Part II
Residence Rules Meet Real Life: Challenges in Defining Residence



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Part II Residence Rules Meet Real Life: Challenges in Defining Residence

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census This page intentionally left blank.

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census –3– The Nonhousehold Population AN INEVITABLE TRUTH in every decennial census is that there are groups of people who are extremely difficult to count. In some cases, their living situations make it difficult to accurately gather data from them by standard enumeration techniques or even to locate them at all. In other cases, they may simply be unwilling or unable to provide accurate information even if a questionnaire reaches them. This is the first of two chapters in which we focus on groups of people who may have multiple residences (making it difficult to specify one “usual residence”) or whose ties to any fixed residence are ambiguous. In addition, we identify groups that are not explicitly covered by current census residence rules or that have historically proven difficult to count by the standard census methods and questionnaires. This listing of complex living situations is by no means exhaustive, but is intended to provide concrete examples of the breadth of difficulties in defining residence. In each case we attempt to give some indication of what is known about the size of the group; this is important because not all the groups are the same in terms of the magnitude of the problems they present to the census count. Ultimately, as the Census Bureau and other agencies work on approaches to reach these problematic groups, some sense of prioritization is needed in order to make effective use of time and resources. However, there are cases in which no real quantitative assessment of a group’s size is possible; instead, we rely on qualitative impressions. For each group, we also indicate how the group was handled under the 2000 census residence rules as well as past censuses.1 1 Our focus throughout this report, particularly in these two descriptive chapters, is on the

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census We begin in this chapter with some major segments of what the Census Bureau has traditionally termed the “group quarters” population (Section 3–A). Among these groups are students at colleges and boarding schools (3–B), patients of health care facilities (3–C), and persons serving terms in correctional facilities (3–D). As we argue in more detail in Chapter 7, we believe that term “group quarters”—indeed, the very concept—deserves reconsideration and revision. We prefer the nomenclature “nonhousehold population,” but still use “group quarters” in this and later chapters for consistency with past work. We also include in this chapter two groups that blend elements of group quarters and standard household enumeration. Children in foster care (3–E) are predominantly placed in individual family homes but may also reside in group home or semi-institutional settings, and they may transition between these settings with some frequency. Likewise, the domestic military population (3–F) includes both barracks housing on military bases as well as housing in surrounding communities. (Of course, the general issue of counting of the military population raises issues related to the enumeration of American citizens stationed or living overseas; we discuss the overseas aspects in Appendix C.) 3–A THE CONCEPT OF “GROUP QUARTERS” Before discussing large classes of group quarters, it is useful to first describe how the general concept has evolved in past censuses. Some tabulations of the 1900 census drew a distinction between “private families” and “families not private,” and the latter category was subdivided into the categories “hotel,” “boarding,” “school,” “institution,” and “other.”2 The 1930 census was the first to make a firm differentiation between places like institutions, prisons, and boarding houses as distinct from more conventional households. Dubbed “quasi-households” in that census and in 1940, and “nondwelling-unit quarters” in 1950, this segment of the population was subsequently renamed “group quarters”: the label has endured since then, even though the exact definition and distinction between group quarters and the household population have evolved from census to census. Most censuses since 1930 have used some numerical standard to define group quarters: the 1930 and 1940 censuses defined 12 unrelated people living census as it is conducted in the 50 U.S. states and the District of Columbia. The parallel set of census processes conducted in the Commonwealth of Puerto Rico raise unique challenges of their own, from demands on Spanish-language resources to differences in street and mail address formatting. 2 “Other,” in turn, “includes groups of laborers at work on farms and plantations, railroads, roads, etc.; groups of miners and lumbermen in camps, etc.; crews of boats and vessels; soldiers and sailors at military posts and stations and on naval vessels; and miscellaneous groups of persons lodging together but having no family relationship in common” (see U.S. Census Office, 1902:clviii–clix; Tables 95 and 96).

