Census
Bureau assigns to or collects from exceptional populations against its master address
list?
accuracy – to what extent does the address assigned to or collected from
exceptional populations artificially inflate populations in communities with prisons, and
artificially dilute populations in communities without prisons?
fairness – to what extent does this method satisfy the fairness concerns of both
Congress and the Census Bureau?

30

The verification problem in the prison count is one of extent, not kind. The task of verifying street
addresses poses nothing novel to the Bureau—all addresses are verified by checking a master database to
determine whether they exist or not. The worst threat any of the options discussed here poses to verification
is an expansion of its breadth; the Bureau would simply need to check more addresses on the master list.

“Home” in 2010
Page 15 of 22
Option 1: Enumerating at the facility
For each of the exceptional populations listed in Box 1, the most immediate solution to
suggest itself is the current one: counting them as residents of their facilities. Indeed, in
some limited number of cases, enumeration at the location of the prison appears the only
viable option. This seems most apparent in the case of people who will never leave
prison.
Counting people at the facility address is clean and easy for a number of reasons. For one,
prison addresses are readily available. It is much less labor-intensive for the Bureau to
simply locate and verify a discrete and limited number of correctional facility addresses
than to do the same for a host of individual addresses belonging to those held in such
facilities.31 The Bureau is already familiar with this, its current practice for the entire
incarcerated population, and would need to make no significant operational changes in
order to proceed with the 2010 enumeration. For these reasons, this option is the most
user-friendly.
The practice of counting people in prison at their facilities withstands evaluation by other
criteria less well. Institutional addresses produce the least accurate results. For example,
in Census 2000, 43,740 New York City residents were counted in upstate prisons miles
from their urban dwellings.32 An Urban Institute Report found that 53 percent of people
released from prison in Illinois returned to Chicago—a city with no state prisons—upon
completion of their sentence.33 For years, data users and demographers have been
flummoxed by distortions in local population counts and demographics due to the current
prison enumeration method—many sufficiently so that their states have opted to exclude
prison counts entirely from redistricting calculations. In Virginia, for example, counties,
cities, and towns with census populations that are comprised of more than 12% people in

31

Institutional addresses, however, would still need to be verified. As noted by the Panel on Research on
Future Census Methods of the Committee on National Statistics, “[t]he Bureau’s Count Question
Resolution program, beginning in 2001 and continuing through 2003, offered local authorities the
opportunity to challenge their census population counts for geocoding and other errors; if successful, they
could receive a letter certifying their revised count, although the program specified that corrected counts
would not affect either reapportionment or redistricting. As the program proceeded, it became clear that a
number of group quarters—including long-established prisons and college dormitories—had been coded to
the wrong town or county.” See REENGINEERING THE 2010 CENSUS: RISKS AND CHALLENGES 151 (Daniel
L. Cork, Michael L. Cohen & Benjamin F. King, eds., 2004).
32
See Peter Wagner, Importing Constituents (Prison Policy Initiative, Springfield, MA) (2002), at 9. Had
the census count occurred on January 1, 2004, that number would be closer to 36,575. See State of New
York Department of Correctional Services, HUB System: Profile of Inmate Population Under Custody on
January 1, 2004 (2004), at 10.
33
See Nancy G. La Vigne & Cynthia A. Mamalian with Jeremy Travis & Christy Visher, A Portrait of
Prisoner Reentry in Illinois, (Urban Institute, Wash. D.C.) (2003); see also Building Bridges: From
Conviction to Employment: A Proposal to Reinvest Corrections Savings in an Employment Initiative,
(Council on State Governments, Lexington, KY) (2003), at 7.

“Home” in 2010
Page 16 of 22
prison are allowed to exclude this population for purposes of redistricting. Colorado,
Mississippi, and New Jersey have all adopted similar such provisions.34
The impetus for these adjustments is completely reasonable, and officials should be
credited for recognizing the serious distortional effect the prison count has on the
distribution of political power and funds in their states. Yet these state-specific provisions
ultimately create a patchwork of interpretations and incongruous state policies, many of
which are tantamount to a de facto undercount. As the primary source of redistricting
data, only the Census Bureau can ameliorate the situation by identifying more accurate
addresses for those in institutional quarters and thereby encouraging uniform redistricting
practices across states.
Counting people in prison as residents of the facility in which they are incarcerated also
contravenes the Census Bureau’s stated goal of fairness. It creates districts where
politicians are by definition unaccountable to their disenfranchised constituents, yet
benefit from their presence. (This is true of every person in prison, even those who will
never leave.)35 And it gives rural communities with prisons an arbitrary advantage over
those without prisons in competition for scarce rural-based federal and state funds.36

