Prison Populations and the Census - FAQ
Last update: May 06, 2015
Show All Answers
- Q: Does the census count incarcerated people as if they were residents of the towns where they are incarcerated?
- A: Yes.
- Q: How does the Census count incarcerated people?
- A: The Census sends forms to every American household with the question "How many people were living or staying in this house, apartment or mobile home on April 1, 2010?" The form instructs respondents to exclude certain residents, including household members incarcerated in correctional facilities. To count incarcerated people, census takers visit prisons and distribute forms or collect the necessary data from administrators. Regardless of how their forms are filled out, the Census assigns prisoners to the address of the prison.
- Q: Have incarcerated people always been counted at the prison?
- A: Yes. However, prior to the 1990 Census, prisoners were not explicitly excluded from the census counts of their actual residences. This mattered little until recently, when the prison population exploded and people were incarcerated further from their actual residence. There are now enough people incarcerated that prisons can be used to shift political power.
- Q: Are people incarcerated in federal prisons counted the same way?
- A: Yes, even though people under the jurisdiction of the Federal Bureau of Prisons are often incarcerated out of state, they are still counted at the location of the prison, not at at their address of residence.
- Q: Why does the Census need to issue instructions on where to count people?
- A: The Census is trying to be sure it counts everyone, and only counts them once. To ensure consistency, the Bureau has developed the "usual residence rule."
- Q: What is the "usual residence rule"?
- A: The "usual residence rule" is how the Census Bureau determines where people should be counted.
"For the 2010 Census, the Census Bureau is committed to counting every person. Just as important, however, is the Census Bureau's commitment to counting every person in the correct place. The fundamental reason the decennial census is conducted is to fulfill the Constitutional requirement (Article I, Section 2) to apportion the seats in the U.S. House of Representatives among the states. Thus, for a fair and equitable apportionment, it is crucial that people are counted in the right place during the 2010 Census."
Source: Census Bureau "Residence Rule And Residence Situations For The 2010 Census".
- Q: What does the usual residence rule mean?
- A: In general, it means that people will be counted where they "eat and sleep most of the time." For example, people on business trips on Census Day will still be counted at home. For most groups, the rule is obvious, but for others it is more complex. And for some groups -- like prisoners, who live in a place temporarily and involuntarily -- it is controversial.
- Q: Is "usual residence" the same as a legal or voting residence?
- A: No. "Usual residence" is the Census' internal definition of residence. The Census Bureau collects and publishes data based on its own definition of residence, but when states use this data as if it reflected people's actual residence for redistricting purposes districts do not reflect actual populations, and that skews the equal distribution of voting power.
- Q: Even though incarceration does not legally change someone's permanent residence, don't many people go to prison for life, and become in effect, residents of the community where the prison is located?
- A: Actually, most people in prison are serving only short sentences. Most are only away from their homes and communities temporarily, for less than 3 years. During their relatively short term of incarceration, in nearly all respects the person remains a member of his or her home community; incarcerated people do not generally establish ties and become integrated with the community where the prison is located.
- Q: Does the "usual residence rule" require incarcerated people to be counted at the facility and not at home?
- A: The Census justifies counting incarcerated people at the prison with the usual residence rule. Prisoners do "eat and sleep most of the time" in prison, but Census and Congress have the power to change this rule, as they have done for other special populations in the past.
- Q: For what groups has the usual residence rule been changed?
- A: It's been changed for college students, missionaries, and overseas Americans and other groups, and often more than once.
- Q: How could the Census count prisoners differently?
- A: The Census could allow prisoners to declare their own addresses, or use Department of Corrections' files to determine their homes of record. The Census has used both methods successfully to count military personnel.
- Q: Would it really cost the Census $250 million to count incarcerated people at home?
- A: That is extremely unlikely. (The Maryland Department of Legislative Services, for example, estimates the cost of counting incarcerated people at their residential address to about $1.60 per prisoner, that would be about $4.2 million for the whole US prison population.) That figure comes from a report the Census Bureau prepared for Congress in 2006: Tabulating Prisoners at Their "Permanent Home of Record" Address.
Congress asked the Bureau for its opinion on changing the prison count, and rather than describe the problem and ask for helped resolving it. The Bureau threw up a smokescreen. The New York Times was blunt: “The Census Bureau tends to stamp its feet and shake its head no when asked to do things differently than it has in the past.” The editorial board neatly summarized the Bureau's report as “obtuse and evasive”.