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census in the same unit as a group quarter, the 1950–1970 censuses used 5 or more unrelated people, and the 1980 and 1990 censuses used 10 or more unrelated people (Ruggles and Brower, 2003:75). However, the 1980 census also began the practice of declaring some types of living facilities—notably, college dormitories, as well as some hospitals and missions or flophouses—as “non-institutional group quarters” regardless of the number or relationship of the people within those units. This approach was used in the 2000 census: types of places were designated as group quarters in advance, without any numerical criterion in the definition of “group quarters.” Table 3-1 lists the 2000 census totals for the group quarters population, divided by group quarters type. The major groups in that listing—college students, patients in health care facilities, and persons in correctional facilities—are also significant cases where residential ambiguity and census error are potential problems. Before discussing issues specific to each group, we note five issues that are common across the groups. The first general issue is competing claims as to where facility population should be counted. The size of the population of group quarters facilities makes the geographic placement of the group in census tabulations a sensitive issue, politically and economically. Colleges and universities, prisons, and military bases (with on-base housing) can account for large shares of the overall population and job base of cities, towns, and counties, and—in the quest for allocated state and federal funds—areas that house facilities have a strong incentive to have the facilities counted in their population tallies. However, the various facility types are not the same in how they fit the specifications of funding programs: for example, by virtue of their ability to move around and seek employment outside the facility, the populations of colleges and military installations are arguably more service needy and factor heavily into local areas’ planning of almost all activities. By contrast, the inherently less mobile and more insular populations of prisons are less likely to play a role in, say, local transit funding and education decisions; however, the prison population may need to be covered by local fire and emergency response personnel, and so be accounted for in those allocations. The second broad issue involves gradations in the length of stay. Large group quarters segments differ greatly in the expected length of stay of their residents. College and boarding school students live in dormitories during the 8–10 months of the academic year, and the dormitories are largely vacant during the summer months. Health care facilities like hospitals and nursing homes can hold patients for stays ranging from hours (emergency room visits and admissions) to weeks (short-term physical rehabilitation stays at nursing facilities) to years (long-term nursing home assignment), and individual patients may cycle back and forth between home and hospital stays. Likewise, local jails may hold inmates for a matter of hours but may have to hold some for weeks or months (e.g., incarceration during a trial), and prison sentences

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Table 3-1 Group Quarters Population by Group Quarters Type, 2000 Census Group Quarters Type Population Institutionalized Population 4,059,039 Correctional institutions 1,976,019 Federal prisons anddetention centers 154,391 Halfway houses 30,334 Local jails and other confinement facilities (including police lockups) 623,815 Military disciplinary barracks 20,511 State prisons 1,077,672 Other types of correctional institutions 69,296 Nursing homes 1,720,500 Hospitals/wards, hospices, and schools for the handicapped 234,241 Hospitals/wards and hospices for chronically ill 40,022 Hospices or homes for chronically ill 7,682 Military hospitals or wards for chronically ill 1,944 Other hospitals or wards for chronically ill 30,396 Hospitals or wards for drug/alcohol abuse 19,029 Mental (psychiatric) hospitals or wards 79,106 Schools, hospitals, orwards for the mentally retarded 41,857 Schools, hospitals, orwards for the physically handicapped 20,765 Institutions for the deaf 2,325 Institutions for the blind 1,885 Orthopedic wards and institutions for the physically handicapped 16,555 Wards in general hospitals for patients who have no usual home elsewhere 23,538 Wards in military hospitals for patients who have no usual home elsewhere 9,924 Juvenile institutions 128,279 Long-term care 71,496 Homes forabused, dependent, and neglected children 19,290 Residential treatment centers for emotionally disturbed children 12,725 Training schools for juvenile delinquents 39,481

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Short-term care,detention or diagnostic centers for delinquent children 29,437 Type of juvenile institution unknown 27,346 Noninstitutionalized Population 3,719,594 College dormitories (includes college quarters off campus) 2,064,128 Military quarters 355,155 On base 310,791 Barracks,unaccompanied personnel housing (enlisted/officer), and similar group living quarters for military personnel 281,416 Transient quarters for temporary residents 29,375 Military ships 44,364 Group homes 454,055 Homes or half way houses fordrug/alcohol abuse 94,243 Homes for the mentally ill 63,912 Homes for the mentally retarded 148,043 Homes for the physically handicapped 16,119 Other group homes 131,738 Religious group quarters 78,102 Dormitories 97,697 Agriculture workers’ dormitories on farms 47,664 Job Corps and vocational training facilities 32,143 Other workers’ dormitories 17,890 Crews of maritime vessels 4,523 Other nonhousehold living situations 95,565 Other noninstitutional group quarters 570,369 Total 7,778,633 SOURCE: 2000 Census Summary File 1, tabulated through http://factfinder.census.gov.

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census can range from months to life. This gradation in the possible length of stay at facilities can create residence conundrums for census respondents and their families—if a household member is expected to return home from a hospital stay in a few days/weeks/months, should they be counted at their home or at the nursing facility? And, as we discuss in more detail later, should prisoners be counted at a “home of record” rather than a prison location and, if so, should remaining time left in a prisoner’s sentence (e.g., 3 months or natural lifetime) be a factor in making that determination? The third general issue concerns the continuum of attachment to a facility. Consistent with the variation in lengths of stay at facilities, the group quarters population spans a continuum of levels of attachment to the facility. The population blends groups of people who are unequivocally in the facility as a usual residence (they have no other residence outside the facility) with those who have strong ties to a usual residence elsewhere. In a different vein, the fourth general issue is the presence of a “gatekeeper” or other barriers to direct enumeration. Residents of large group quarters also vary greatly in the degree to which direct enumeration—distribution of questionnaires to be filled out directly by residents—is either feasible or permissible. Some college campuses, eager to get as complete a count as possible, may encourage such direct contact. Others—citing resource constraints or the magnitude of the job—may insist that enumeration not be done directly and that university records be used. Hospital and group home residents may be physically incapable of completing forms on their own, requiring special enumeration methods or reliance on administrative records. And, finally, prison administrators may cite safety concerns in prohibiting any direct contact with their inmate populations. Lastly, there are complicated address listing issues that affect all groups. As examined in greater detail in Chapter 7, it can be difficult to maintain an accurate and up-to-date roster of group quarters units, and—if the lists of group quarters and “regular” housing units are kept separate—duplication or omission can occur if the two lists are not reconciled. An analysis of the 2000 census public-use microdata sample (Ruggles, 2003:481) concluded that the Census Bureau’s decision to scrap a rule that would define a group quarters on the basis of a set number of unrelated individuals living together was largely inconsequential because only 42 households in the sample would have been declared group quarters using such a rule (using the 1990 cutoff). However, many of these stray households appear to be clear cases in which buildings were misclassified and errors were made in listing group quarters facilities: these “large households” consisted entirely of college students, farm laborers, or construction laborers.3 3 Among these is “one unit with 53 female housemates aged 18 to 24, all of whom were attending a public university” (Ruggles, 2003:481).

