This time next year, the 2020 census will be in full swing. For the first time, the Census Bureau hopes to conduct much of the census count online. A lot is at stake and public and private organizations are mobilizing to assure that the new method counts “everyone once, only once, and in the right place.”
The census is a huge challenge, but the basic idea is straightforward: First, inventory with complete specificity all of the housing units and group quarters — prisons, dormitories, nursing facilities, etc. — in the United States. Second, reach each of those housing units and group quarters and find out who lives there as of Census Day (April 1 of the decade year).
For some decades, the Census has done the second step in two basic phases. They have started by mailing surveys to each housing unit. If there is no mail response from a housing unit, then the census sends an enumerator to determine if someone is living there and attempt to get a response.
In 2020, people will receive still receive mail from the census, but the mail will request that they go online and complete a census survey. It is not entirely predictable how well this will work. However, the census is not jumping in blindly — they are doing extensive field testing of their methods.
The per person costs of the count for those who respond to the initial mail approach are low. The census will be using multiple channels, including social media, to encourage people to respond to the mail.
Based on prior experience, however, the expectation is that only about 70% of the population will respond to mail, and in-person visits will be required for the rest. The visits are vastly more expensive and difficult to complete.
Neighborhoods vary in their responsiveness to census mail. Low literacy, low trust, and high mobility are all factors that depress mail response. Renters and “non-family households” ( like a group of students sharing an apartment) tend to have lower response rates.
“Hard to Count” census tracts are those with response rates in the lowest 20% nationally. The map of “Hard to Count” census tracts looks a lot like the map of lower income neighborhoods, but also includes some higher-income high-mobility neighborhoods. For example, many Allston and Brighton census tracts and even one tract in Back Bay are “Hard to Count”. Belmont and Watertown have no “Hard to Count” tracts.
Overall in the 2010 Census, follow up recounts suggest that the total count was just a little high — by 0.01% or 36,000 people. The census overcounted homeowners through duplication by 0.6% but undercounted renters by 1.1%. The rental difference and other differences translated into an overcount of non-Hispanic whites by 0.8%, while non-Hispanic blacks were undercounted by 2.1% and Hispanics by 1.5%.
Undercounts weaken representation in Congress and in state and local legislative districts. They also diminish the allocation of many federal funding programs.
A strong coalition is developing to assure a full count in Massachusetts. Community outreach in a variety of forms will help strengthen the response to the Census. I intend to help the outreach within my senate district. When the counting is done, I will be making heavy use of the statewide results as chair of the Senate’s Committee on Redistricting.
If anyone has proposals for how to reach out to any likely undercounted local groups, I am eager to hear them.
Questions and Answers
Is Massachusetts at risk of losing a congressional seat?
Assuming that current population estimates are validated by the decennial census, it appears that the answer is no. Massachusetts lost one seat in 2010 and has experienced reasonable population growth since then. A careful analysis by Election Data Services shows that Massachusetts would neither gain nor lose a seat based on the Census estimates as of July 1, 2018. The EDS analysis also extends demographic trends to 2020 on three alternative growth paths, but none of the scenarios show Massachusetts close to losing representation.
The congressional seat apportionment process basically works as follows: Every state is allocated a first seat, then the remaining 385 seats are allocated in turn according to a ranking methodology, where each additional seat for each state is ranked by the state’s population divided by the geometric mean of the additional seat number and the prior seat number.
So, for example, in 2018, the 9th seat for Massachusetts has a priority level of 6,902,149/ square_root_of( 9*8). This priority level makes the 9th seat of Massachusetts the 403rd seat allocated — in other words, Massachusetts 9th seat would have to drop 32 positions before it would be lost. The 9th seat is at worst 405th in the three alternative Election Data Services scenarios. On a population percentage basis, the drop from 403rd to 436th (losing a seat) equates to an 8% population loss over the next two years. For a sample computation, see this spreadsheet.
How much does each undercounted person cost the state?