The $250 million figure is the most often cited piece of the report, but the Bureau provides no breakdown of the costs. Worse, in the interest of stopping reform, the Bureau inflates minor methodological issues into unnecessary work. For example, while some incarcerated people – like people anywhere – will be incapable of filling out forms by themselves, that does not mean that each and every able-boded incarcerated person would have to be individually interviewed in a private room. Similarly, the Bureau conflates the simple problem of making sure that an address provided by an incarcerated person actually exists on the Bureau's maps with an extremely difficult and completely unnecessary process of interviewing people at the alternative address to make sure they know the incarcerated person.
The real tragedy is that this 2006 report reflects a missed opportunity to improve the 2010 Census. Instead of raising doubts and worries, the Bureau should have been seeking solutions. The Bureau should have described the problem and asked for a small appropriation to study the best way to collect and process home address information from people incarcerated in federal, state and local prisons and jails. There are indeed methodological questions that must be resolved with research prior to changing where incarcerated people in prison are counted. It is not rocket science, but even simple problems can loom impossibly large when you wait too long.
The Bureau should have spent the previous decade seeking ways to improve the prison count for 2010 Census. Instead, because the Bureau squandered the necessary planning time on obstructionist arguments, our focus must shift to changing how the 2010 Census data is used and having incarcerated people counted at home in the next Census in 2020.
- Q: Prisoners can't vote. Why does the Census count them?
- A: The Census counts everybody, including children and non-citizens, neither of whom can vote.
- Q: If prisoners could vote, would they vote at home or prison?
- A: No. The two states that allow prisoners to vote (Maine and Vermont), require them to vote by absentee ballot; states that have only recently disenfranchised prisoners (Massachusetts and Utah), also required the same absentee voting procedure. Most states also have constitutional provisions stating that incarceration does not add or change a residence for voting purposes. (For more information, see Absentee Prisoner Voting in Maine and Vermont.)
Aditionally, people in jail often still have the right to vote, people in jail are required to vote absentee using their residential address.
- Q: Since incarcerated people can't vote, who suffers when districts are drawn based on prison populations?
- A: Counting prison populations as if they were actual constituents of the prison location gives a few small communities more political power at the expense of everyone who does not live near a prison. In effect, everyone who does not live in a district that contains a prison has their vote diluted by these artificially inflated populations.
- Q: What about college students?
- A: College students are counted at their colleges, but prior to the 1950 Census, they were counted at home. Students in boarding schools below the college level are still counted at home.
- Q: How are students different from prisoners? If we change how prisoners are counted, wouldn't we also need to change how we count students?
- A: Changing how prisoners are counted would not require a change in how students are counted. Students are welcome and encouraged to purchase local goods and services and to rent apartments in town. They can register to vote in the community and may decide to stay after graduation. While the majority of students will eventually leave their college towns, there is no evidence to suggest that they are likely to return to their parents' communities. For the duration of their time at the college, the college is the place they willingly live: that's the very definition of residence.
Unlike college students, prisoners are not a part of the local community, and while some colleges are partially self-reliant communities, prisons are entirely so. Prisoners are barred from leaving the facility without official permission and cannot take part in local elections. Most return to their home communities after release, as parole regulations frequently require prisoners to return to their county of conviction.
- Q: Do state constitutions define residence differently than the U.S. Census's usual residence rule?
- A: Yes. Most states define legal residence as the place that you choose to be until you intend to go somewhere else. Many state constitutions - Arizona, Colorado, Michigan, Minnesota, Missouri, Nevada, New York, Oregon, and Washington - specifically declare that incarceration does not change a residence. Most other states have this definition codified in their statutes.
- Q: If the Census has counted prisoners this way since 1790, why should it change?
- A: The Census Bureau's mission and American demographics have changed drastically in the 217 years since the first Census, and the Census has evolved in response to many of these changes. The Census' original mission was to count the number of people in each state for purposes of Congressional apportionment. It only mattered how many people New York had compared to Pennsylvania, not whether the Census Bureau counted New Yorkers at Attica Prison or in the Bronx.
Times have changed. Not only has the incarceration rate quadrupled since 1980, the Census's Bureau's purpose has broadened considerably. A series of Supreme Court cases in the 1960s established the "one person, one vote" rule, which requires states and localities to redistrict at least once per decade to match evolving population patterns. The Bureau now provides state and local information suitable for use in local redistricting, as well as the data for all kinds of government planning.