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Large institutional group quarters may also have housing units that should rightly be contacted by usual census methods like mailout/mailback: college dormitories may include rooms for resident assistants who maintain a permanent residence there; boarding schools may likewise have permanent onsite resident faculty and staff quarters; and contemporary nursing homes may blend apartment-style assisted living units with more institutional, hospital-like nursing beds in the same facility. It should also be noted that the questionnaire used to enumerate group quarters residents differs somewhat from the general household questionnaire. Group quarters resident information was to be collected on Individual Census Reports (ICRs), or customized versions for the military and shipboard populations; see Box 3-1. In addition to the ICRs—one per person—enumerators of group quarters facilities were expected to complete a general roster of residents. The general roster and every ICR were supposed to be marked with the same group quarters identification number (the space for which is shown on page 2 of the ICR in Box 3-1). 3–B STUDENTS 3–B.1 Colleges and Universities As Lowry (1987:15) observed, “[college] students are individually transient—each year many leave never to return—but as a class they form a permanent component of the local population that must be served in various ways by local government.” As a large group of people with strong ties to two locations, college students have understandably been prominent in census residence issues from the outset. College students were the focus of one of the first formal census residence rules (see Section 2–B.1), and were the focus of one of the most significant changes to the rules in recent decades. Enumerator instructions for 1850 directed that college students be “enumerated only as the members of the family in which they usually boarded and lodged on the 1st day of June.” Instructions for the 1870 census clarified that “children and youth absent for purposes of education on the 1st of June, and having their home in a family where the school or college is situated, will be enumerated at the latter place.” Whether the student was to be counted at home or school would depend on the course of study and whether students returned home at the end of the academic calendar. The 1880–1900 census instructions signaled a clearer inclination for counting students at their parental homes, although that specific language was not explicitly used. A college student was identified by the 1880 instructions as a person for whom the enumerator “can, by one or two well-directed inquiries, ascertain whether [there] is any other place of abode within another district at which he is likely to be reported.” (Similar text is included in the subsequent census instructions.) The

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Box 3-1 Individual, Military, and Shipboard Census Reports The census questionnaire used to gather data from people in group quarters facilities is similar to, but not exactly the same as, the questionnaire mailed to the general household population. In the 2000 census, four separate forms were used for group quarters. The Individual Census Report (ICR), short form, was a two-page questionnaire, shown here. A six-page ICR, long form, was administered to those people included in the long-form sample; it presented the ICR short-form questions (though in less space), along with the social and demographic long-form questions. The Military Census Report (MCR) and Shipboard Census Report (SCR) were based on the ICR long form, for enumeration of military personnel and maritime vessel crew, respectively. The MCR and SCR forms begin by asking for context-specific information—installation/base name and unit name for the military, name and operator of ship for the shipboard population—after requesting the respondent’s name in question 1.

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census A particular feature of note in the ICR is that it asks for information on a “usual home elsewhere” (“What is the address of the place where you live or stay MOST OF THE TIME?”), even if the respondent lived in a group quarters type for which the Census Bureau did not allow “usual home elsewhere” information to be used. The MCR adds a filtering question and is more specific in defining a time period: “Is the place where you stay at least 4 nights a week a barracks, BOQ [bachelor officers’ quarters], disciplinary barracks, hospital, etc., or a house, apartment, or mobile home?” If the answer is “house, apartment, or mobile home,” the person is asked for the address; if not, he or she is routed to a question asking for a telephone number. The SCR asks, “Do you have a residence (house, apartment, or mobile home) where you usually stay when off duty?” If yes, the address is requested.