George Washington University has calculated that Massachusetts received $2,372 per capita in Fiscal 2015 based on statistics derived from the decennial census. However, the question of how much an individual undercount costs is much more complex than that. Program disbursements depend on many complex variables in addition to the raw counts.
For one example of the complexity, Medicaid accounts for over half of the amounts identified by GWU. MassBudget has done a careful analysis of the particular programs affecting of children in Massachusetts, including Medicaid. This analysis points out that for a wealthy state like Massachusetts, the reimbursement under Medicaid is not directly sensitive to an undercount.
We can say this much with confidence: The cost of an undercount depends on who is undercounted — undercounting people in economic need has the greatest consequence. At the same time, it is people in economic need who are most likely to be undercounted and towards whom most of the outreach efforts are targeted. To the extent program disbursements are, in fact, driven by the decennial census, the effects of an error will last a decade.
How does the census determine where to count people?
What is the general rule?
One fundamental use of the census is to apportion the Representatives in Congress. The constitution, as adjusted by the Fourteenth Amendment following the end of slavery, frames the base of apportionment as:
. . . the whole number of persons in each State . . .
In the decennial census, individuals are counted at their “usual residence”:
The Census Act of 1790 (See Section 5) established the concept of “usual residence” as the main principle in determining where people should be counted, and this concept has been followed in all subsequent censuses. “Usual residence” has been defined as the place where a person lives and sleeps most of the time. This place is not necessarily the same as the person’s voting residence or legal residence.2020 Census Residence Criteria and Residence Situations.
The phrase “the place where a person lives and sleeps most of the time” has been interpreted literally with the result that most people in group quarters are counted at the location of the group quarters, not at why they might think of as their home address. The final residence criteria for 2020 were published in the federal register in February 2018. The summary rules are:
- Count people at their usual residence, which is the place where they live and sleep most of the time.
- People in certain types of group facilities on Census Day are counted at the group facility.
- People who do not have a usual residence, or who cannot determine a usual residence, are counted where they are on Census Day.
How does the “usual residence” concept apply to college students?
In general, students are counted in their dormitories. This rule has been questioned but has been upheld in litigation. The rule was not the subject of much comment in the most recent review. Reproducing the exact language from the federal register:
(a) College students living at their parents’ or guardians’ home while attending college in the United States—Counted at their parents’ or guardians’ home.
(b) College students living away from their parents’ or guardians’ home while attending college in the United States (living either on-campus or off-campus)—Counted at the on-campus or off-campus residence where they live and sleep most of the time. If they are living in college/university student housing (such as dormitories or residence halls) on Census Day, they are counted at the college/university student housing.
(c) College students living away from their parents’ or guardians’ home while attending college in the United States (living either on-campus or off-campus) but staying at their parents’ or guardians’ home while on break or vacation—Counted at the on-campus or off-campus residence where they live and sleep most of the time. If they are living in college/university student housing (such as dormitories or residence halls) on Census Day, they are counted at the college/university student housing.
(d) College students who are U.S. citizens living outside the United States while attending college outside the United States—Not counted in the stateside census.
(e) College students who are foreign citizens living in the United States while attending college in the United States (living either on-campus or off-campus)—Counted at the on-campus or off-campus U.S. residence where they live and sleep most of the time. If they are living in college/university student housing (such as dormitories or residence halls) on Census Day, they are counted at the college/university student housing.(f) Staff members living in college/university student housing (such as dormitories or residence halls) on Census Day—Counted at the residence where they live and sleep most of the time. If they do not have a usual home elsewhere, they are counted at the college/university student housing.
How does the “usual residence” concept apply to prisoners?
This was the single subject of greatest controversy in the review of the criteria and the subject of a significant advocacy campaign. 77,863 comments out of a total 77,995 comments received on the criteria were comments urging the adoption of a rule that prisoners would be counted at their previous home address. The Massachusetts Senate was among the organizations urging the adoption of a new rule.