Census policy is not fixed; instead it responds to changing needs. When evolving demographics meant more college students studying far from home and more Americans living overseas, Census policy changed in order to more accurately reflect Americans' diverse residences. Today, the growth in the prisoner population requires the Census to update its methodology again.
- Q: Does the movement of prisoners across state lines affect Congressional apportionment?
- A: No. Our review of the apportionment numbers, and our research on the flow of correctional populations across states lines says that in 2010, apportionment was not affected by the prison count.
- Q: Besides state constitutions and voting rights law, what do definitions of residence in other contexts say about the actual residence of people in prison?
- A: Most states have constitutional clauses or statutes that define residence for incarcerated people to be the place they lived prior to incarceration. Similar rules exist in other contexts, such as determining where a person may file a lawsuit. Judge Wade H. McCree most clearly summarized the rule and its rationale in a 1973 decision:
It makes eminent good sense to say as a matter of law that one who is in a place solely by virtue of superior force exerted by another should not be held to have abandoned his former domicile. The rule shields an unwilling sojourner from the loss of rights and privileges incident to his citizenship in a particular place....
Stifel v. Hopkins 477 F.2d 1116, 1121 (1973).
- Q: Where can I learn more about redistricting and the census?
- A: The Census Bureau has an excellent 11 page overview: Strength in Numbers: Your Guide to Census 2000 Redistricting Data [PDF]. The first half is an introduction to the Census Bureau and its role in redistricting, and the second half explains the use of the data.
We've also written two articles that explain how redistricting works:
- Q: Hasn't this been litigated? What about Rodriguez v. Pataki and D.C. v. Commerce?
- A: The prisoner miscount's impact on redistricting has never been the subject of litigation as a voting rights case above the county level. The impact on state legislative redistricting, which is the focus of our work, has never been litigated. Rodriguez, a 2003 lawsuit challenging the New York State Senate districts as a dilution of minority voting strength in New York City, did not raise the prisoner issue as a claim. Instead, the lawsuit discussed the prisoner miscount as an important context within which to view the existing numerical disparity between upstate and downstate districts.
D.C. v. Commerce, 789 F. Supp. 1179 (1992), was a suit filed by the District of Columbia against the Census Bureau because Washington D.C.'s prisoners were counted as residents of Lorton, VA, the location of the D.C.-owned and -operated prison. D.C. sued the Census Bureau for allegedly costing the district Congress-distributed federal funds. Voting rights were not a part of this lawsuit.
The Census data has many uses, but its only constitutional mandate is congressional reapportionment. Because federal funds are not a fundamental right, the court viewed the case under a standard called "mere rationality," meaning that the District would need to show that the Census' usual residence rules were "arbitrary and capricious." This is a difficult standard to meet, and the lawsuit was correctly dismissed. Had this been litigated as a voting rights case, however, the standard and the outcome would have been very different.
- Q: Who would the ideal parties be in a voting rights suit based on the Census' prisoner miscount?
- A: The ideal lawsuit would be brought on behalf of voting-eligible members of high-incarceration neighborhoods against a state legislature. The argument would be that the state legislature violated the 14th Amendment's equal protection clause and -- in most states -- the state constitution by using Census Bureau data. "One person, one vote," and most states' definition of residence, require a redistricting dataset that counts people at their actual homes.
Our position, and that of D.C. v. Commerce, is that the Census Bureau has a very limited constitutional mandate: to count the population of each state for purposes of reapportionment. Congress has expanded this mandate to include providing redistricting data to the states and other research purposes, and it is possible that this expansion of the Census' role could give rise to Bureau liability for aiding the dilution of urban and minority voting strength.
- Q: Are states required to use the federal census when conducting redistricting?
- A: While states must redistrict on the basis of actual population, the U.S. Constitution does not require states to use the federal census for its own redistricting:
"Although a state is entitled to the number of representatives in the House of Representatives as determined by the federal census, it is not required to use these census figures as a basis for apportioning its own legislature." Bethel Park v. Stans, 449 F.2d 575, 583 (3rd Cir. 1971)
States are therefore free to use their own census or to correct how the federal census counts prisoners. In fact, making adjustments is quite common.
- Q: Has this been litigated as a voting rights case on the county level?
- A: Yes. There are at least six cases in New York (and perhaps in other locations), although many of these issues are complicated by questions of standing and courts' reluctance to get involved after the fact. Many counties have dropped imported prison populations from their county redistricting dataset after a public outcry.