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Facilities may be of many different types, ranging from “boot camp” and “boys ranch” (or “girls ranch”) settings to highly secure, fenced, and locked facilities (Greenwood, 1995). Generally, “national accreditation standards for juvenile facilities express a preference for relying on staff, rather than on hardware, to provide security. The guiding principle is to house juvenile offenders in the ‘least restrictive placement alternative’ ” (Sickmund, 2004:16). As with other correctional facilities, recent decades have seen the emergence and expansion of privately owned and operated facilities for the treatment and housing of juveniles (Greenwood, 1995). Time spent in juvenile facilities is difficult to quantify because—much like jails—juvenile facilities play different roles in sentencing and adjudication. Some juveniles may be committed to facilities—that is, they are sentenced to the facility for some fixed term by a juvenile or criminal court; others may be detained at a juvenile facility while waiting on further action (e.g., pending adjudication or waiting on sentencing); still others may be diverted to serve a period at a facility as part of a legal agreement in order to avoid adjudication (in the hopes of deterring the juvenile from further delinquent acts). A National Research Council and Institute of Medicine (2001:156) report finds that no national data on sentence length, time in confinement, and time spent in parole are published. Parent et al. (1994) estimate that juveniles who were released from correctional facilities in 1990 had average stays ranging from 15 days to 32 weeks, depending on the type of facility. They note, however, that their measure gives greater weight to facilities with high turnover, thus understating the lengths of stay of juveniles confined on any given day. The Office of Juvenile Justice and Delinquency Prevention (OJJDP) operates several survey programs to assess the size and living conditions of the incarcerated juvenile population. Since 1997, OJJDP has contracted with the Census Bureau to conduct the biennial Census of Juveniles in Residential Placement (CJRP).28 The CJRP is a mail survey sent to all known juvenile residential custody facilities in the United States,29 excluding facilities that are exclusively for drug or mental health treatment (which may house some juveniles) or for abused or neglected children. The CJRP also, by design, does not cover juveniles that may be housed in predominantly adult facilities. In 2000, OJJDP initiated the separate Juvenile Residential Facility Census (JRFC). Focused on the capacity and quality of the facilities, the JRFC shares the same limitations (excluded facilities) as the CJRP. The 2000 JRFC found that 110,284 juvenile (under 21 years old) offenders 28 This program is the successor to the Census of Public and Private Juvenile Detention, Correctional, and Shelter Facilities—better known as the Children in Custody Census—that was first conducted in the 1970s. 29 “Residential” means that they are capable of holding juvenile offenders overnight; some facilities to which the CJRP instrument was mailed turned out to not have this capacity, and so were identified as out-of-scope.

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census lived in 3,061 facilities; about 70 percent of them were concentrated in the 40 percent of facilities that are government owned (Sickmund, 2002). The 1999 administration of the CJRP (Sickmund, 2004) arrived at a higher number of juvenile offenders—134,011, itself an increase from 125,805 in 1997. About 74 percent of the juveniles encountered in the CJRP have been committed while about 25 percent are detained; only 0–1 percent are under diversion agreements. The total population of juvenile offenders in residential placement is disproportionately composed of racial and ethnic minorities, with blacks (39 percent) and Hispanics (18 percent) being large shares (38 percent are reported as white). 3–E CHILDREN IN FOSTER CARE Foster children are those who, for a variety of reasons (including abusive or criminal activity by parents, abandonment, or death of parents) are in the care of a state government agency. These children are a unique and challenging population for censustaking purposes because of the variety of their living arrangements. Foster care may be provided in state-funded facilities or through placement of children in the homes of individual families; in the latter case, the state government agency reciprocates by providing regular payments or the foster child’s care.30 The Pew Commission on Children in Foster Care found that 523,062 children were in some form of foster care arrangement in the United States and Puerto Rico in fiscal 2003; California’s 97,261 cases accounted for the largest share of these (18.6 percent), with New York state’s 37,067 cases (7.1 percent of the total) ranked second.31 Demographically, Pérez et al. (2003) found that, for fiscal 2000, approximately 30 percent of foster children are between ages 11 and 15; roughly 25 percent each are in the 1–5 and 6–10 age groups; and about 4 percent are 1 year old or younger. Only 2 percent are aged 19 and older; typically, foster children “age out” of the system on reaching legal adulthood at 18. Pérez et al. (2003) also found that the foster child population is disproportionately African American (41 percent); American Indian and Alaska Native children are also somewhat overrepresented in the foster child population relative to their prevalence in the general population. 30 Schwede (2003:xiii) summarizes ethnographic observation from the 2000 census that suggests that—for some non-English-speaking households—foster children may be miscounted due to terminological confusion. Specifically, the ethnographers found that the Spanish-language questionnaire used the phrase hijo de crianza for “foster child”; however, the Spanish phrase can be interpreted as a child one is raising for a friend or relative—losing the connotation of placement and supervision by the government. Conversely, the Korean-language questionnaire used a term for foster child that translates literally as “child under trusteeship,” which confused respondents. 31 See http://pewfostercare.org/research/docs/Data102705a.pdf [8/1/06].