34

For a full list with explanations, see Peter Wagner, State officials tell counties to exclude prison
populations from county level redistricting, available at http://www.prisonersofthecensus.org/news/fact-111-2004.shtml.
35
While Maine and Vermont allow all people with felony convictions—even those currently
incarcerated—to vote, incarcerated people vote by absentee ballot in their home districts. Council on State
Governments, Eastern Regional Conference, Felons and the Vote: Lawmakers Examine Issue as Elections
Draw Near, Weekly Bulletin, Issue 43, September 22, 2004, available at:
http://www.csgeast.org/page.asp?id=weeklynewsbulletin43.
36
See Peter Wagner, Financial burden of how prisoners counted in Census falls on rural—not urban—
communities, December 8, 2005, available at http://www.prisonersofthecensus.org/news/fact-8-122005.shtml.

“Home” in 2010
Page 17 of 22
Option 2: Enumerating in the county of commitment
One solution to the predicament posed by those exceptional cases in which people have
no proper address is to assign them to their county of commitment—the county in which
they were convicted and from which they were sent to prison. County of commitment is
considered by many research groups to be the best geographic measure of postincarceration address short of “home of record.” According to a recent Justice Policy
Institute report, “[w]hile it does not precisely pinpoint where people lived prior to
imprisonment, commitment data provides a reasonable geographical overview.”37 While
certainly a second-best solution, using county of commitment is more accurate at the
collective level than counting people as residents of a prison miles from the “home” to
which they will return. Because parolees are supervised from and required to regularly
report to their county of commitment, in most cases their home address will be close by.38
For this reason, enumerating certain exceptional populations at their county of
commitment would in many ways minimize distortions produced by the current count.
There exist, however, select geographies for which this method of address determination
is seriously flawed. In New York City, for example, many people returning from prison
will be released to Manhattan, the city of their commitment, yet reside in Brooklyn
(King’s County).
In addition, county of commitment determinations do not attenuate the problems posed
by people in prison for life, nor those who will move to a relative’s new address. Nor do
they provide the block-level specificity necessary for redistricting. Because it is
ultimately only useful, then, for determining federal congressional apportionment, using
county of commitment information is effectively no fairer than counting people without
a prior address at the facility.39 Further, what we know about the state of commitment
records suggests that while it is available, in some form, in all states, questions remain
about its accessibility in court or DOC records.40
Finally, county of commitment information is unlikely to meet the Bureau’s standards of
verifiability: unmoored to a precise physical location, like street address, it cannot be
verified in the manner on which the Bureau insists.

37

Eric Lotke & Jason Ziedenberg, Tipping Point: Maryland’s Overuse of Incarceration and the Impact on
Public Safety, (Justice Policy Institute, Wash. D.C.) (2005), at 10.
38
See supra note 21.
39
The clear exception to this rule, of course, is people in federal prison, many of whom would be nominally
rerouted to their home states.
40
See Peter M. Brien, Improving Access to and Integrity of Criminal History Records (Bureau of Justice
Statistics, July 2005).

“Home” in 2010
Page 18 of 22
Option 3: Creating a correctional facility tract
Another possible way to handle exceptional cases derives from a suggestion made by
William Cooper, a Virginia-based demographer with FairData 2000.41 Cooper
recommended the following:
Rather than assigning inmates to census blocks, maybe the entire
incarcerated population could be taken out of the block-level PL94-171
and assigned to a separate census entity for each state that would be
equivalent to a county. This would be analogous to the 1990s practice of
putting certain crews on vessels into off-shore census tracts.
This way, the incarcerated population could still be used for determining
[C]ongressional apportionment, but would not be a factor in
[C]ongressional and state/local redistricting.42
This solution seems most appropriate not for the entire incarcerated population, but
instead solely for those 20% for whom the Bureau cannot locate or verify a street-level
address. The appeal of Cooper’s suggestion (hereinafter the “prison tract” option) lies in
its offer of a fixed standard for dealing efficiently with anomalies and outliers throughout
the state—those who were homeless or living in a shelter at the time of arrest, those
sentenced to life without parole, and those for whom the provided address cannot be
verified or geo-coded. In accordance with “one person, one vote” principles, 43 it is fairer
that these groups count only for the purposes of Congressional apportionment and not for
the benefit of local and state politicians who are not accountable to them.44 Perhaps more
important, from a data user’s perspective, it is better that they not distort local
41