- Q: What are the important Supreme Court cases on redistricting?
- A: Although many states required their districts to be redrawn periodically to keep them equally sized, there was no federal requirement and these state-law requirements were frequently ignored. Until 1962, the Supreme Court maintained that these were non-justiciable political questions and refused to rule. These are the major cases that define Supreme Court jurisdiction and the U.S. constitution's requirements for legislative democracy.
Establishing federal jurisdiction and the underlying principles:
Baker v. Carr, 369 U.S. 186 (1962). Baker involved the Tennessee Assembly and its failure to redistrict in 60 years, despite significant population growth. This was the first case to decide that malapportionment violates the equal protection clause of the U.S. constitution.
Wesberry v. Sanders, 74 U.S. 1 (1964). This case struck down Georgia's congressional districts, which had the 5th district two to three times larger than other districts in the state.
"We hold that, construed in its historical context, the command of Art. I, 2, that Representatives be chosen "by the People of the several States" means that as nearly as is practicable one man's vote in a congressional election is to be worth as much as another's."
Reynolds v. Sims, 377 U.S. (1964). Reynolds was the first case to declare the "one person, one vote" principle, deciding the case under the equal protection clause:
"Legislators represent people, not trees or acres. Legislators are elected by voters, not farms or cities or economic interests. As long as ours is a representative form of government, and our legislatures are those instruments of government elected directly by and directly representative of the people, the right to elect legislators in a free and unimpaired fashion is a bedrock of our political system"
Reynolds did not, however, bar all deviations from strict population equality.
Establishing limits on population deviations between districts:
Congressional districts are not allowed to deviate from population equality at all. State legislative districts are allowed a limited amount of flexibility. Generally speaking, states can be required to justify any deviations in their state legislative districts over 10%.
Abate v. Mundt, 403 U.S. 182 (1971) upheld a 16.4% deviation between districts because there was a compelling state interest in "maintaining the integrity of political subdivision lines." The alternative plans presented were not able to draw the districts more equally without splitting up many towns and counties.
White v. Regester, 412 U.S. 755 (1973) held that a state with district size deviations of less than 10% need not be supported by any state objective. However, a recent District Court for the Northern District of Georgia case (Larios v. Cox) about a deliberate regional bias in Georgia's redistricting plan suggests that the 10% rule is not a safe harbor.
Karcher v. Daggett, 462 U.S. 725 (1983) requires Congressional districts in each state to be exactly equal in size.
- Q: What is Gerrymandering?
A: "Since a single party usually controls each state legislature, it is in the best interest of the party in power to redistrict their state so that their party will have more seats ... than the opposition party. This manipulation of electoral districts is known as gerrymandering....
"The term gerrymandering is derived from Elbridge Gerry (1744-1814), the governor of Massachusetts from 1810 to 1812. In 1812, Governor Gerry signed a bill into law that redistricted his state to overwhelmingly benefit his party, the Republican Party. The opposition party, the Federalists, were quite upset. One of the congressional districts was shaped very strangely and, as the story goes, one Federalist remarked that the district looked like a salamander. No, said another Federalist, it's a gerrymander. The Boston Weekly Messenger brought the term gerrymander into common usage when it subsequently printed an editorial cartoon that showed the district in question with a monster's head, arms, and tail and named the creature a gerrymander."
- Q: As a redistricting professional, how can I best access the Census Bureau's prison count data?
- A: In 2011 the Census Bureau published the Advance Group Quarters Summary File, which is still available in the original through the Bureau. We have since, however, produced several easier ways to access this data, including a web-based county search tool, and a shapefile. Each of these contains the correctional populations by block. The web-based tool contains additional annotations (facility identification and basic demographic data) for each block with a correctional facility. The shapefile contains data for all types group quarters, but includes links to our block-level annotations for the blocks that contain correctional facilities.
- Q: As a redistricting professional, how can I know what correctional facilities are in particular Census Blocks?
- A: The Census Bureau publishes its data aggregated to individual blocks, block groups or tracts; but sometimes it's helpful to know and be able to talk about a particular correctional population. For example, it might be helpful to distinguish a state prison in one block from a federal prison or local jail in another nearby block. For that reason, we've annotated most of the blocks that contain correctional facilities with the facility name and type. As of 2013, these annotations are largely complete, but we do continue to accept corrections and additions. Our annotations were drawn from comparisons with government surveys of federal prisons, state prisons, local jails and Indian Country jails.