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Individual states vary in the number of children in foster care and in the growth trends of foster care caseloads. The number of “children in substitute care” in Illinois peaked at 51,331 in 1997, according to the state’s Department of Children and Family Services, declining to 17,735 by July 2005.32 The state of New York has experienced a significant decline in the number of foster cases in recent years; the number of children in foster care decreased smoothly from 42,921 in 2000 to 29,680 in 2004 (New York State Office of Children and Family Services, 2004:11). Texas has seen its foster care caseload grow from 18,626 to 26,133 between fiscal 1999 and 2003 (Strayhorn, 2004:3). For the nation as a whole, Pérez et al. (2003) found that the majority of foster children, 72 percent, are housed in individual families’ homes: 47 percent are in foster homes in which they have no relation to the head of the household, and 25 percent live with a relative. Another 4 percent of foster children in fiscal 2000 were living with families while adoption processes were pending, and 3 percent were housed with families on a trial visit basis. By comparison, a significant portion of foster children lived in institutional settings, either designated institutions (10 percent) or group homes (8 percent). The exact mix of in-family placement versus more institutional settings varies by state. New York state reported 74.6 percent of its 2004 foster cases as living in individual homes and 23.9 percent in “congregate care,” institutional settings (New York State Office of Children and Family Services, 2004:16). Figures from Texas suggest only about 12.1 percent of foster children living in institutional “residential treatment facilities” in fiscal 2003, but group homes are combined with general foster homes in the tabulations (Strayhorn, 2004:7). January 2006 statistical reports by the Illinois Department of Children and Family Services showed 11.4 percent of the state’s foster children living in institutional/group home settings.33 The ways in which children can transition out of foster care—including “emancipation” (reaching legal adulthood), adoption, or reunion with parents or caretakers—are widely varied, as are the circumstances by which children enter foster care. For census purposes, foster children can be a tricky population group to count accurately because of transitions within the foster care system. Although researchers suggest that the notion of “foster care drift”—“children trapped in foster care and constantly moving from one placement to the other” (Usher et al., 1999:22)—can be overstated, national-level records do suggest that foster placements can be transitory. On average, foster children experience three different placement sites during their state supervision (Pew Commission on Children in Foster Care, 2004). The Texas state comptroller (Strayhorn, 2004:7) found that children in the state’s foster care 32 See http://www.state.il.us/dcfs/foster/index.shtml [8/1/06]. 33 The department’s monthly tallies are regularly updated at http://www.state.il.us/DCFS/docs/execstat.pdf [8/1/06].

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census system in fiscal 2003 had averaged 4 different placements; “those remaining in foster care for a decade or more could expect to be moved about once a year.”34 Because of the movement of children in foster care placements, the ties between children and households (when foster children are placed in family homes) could be viewed as tenuous or transitory; this could affect the householder’s notion of whether the child is a “usual resident” there. If foster caregivers know that a child’s placement in their home is likely to be short term (as opposed to a full adoption), they might believe they should not include the child in a household roster. Children in “congregate foster care”—group-based settings—may present difficulties for enumeration because of the variety of housing arrangements in which they may be found. These congregate living arrangements run the gamut from “institutional” approaches such as dedicated orphanages and youth centers to group home situations that are indistinguishable from other housing stock. Indeed, a goal of some state-sponsored homes for these youths is to make housing as home-like and nonclinical as possible (e.g., providing access to regular schools and treatment at medical clinics outside the living facility). Address lists generated without full participation of child welfare agencies could lead to operational problems and data oddities that may challenge editing and imputation routines in the census. For instance, the census return from what seems to be a typical suburban house may appear to be a grouping of 6–10 minor children with no adult as a permanent resident (even though staff would be always present at the home, they would not likely reside there full time). Given the age of the children, direct delivery and administration of questionnaires to the children in the facilities would likely not be tenable. As with other group quarters, the accuracy of enumeration in these settings would depend on the willingness and ability of facility staff to provide responses or on the quality of facility or administrative records. 3–F MILITARY AND SEABORNE PERSONNEL The 2000 census recorded about 282,000 residents of military barracks and other on-base residences, another 30,000 at short-term or transient housing on military bases, and 44,000 men and women on military ships (see Table 3-1). These counts reflect only a part of the broader military population (and their dependents) serving in the United States, in its waters, and overseas. The military population is challenging to count because of the diversity of service locations and the changing nature of assignments. As Hollmann (1987:279) notes, the “usual residence” principle that “is readily applicable to the vast majority of the population” is less suited to the military because they “are by 34 As extreme cases, 12 Texas foster children in fiscal 2000 were found to have gone through at least 40 different foster placements.