http://www.fairdata2000.com/.
Email from William Cooper, Demographer, FairData 2000, to the Brennan Center for Justice (January 1,
2006) (on file with the Brennan Center for Justice). “For the 1990 census, the .99 suffix was reserved for
census tracts/block numbering areas (BNAs) that contained only crews-of-vessels population; for Census
2000, the crews-of-vessels population is part of the related census tract.” See
http://www.census.gov/geo/www/cob/tr_metadata.html. “In 1990, the Census Bureau represented crewson-vessels in their electronic boundary files. This differed from previous and later decades in which the
crews-on-vessels populations were added to the nearby land census tracts with which the vessels were
associated. In an attempt to provide consistency over time, crews-on-vessels have been removed from all
tract boundary files. Separate statewide point shapefiles have been created for 1970, 1980 and 1990
representing crews-on-vessels within the state in which the crews-on-vessels were located in that census
year. These shapefiles are designated by state + decade + ‘cov.’” See
http://beta.nhgis.org/geography/help.html#cov.
43
Gray v. Sanders, 372 U.S. 368 (1963) (Douglas, J.) (“The conception of political equality from the
Declaration of Independence, to Lincoln's Gettysburg Address, to the Fifteenth, Seventeenth, and
Nineteenth Amendments can mean only one thing—one person, one vote.”).
44
“Assuming perfect voter turnout, the victorious candidate in a prison system district will have been
elected by fewer people than the candidate in a district with no prison. Even considering the fact that voter
turnout will vary across districts, an elected official in a no-prison district will effectively be responsible
for, and accountable to, more constituents than the official whose district contains a large, disenfranchised
prison population. The result is the unequal weighting of votes across district lines—a practice that does not
stand on firm constitutional ground.” Rosanna M. Taormina, Defying One-Person, One-Vote: Prisoners
and the ‘Usual Residence’ Principle, 152 U. PA. L. REV. 431, 433-34 (2003).
42

“Home” in 2010
Page 19 of 22
demographics in prison-housing communities. In effect, therefore, this method is as
accurate as using county of commitment information, which, too, is helpful only for
determining Congressional apportionment.
Also working in favor of the prison tract option is the practical irrelevance of standards of
availability and verifiability. Unlike county of commitment standards, which present the
Bureau with the arduous task of reconciling heterogeneous court and state DOCs records,
assigning individuals to a prison tract requires no further address specification than the
state in which the individual is incarcerated. And because the prison tract is a fictional
location, it cannot and need not be verified. Finally, as Cooper makes clear, this is a
method with which the Bureau is already familiar.
Option 4: Interviewing and distributing forms in prison
A final solution for exceptional cases lies outside the reach of extant administrative
records; here the definitional dilemma is resolved in a procedural fix. Instead of relying
on divergent state data sets, the Bureau could interview and distribute the short form to
people in prison at their respective facilities, as described above. This would allow people
soon to be released yet with no prior address the opportunity to identify the street address
to which they will return. Likewise, someone whose family has moved since their
incarceration could report their family’s current address. Working with correctional
officials familiar with those incarcerated in their facilities, census field workers normally
assigned to rural areas could include the prison as part of their regular enumeration. In
past censuses, and under contract with the Bureau of Justice Statistics, the Census Bureau
has sent enumerators into prisons to conduct interviews.45 Here, again, the prison tract
could absorb all unverifiable addresses, and serve as a repository for tabulating those
serving life without parole.
While the interview method would force the Bureau to increase the number of people
interviewed, it offers a reward of high response rates and uniform data consistency. It
requires the Bureau to verify more addresses, a larger job than simply assigning all
exceptional populations to a fictional prison tract, but hardly an impossible one,
especially within the scope of its usual enumeration. Again, adopting this method would
represent not a shift or change in practice but simply an expansion of interview methods
already known to the Bureau.46 In other words, the information is eminently available;
the Bureau simply needs to go out and collect it.
45

According to Bureau of Justice Statistics (“BJS”) Statistician Allen Beck, BJS regularly employs the
Census Bureau to go into prisons and jails and conduct interviews for their inmate surveys. Beck estimated
that the most recent survey, conducted in 2003 and 2004, required the Bureau’s full field staff of about 250
employees. The results of these interviews form the content of BJS’s periodic “Survey of Inmates in Adult
State and Federal Correctional Facilities,” upon which any subsequent reports focused on the
characteristics of people in prison—for example, “Incarcerated Parents and Their Children” and “Women
Offenders”—are based. The most recent publicly available survey was conducted in 1997. For more
information on the State and Federal Adult Correctional Facilities Census, see
http://www.census.gov/econ/overview/go2400.html.
46
In 2000, the Census Bureau “used enumerators to take the census in rural areas and check on
questionnaires that had not been returned by mail in more populous areas.” U.S. Census Bureau, Strength