- Q: As a redistricting professional, how can I determine if prison gerrymandering — or a decision to avoid prison gerrymandering — impacts the voting rights of communities of color?
- A: In order to perform this analysis you need to know the race and ethnicity of the incarcerated population. Traditionally, redistricting professionals have had to estimate this data, but we've published the race and ethnicity of the correctional population for every Census Block that contains a correctional facility as counted by the U.S. Census. Previously, this data was unavailable because the Census Bureau doesn't publish this data in one table, but we discovered that by comparing 5 separate tables we could derive 9 race and ethnicity categories for the reported incarcerated population for virtually every block. We published this race and ethnicity data — and identify the specific source tables — for each block with a correctional facility. This data is available in the block level results of our county search tool through the "detail" link in the right-hand column of your search results, and through links within the national group quarters shapefile.
- Q: Does state legislation to end prison gerrymandering create the risk that the urban legislative districts would be unable to elect an African-American or Latino candidate?
- A: No. Despite the seemingly counter-productive addition of non-voting populations to those districts, legislation that would reallocate incarcerated people to their home addresses for redistricting purposes would not create a risk of urban districts being unable to elect African-American or Latino candidates.
This legislation provides the data necessary to account for the presence of incarcerated populations in the home communities: We’ll know exactly how many people in any given district can’t vote because they are incarcerated, and we’ll know the race and ethnicity of the adjusted population for each block. In fact, compared to the estimates on citizenship and political turnout data that are currently used to evaluate how districts will actually perform, the data produced by the bill will be more precise. There is no risk that these districts will not perform as intended.
In fact, such legislation would likely help draw majority-minority districts in high-incarceration communities because properly crediting these districts with their true populations will make it easier to draw districts around their African-American or Latino cores.
Prison-based gerrymandering in other contexts:
- Q: Your research shows how specific rural prison communities get extra voting power. Can you be specific about how individual urban neighborhoods see their votes diluted?
- A: Unfortunately, the available data only allows us to demonstrate the harm done to specific counties and cities. Prisoner miscount cost New York City 43,740 residents, for example, but we can't go more deeply that that. Other research establishes that incarceration is highly concentrated in particular urban neighborhoods, but we can't yet say how many people should have been counted within each legislative district.
Although most states collect home residence information, this is not generally available or easily analyzed. (And, most critically, the U.S. Census does not currently gather it.) Thanks to other researchers, we can compare incarceration rates by neighborhood in Brookyln, NY [PDF] and again in more detail, New York City, New Haven, CT [PDF], New Haven and Hartford, CT [PDF], Baltimore, MD [PDF], Newark and Camden, NJ [PDF], Cleveland, OH [PDF], Dallas, Forth Worth, San Antonio, Austin, and Houston, TX [PDF], Providence, Rhode Island [PDF], and New Orleans, LA [PDF].
These reports allow us to compare incarceration rates in these cities with some specificity. Each of these reports has a different methodology, but it's not possible to directly put this data into our Census districting analysis for a number of reasons:
Things would be simpler if we could count the incarcerated residents lost by a particular urban New York State district, but right now we can precisely document the site of the largest distortions: in the rural prison districts. Our existing research allows us to specifically describe how prisons artificially boost the population of prison towns, about how that increases the weight of a prison-town vote, and how that dilutes the weight of a vote by the resident of an average district. For example, New York State Assembly District 114 is 7% prisoners, so the votes of 93 residents in that district carry the weight of 100 residents elsewhere in the state.
- They are generally some years removed from the April 2000 Census.
- The reports are often based on releases from or commitments to prison in a year, rather than the number of people in prison from a certain place on a particular day. Disparities in sentencing make it difficult to extrapolate population numbers.
- These reports generally count incarcerated people by zipcode, neighborhoods, police precinct, Census tract, or Census block group. In order to do a districting analysis, we need much more specific data, down to the Census block level.
Impact on funding
- Q: When states adjust Census population data for redistricting, is funding affected?
- A: There is no effect on the distribution of federal or state funds, because all funding programs have their own data sources that do not rely on redistricting data.
- Q: What about the American Community Survey?
- A: The American Community Survey (ACS) is counts incarcerated people at the location of the prison. The Survey is conducted by the Census Bureau, but is conducted more often and for different reasons, it uses different rules of residence than for the decennial census. The Prison Policy Initiative does not seek to change ACS residence rules.