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census their essential nature more mobile, less uniformly distributed and often occupying less typical forms of shelter. At the same time, all but the youngest, most recently inducted and lowest ranking of service people are also members of more typical households, from which they may often be absent.” The military is also uniquely prominent in the context of census residence rules because of the weight that has been put on including members of the military in apportionment totals; along with federal government employees on overseas assignment, military personnel stationed overseas have been included in census apportionment totals in recent censuses (they are only assigned to a state, though, and are not included in redistricting totals). The overseas dimensions of counting the military are a major part of debates on census residence issues in recent decades, and we describe them in greater detail in Appendix C. In this section we also discuss the special problem of counting people working and stationed on boats, and so not directly tied to any location on land. The treatment of people on ships—whether military, privately owned, or the Merchant Marine fleet that transports goods during peace time but serves as an auxiliary to the navy during war—has been a long-standing concern in the census and practice has varied over the years. 3–F.1 Personnel Stationed at Domestic Bases or Living in Nearby Housing The 1880, 1890, and 1900 census enumerator instructions were the first—and among the relatively few—sets of census rules to explicitly mention the task of counting people at military stations in the United States (although the direction on how exactly to count them was not clear). The 1880 rule stipulated that “all soldiers of the United States Army, and civilian employees, and other residents at posts or on military reservations will be enumerated in the district in which they reside, equally with other elements of the population.” The 1890 rule modified the rule to read that the personnel “will be enumerated in the same manner as has been provided for institutions.” That rule seemed to tip the decision toward counting military personnel assigned to domestic bases at the base, but the 1900 census instructions reversed that decision: If a soldier, sailor, or marine (officer or enlisted man), or civilian employee in the service of the United States at a station at home or abroad, is a member of a family living in your district, he should be enumerated as a member of that family, even though he may be absent on duty at the time of enumeration. Specific directions on counting the military population are then absent from enumerator instructions until 1950; in the interim, the general process of counting military stationed at land bases in the United States (by means of specially appointed enumerators) seems to have been the norm.

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Beginning in 1950, census practice shifted toward an approach that put emphasis on separate enumeration, using deputized personnel on military bases and vessels to conduct the count, and, ultimately, toward uses of service records. A dual approach took shape in the 1950–1980 censuses: military personnel posted to a domestic base would receive (and be required to return) an individual census report, but off-base households (where the same service members could live with their families) would be covered by the normal household enumeration. The 1950 enumerator instructions for completing the basic population schedule indicated that “soldiers, sailors, marines, and airmen” were only to be enumerated as residents on the standard form if they “are stationed in your vicinity and [live] and sleep off post in your [enumeration district]” (see Box 5-5); instead, military personnel stationed elsewhere were identified as a “special class” to be enumerated by alternate procedures. The question of overlap—of the service members being counted in both places—was left ambiguous; census materials do not suggest any provision for reconciling the individual service reports with household returns from nearby areas. The Bureau’s procedural history of the 1970 census, for instance, says only that the “Report for Military and Maritime Personnel” questionnaires collected from on-base service members “were included in the preliminary population counts for the places where they were stationed” (U.S. Bureau of the Census, 1970:8–25); if and when in processing these records were compared with off-base reports is not specified. The 1990 and 2000 censuses modified these long-standing procedures by permitting personnel living on base to claim a “usual home elsewhere” (see rules 4 and 13 in the 2000 rules, Appendix A). In practice, what resulted in the 1990 and 2000 approach was a multiple-form process. All service personnel stationed to bases were required to fill out individual MCRs, distributed and collected by base personnel. At the same time, housing units located off base received questionnaires in the usual manner (e.g., mostly by mail). Both forms were expected to be returned. If a service member reported an off-base household as his or her usual residence, the MCR record could be linked to the census return from the off-base household to ensure that no duplication had taken place.35 This two-form approach suggests one possible source of error in collecting residence information from military personnel and dependents at or near 35 The MCRs used in the 1980 and 1990 censuses are unusually direct and specific in asking the question about another residence. Question 2b of the 1990 MCR form asks, “What is the address where you usually stay at least 4 nights a week?”—a majority of days in a week as the criterion for usual residence, and not a more generic “most of the time” query. The questionnaire includes “building or barracks number” as a field, so the service member could report on-site housing as applicable; question 2c confirms whether or not the address is on a military base or not. Hollmann (1987) indicates similar text on the 1980 MCR. The 1990 SCR administered to sailors asked the more general question, “Do you have a residence (house, apartment) where you usually stay when off duty?”

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census domestic bases—confusion in the data collection process. Census materials tried to make clear that completing two forms (the MCR by the service member and, if applicable, the normal census questionnaire at the off-base home) was normal and expected. Instructions on the 1990 MCR advised that: Military personnel living away from this installation, but within the census area, also will receive a census form at home. To ensure that such personnel are assigned to the correct jurisdiction, it is important that YOU MAKE SURE YOU ARE INCLUDED ON BOTH FORMS—the report and the census form sent to your home. Likewise, press releases emphasized the completion of both forms.36 Still, these efforts may have left some military families confused and claiming that—one form having been filled out—they saw no need to complete another. For instance, public affairs officials at Fort Detrick, Maryland, reported particular confusion following reports in a base newsletter that “the census at Fort Detrick was 100 percent completed.” That report referred only to the distribution and collection of MCRs; regular census enumeration and nonresponse follow-up were still in process off base (Duble, 2000). In addition to possible confusion over two forms, counting personnel at or near domestic military bases is also complicated because bases include a variety of housing arrangements and styles that defy easy categorization. Barracks and dormitories: A large component of military housing is functionally equivalent to other types of group quarters, with groups of service men and women sharing common facilities, as well as bedroom-bathroom combinations. These quarters range from the highly institutional barracks in boot camps and basic training facilities to the emerging standards of “four plus one” (four people with separate sleeping quarters but shared commons) or “one plus one” (two people with connected, shared living area but possibly separate bedroom and bathroom) configurations. Officer housing: The military services may offer housing options to officers that lack the “group quarters” aspects of barracks or dormitories; these can be private apartment units that are functionally equivalent to civilian apartment or condominium complexes or stand-alone houses that are likewise indistinguishable from conventional housing stock. Depending on where exactly these units are located (within base confines or on the periphery), there may be ambiguity as to whether the housing units are included on the Master Address File and, hence, 36 See, e.g., an article from a Fort McCoy, Wisconsin, newsletter (http://www.mccoy.army. mil/vtriad_online/03102000/Military Census.htm [8/1/06]) for a base census project officer’s explanation of the two-form approach.