The standing rule has been that prisoners should be counted like everyone else — where they usually live and sleep on Census Day, in other words, in prison. The Census decided not to change this rule for 2020. All of the following will be counted where they are sleeping on Census Day: adult prisoners, jail inmates, ICE detainees, and people at halfway houses and similar facilities, including non-correctional residential treatment facilities. This rule also applies to juvenile correctional facilities.
The federal register includes a detailed summary of the arguments against this rule. The funding arguments against the rule are complex and program specific and not necessarily convincing, but the representation concern is clear. The Census recognizes the representation concern, but leaves it to the states to resolve by reallocating prisoners for within-state redistricting purposes:
States are responsible for legislative redistricting. The Census Bureau works closely with the states and recognizes that some states have decided, or may decide in the future, to `move’ their prisoner population back to the prisoners’ pre-incarceration addresses for redistricting and other purposes. Therefore, following the 2020 Census, the Census Bureau plans to offer a product that states can request, in order to assist them in their goals of reallocating their own prisoner population counts. Any state that requests this product will be required to submit a data file (indicating where each prisoner was incarcerated on Census Day, as well as their pre-incarceration address) in a specified format. The Census Bureau will review the submitted file and, if it includes the necessary data, provide a product that contains supplemental information the state can use to construct alternative within-state tabulations for its own purposes. However, the Census Bureau will not use the state-provided data in this product to make any changes to the official decennial census counts.Census response to comments.
It is not easy, in many instances, to determine home location for prisoners and it adds complexity to recompute census block populations for redistricting, but a few states have decided to attempt to do so, and several more have adopted measures excluding prison populations from local district counts.
Where are the homeless counted?
Reproducing the language from the federal register:
(a) People in domestic violence shelters on Census Day—People staying at the shelter (who are not staff) are counted at the shelter. Staff members are counted at the residence where they live and sleep most of the time. If staff members do not have a usual home elsewhere, they are counted at the shelter.
(b) People who, on Census Day, are in temporary group living quarters established for victims of natural disasters—Anyone, including staff members, staying at the facility is counted at the residence where they live and sleep most of the time. If they do not have a usual home elsewhere, they are counted at the facility.
(c) People who, on Census Day, are in emergency and transitional shelters with sleeping facilities for people experiencing homelessness—People staying at the shelter (who are not staff) are counted at the shelter. Staff members are counted at the residence where they live and sleep most of the time. If staff members do not have a usual home elsewhere, they are counted at the shelter.
(d) People who, on Census Day, are at soup kitchens and regularly scheduled mobile food vans that provide food to people experiencing homelessness—Counted at the residence where they live and sleep most of the time. If they do not have a usual home elsewhere, they are counted at the soup kitchen or mobile food van location where they are on Census Day.
(e) People who, on Census Day, are at targeted non-sheltered outdoor locations where people experiencing homelessness stay without paying—Counted at the outdoor location where they are on Census Day.
(f) People who, on Census Day, are temporarily displaced or experiencing homelessness and are staying in a residence for a short or indefinite period of time—Counted at the residence where they live and sleep most of the time. If they cannot determine a place where they live most of the time, they are counted where they are staying on Census Day.
Where are foreign citizens counted?
Citizenship and documentation status are not factors that affect census counting — the basic principle are the same as for all categories . . . reproducing the language of the federal register:
(a) Citizens of foreign countries living in the United States—Counted at the U.S. residence where they live and sleep most of the time.
(b) Citizens of foreign countries living in the United States who are members of the diplomatic community—Counted at the embassy, consulate, United Nations’ facility, or other residences where diplomats live.
(c) Citizens of foreign countries visiting the United States, such as on a vacation or business trip—Not counted in the census.
How are people in group quarters counted?
Group quarters include facilities like college dormitories, prisons and nursing homes. In the 2020 Census, group quarters will be able to submit electronic lists of residents minimizing the need for individual questionnaire completion or interviews of occupants in many cases. Facilities that serve transient populations raise special challenges. Arrangements for the count are being made with systematically with institutions in advance of the actual count date.