“Home” in 2010
Page 20 of 22

IV. The Importance of Uniform, Accessible Prison Data
Information for Policymakers
In recent years, concern about prison reentry has blossomed. Perhaps the clearest
indication of this sea change in official public rhetoric was President Bush’s
announcement of a four-year, $300 million reentry initiative in his 2004 State of the
Union address.47 In order for census data to be useful for reentry purposes, each
reassigned address—with the exception of those aggregated and released with PL94-171
tables—should carry with it an institutional tag to denote that the person to whom it
belongs is currently in a correctional facility. Without such a provision, census data users
will be unable to determine, at a collective level, which neighborhoods should be the
focus of reentry efforts.
This point raises the further question of whether it is important for data users to know,
through the census, the population and location of a given prison or jail on census day.
One can imagine reasons why this information would be useful—the clearest being the
benefit of a uniform facility-level enumeration across states occurring on a single day.48
University of Pennsylvania Law Professor Nathaniel Persily suggests having “a special
form for prisoners, which allows them to specify both their original residence and their
prison residence. More importantly,” Persily continues, “the Census Bureau would need
to offer (yet again) an additional dataset that would allow states and localities to redistrict
based on the reassignment of prisoners to their original communities.”49
Persily recommends that the Bureau provide both sets of data, leaving it up to
states to decide how to use the data. We recommend instead that the Bureau first
release PL94-171 data using prisoners’ home address for purposes of redistricting
and then, at the conclusion of redistricting determinations, follow this with a
subsequent release of both institutional and home addresses in Summary Files 1
and 2. Alternatively, the Summary File 1 could mirror the format of PL94-171,
and an additional data set including only the addresses and populations of
correctional facilities on census day—call it SF-1A—could be created to

in Numbers, Your Guide to Census 2000 Redistricting Data From the U.S. Census Bureau, July, 2000
available at http://www.census.gov/clo/www/strenghth2.pdf, 2.
47
“Tonight I ask you to consider another group of Americans in need of help. This year, some 600,000
inmates will be released from prison back into society. We know from long experience that if they can't
find work, or a home, or help, they are much more likely to commit crime and return to prison. So tonight, I
propose a four-year, $300 million prisoner re-entry initiative to expand job training and placement services,
to provide transitional housing, and to help newly released prisoners get mentoring, including from faithbased groups. (Applause.) America is the land of second chance, and when the gates of the prison open, the
path ahead should lead to a better life. (Applause.)” President George W. Bush, State of the Union Address,
(January 20, 2004), available at: http://www.whitehouse.gov/news/releases/2004/01/20040120-7.html.
48
Besides employing different enumeration methods, state DOCs differ in the day on which they provide
annual facility-level population estimates.
49
Nathaniel Persily, “The Law of the Census: How to Count, What to Count, Whom to Count, and Where
to Count Them,” in Title TBA, 35 (Russell Sage Press, forthcoming 2005) available at
http://www.law.upenn.edu/fac/npersily/workinprogress/persily.census.8.6.doc.

“Home” in 2010
Page 21 of 22
complement the PL94-171 and Summary Files.50 Such a strategy is properly
homologous to the primary and secondary purposes of the census: first, to allocate
political power and funding and second, to provide detailed aggregate statistics
for data users. As former Bureau Director Kenneth Prewitt explains, “The initial
political purpose of the census is to adjust the regional distribution of power to
match changes in the distribution of the population…The laws under which
federal funds are distributed specify what groups and areas will receive them.
Spending formulas routinely incorporate population size as a factor.”51

50

We owe this insight, as well, to William Cooper of FairData 2000.
See Kenneth Prewitt, Politics and Science in Census Taking, October 2003, available at
http://www.prb.org/PrintTemplate.cfm?Section=PRB&template=/ContentManagement/ContentDisplay.cf
m&ContentID=10113.

51

“Home” in 2010
Page 22 of 22

Conclusion
Congress has directed the Census Bureau to re-evaluate the manner in which it counts
people in prison, a process that requires consideration of several definitional and
operational issues. Developing conclusive findings on the feasibility of using home
addresses for people in prison requires the Bureau to consider how it should define
“home” for the purpose of enumerating the prison population, what methods are best for
counting the majority of people in prison—those with accessible and verifiable home
addresses—and what methods are best for counting exceptional cases in which people in
prison cannot be counted at their home addresses.
Securing additional funding and resources is a logical and necessary next step in
determining how to count people in prison back home. But criminal justice and DOCs
data experts agree: though there are many practical details to work through, if the
commitment to do so exists, people in prison on April 1, 2010 can and should be
enumerated at their home addresses.