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census whether they may be picked up by normal census operations in addition to military count operations. On-base and off-base housing for families and dependent children: Depending on rank and marital status and dependent children, military personnel may live in on-base housing or receive a monetary basic allowance for housing to be applied to off-base housing. As discussed earlier, off-base housing raises the potential of confusion of overlapping operations: if family members live in quarters within the physical confines of the base, they could be missed by the normal household census operations (questionnaire mailout) if address listing is not done carefully. Public partnerships: In response to long-standing concerns on the quality and quantity of affordable military housing, the Military Housing Privatization Initiative was begun in 1996 in order to facilitate private-sector financing, construction, and maintenance of housing for military servicemembers.37 Under the initiative, the military services are authorized to enter into agreements with private developers, who may own and operate military family housing under 50-year lease agreements. This private ownership may affect the information that base personnel have on the occupancy and exact structure of housing units during address listing operations. Like their counterparts serving overseas, domestic-based military personnel have some “home of record” on file in Department of Defense records. Periodically, questions have been raised as to whether military personnel should be counted using administrative records, relying on that “home of record” information. That proposition was raised at the same 1999 congressional hear-ing that discussed pending legislation on the counting of prisoners. Speaking against the idea, census director Kenneth Prewitt commented that “if we had to match completed census forms for armed forces members to Defense Department administrative records, that would require a massive, costly, and time-consuming operation that we could not undertake without putting the census at risk.” Further, Prewitt questioned the nature and consistency of “home of record” coding; in some cases, it could reflect place of birth and not more recent residence information (U.S. House of Representatives, Committee on Government Reform, 2000). One state—Kansas—disagrees with the Bureau’s default placement of military personnel as well as college students and fields its own survey in order to adjust census totals so that military personnel stationed in Kansas but whose home of record is elsewhere are excluded from the counts used to redraw legislative districts (we discuss this further in Section 7–E). 37 See http://www.acq.osd.mil/housing/mhpi.htm [8/1/06].

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census 3–F.2 Shipboard Personnel The 1850 census enumerator instructions—the first to begin to include detailed residence rules—sought to establish a clear policy for shipboard personnel: “errors necessarily occurred in the last census in enumerating those employed in navigation, because no uniform rule was adopted for the whole United States.” Specifically, the rules directed that people should be counted at their homes on land, if possible, and at the ship’s “homeport” otherwise: The sailors and hands of a revenue cutter which belongs to a particular port should be enumerated as of such port. A similar rule will apply to those employed in the navigation of the lakes, rivers, and canals. All are to be taken at their homes or usual places of abode, whether present or absent; and if any live on board of vessels or boats who are not so enumerated, they are to be taken as of the place where the vessel or boat is owned, licensed, or registered. Instruction language in 1860 repeated this basic rule, adding the specific instruction that “persons on board any description of ships or vessels accidentally or temporarily in port … are not to be enumerated in your district,” but rather at their land homes or at “the place where they have been engaged, shipped, or hired.” The presumption that sailors were to be counted at their land homes was sharpened further in 1870, with the instruction that “seafaring men are to be reported at their land homes, no matter how long they may have been absent, if they are supposed to be still alive.” This latter wording was directly repeated in the instructions for several subsequent censuses, the 1890 instructions being the first to specifically acknowledge that military sailors and marines, as well as navy yard personnel, would be counted by special enumerators.38 As it did for counting the domestic military population, the 1950 census marked the arrival of a revised approach for counting naval personnel and the seaborne population generally. The enumerator instructions pointedly told household enumerators not to enumerate “officers and crews of ships” when “they are reported as absent members by their families.”39 Instead, the 1950 census practice signaled a stronger partnership with the Department of Defense and related agencies (U.S. Bureau of the Census, 1955:12): 38 As mentioned in the previous section, enumerator instructions for 1900 indicated that sailors “at a station at home or abroad”—presumably including assignment to a ship—be enumerated at their land home. The 1900 rules added a general provision that “all persons who claim to be residents of the United States” who can be found “in vessels, steamboats, and house boats at wharves and piers or river landing[s]” on the morning of Census Day be enumerated “by the enumerators of the districts contiguous to the water front.” Whether those residents could be attributed to land home is not specified. 39 However, “men employed on vessels on the inland waters (rivers, canals, etc.) of the United States, other than the Great Lakes,” were to be “report[ed] in the regular way” (that is, on the household form).

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Naval, merchant marine, and other vessels were enumerated with the cooperation of the Navy Department, Coast Guard, Coast and Geodetic Survey, and the Merchant Marine. Mailing registers were established listing the vessels and the approximate number of crew members aboard each vessel. Packages of the special enumeration forms for crews of vessels (P4) with letters of instructions were mailed directly to the captains of all vessels in the Navy and Coast Guard and to other government-operated vessels. Those for the merchant marine were grouped and mailed to the companies operating the ships and then reshipped by them to each vessel. Just as the revised military counting procedure brought with it some ambiguity (in the form of uncertainty as to how military person records would be unduplicated), the revised practice for enumerating shipboard personnel left it uncertain as to whether ship crews would be counted at a port or at individual homes. The procedural history of the 1970 census (U.S. Bureau of the Census, 1970) indicates that census forms from military ships and other American flag vessels were routed to the Bureau’s processing center and sorted (by reported location on a control card) into “U.S.” and “At Sea” groups. The returns from each “U.S.” vessel were then “coded to an enumeration district … and a census block assigned to the pier area of the port where the vessel happened to be or, in the case of a coastwise vessel, its home port.” That is, the physical location of the ship around Census Day appeared to take precedence over the residences reported on individual reports. Accordingly, the physical location of these ships—each of which could house hundreds of service men and women—became a point of concern in communities with seaports. Hollmann (1987:280) recounts that “during the 1960s there was an abundance of stories, most of them no doubt apocryphal, of the efforts of city officials to get as many of the home-ported vessels and as many transients as possible into port to tie up and be counted on Census Day. A major storm which would make life difficult for door-to-door enumerators was supposedly a heaven-sent gift to the local finance officer of chamber of commerce of the port city.” The 1980 census revamped the process for enumerating personnel on ships. Naval ships in the 6th or 7th Fleets were automatically designated part of the overseas population, because deployment to those Mediterranean-and Pacific-based fleets is generally long term. Other naval vessels were “attributed to the municipality that the Department of the Navy designated as its homeport.” If the ship had fewer than 1,000 personnel, the crews were counted at the physical homeport location; personnel on ships with crews of 1,000 or more with reported usual residences within 50 miles of the home port were allowed to be counted at those residences, and at the home port location otherwise. Crews of merchant vessels were not allowed to report usual residences on land: they were counted at the U.S. port at which they were berthed on Census Day, at their port of destination if their vessel was not berthed but

OCR for page 59
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census was still in U.S. territorial waters, and at their home port if they were still in U.S. waters but headed overseas. The 1990 and 2000 censuses dropped the 1,000-personnel and 50-mile provisos of the rule for counting naval vessels. They replaced the 50-mile radius with the ambiguous “nearby,” and permitted reports of household locations on land and counting military shipboard personnel there if possible. The two censuses also permitted crews of merchant vessels to report off-ship residences; those who did not were counted at the berthing port (if berthed on Census Day), the U.S. port of departure (if at sea), the U.S. port of destination (if headed back to the United States from a foreign part), or as part of the overseas population otherwise. Citro (2000:205) relates: For the 2000 census, the Census Bureau worked with the U.S. Department of Defense and U.S. Coast Guard to identify both housing units and group quarters on military installations and ships assigned to a home port in the United States. Questionnaires were mailed to housing units on military installations; other methods, such as visiting installations and ships to enumerate people at their work stations, were used to count the military population in group quarters. The U.S. Maritime Administration assisted in identifying and arranging to mail questionnaires to other maritime vessels in operation around Census Day; these vessels included “factory trawlers, floating processors, tuna boats and National Oceanic and Atmospheric Administration vessels” (U.S. Census Bureau, 1999:1). Discussion of the military seaborne counting procedures at the 1986 CO-PAFS conference on residence rules suggested general agreement that the use of home port information rather than current location was generally good. However, accuracy depends on how home ports are recorded and whether they are shifted, for instance, when a ship is detailed to a navy yard for months for extensive repairs or upgrades. The discussion also pointed out remaining ambiguities in the interpretation of what a home port location means. With specific reference to San Diego, Hollmann (1987) describes how different interpretations where a ship is located (e.g., physically tied to a pier within city limits, even though the land below “mean low water” underneath the ship’s hull might belong to another jurisdiction) could credit ships whose home port is San Diego to the cities of San Diego, National City, or Coronado. That specific case was argued over most of the 1980s, culminating in a California court ruling that credited the ships to Coronado even though the waters in which they float may be in San Diego’s jurisdiction.