–7–
Nonhousehold Enumeration

THOUGH RELATIVELY SMALL in magnitude—just below 3 percent of the total population in both the 1990 and 2000 censuses—the population living in group quarters (nonhousehold) settings includes many of the most prevalent living situations that complicate interpretation of “usual residence,” among them college students, hospital and nursing home patients, and people in correctional facilities. In the difficult task of censustaking, group quarters data collection is surely one of the most challenging parts. Group quarters are hard to define, and collection of the data is highly likely to face both physical and legal impediments. Collection of data from individual residents may be flatly denied by prison wardens, college administrators, or facility administrators, citing safety or logistical concerns; the quality of data from these facilities then hinges on the quality of administrative or facility records. Legal impediments include concern over confidentiality, particularly in such settings as medical facilities or work camps where sensitivity to medical records privacy and citizenship status may loom large. We have discussed conceptual aspects of residence data collection from the nonhousehold population in Chapter 3; in this chapter, we describe operational improvements to nonhousehold enumeration and the research needed to attain them.

It is important to get group quarters enumeration right in the decennial census, not only because of the consequent problems of omission and duplication, but also because the decennial census is currently the only comprehensive data source on this segment of the population. As of this writing, most group quarters types have been included in the long-form replacement American Community Survey (ACS); however, it is unclear how accurate ACS-based data on group quarters will be. As we will discuss below, obtaining



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 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census –7– Nonhousehold Enumeration THOUGH RELATIVELY SMALL in magnitude—just below 3 percent of the total population in both the 1990 and 2000 censuses—the population living in group quarters (nonhousehold) settings includes many of the most prevalent living situations that complicate interpretation of “usual residence,” among them college students, hospital and nursing home patients, and people in correctional facilities. In the difficult task of censustaking, group quarters data collection is surely one of the most challenging parts. Group quarters are hard to define, and collection of the data is highly likely to face both physical and legal impediments. Collection of data from individual residents may be flatly denied by prison wardens, college administrators, or facility administrators, citing safety or logistical concerns; the quality of data from these facilities then hinges on the quality of administrative or facility records. Legal impediments include concern over confidentiality, particularly in such settings as medical facilities or work camps where sensitivity to medical records privacy and citizenship status may loom large. We have discussed conceptual aspects of residence data collection from the nonhousehold population in Chapter 3; in this chapter, we describe operational improvements to nonhousehold enumeration and the research needed to attain them. It is important to get group quarters enumeration right in the decennial census, not only because of the consequent problems of omission and duplication, but also because the decennial census is currently the only comprehensive data source on this segment of the population. As of this writing, most group quarters types have been included in the long-form replacement American Community Survey (ACS); however, it is unclear how accurate ACS-based data on group quarters will be. As we will discuss below, obtaining

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census complete long-form data for group quarters residents in the census has been difficult in the past, and it remains to be seen how successful a long-form-only collection will be for group quarters facilities. 7–A IMPLEMENTATION PROBLEMS IN THE 2000 CENSUS Our predecessor panels on the 2000 census and the 2010 census plan (National Research Council, 2004c:152–155; 2004b:150–152) reviewed the many implementation problems that plagued group quarters data collection in the 2000 census. In this section, we draw from and expand on their analyses in our own account of the group quarters process in 2000 (see also Jonas, 2003; Abramson, 2003). Failure to Reconcile Group Quarters Roster with MAF The Census Bureau’s inventory of group quarters was developed independently from the Master Address File (MAF). The two were developed by separate internal divisions of the Bureau, and were not cross-checked with each other prior to November 1999. An adjunct of the Local Update of Census Addresses Program was instituted to allow local and tribal governments to review group quarters listings, but the effectiveness of that program was severely compromised by an 18-month-late start. Since the program began just 4 months prior to Census Day, participation was low given competing demands on local resources for census outreach and other activities. The failure to adequately reconcile the group quarters roster with the MAF and with local authorities hurt the Bureau’s ability to use an accurate, nonoverlapping, and comprehensive address list as the base of the 2000 census. Poor Handling of Geographic Location The group quarters roster was also not well reconciled with the Census Bureau’s Topologically Integrated Geographic Encoding and Referencing geographic database. For unknown reasons, a number of group quarters—including very long-established prisons and college dormitories—were given inaccurate geographic location codes. These inexplicable errors led to sizable local shifts in population, as group quarters facilities were inadvertently assigned to neighboring cities and counties. These group quarters discrepancies account for a large fraction of the population counts that local jurisdictions challenged in the Census Bureau’s Count Question Resolution Program.1 Of these discrepancies, the highest-profile case arose when the Census Bureau acknowledged that 2,696 students at the University of North Car- 1 Though the Count Question Resolution Program could result in the Census Bureau issuing an errata statement and a certificate of revised population, a condition of the program was that the revised counts could not be used for apportionment or redistricting.

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census olina at Chapel Hill—representing 1,583 dormitory rooms in 26 separate buildings—had been double-counted. These dormitory rooms were included in the U.S. Postal Service’s Delivery Sequence File (a primary source of updates to the MAF) and subsequent operations provided conflicting information on whether the rooms were valid housing units, but not enough negative information for them to be deleted.2 The admission was sensitive because North Carolina edged Utah (by 857 residents) for the 435th and final seat in the U.S. House of Representatives; Utah had already conducted two lawsuits challenging the census count.3 Though the North Carolina case was the highest profile, group quarters also accounted for other sizable discrepancies (U.S. Government Accountability Office, 2005), including: Dormitories at Morehead State University were incorrectly placed outside the city limits of Morehead, Kentucky, in the original 2000 census counts; the city’s population was revised from 5,914 to 7,593. Similarly, misplacement of dormitories at the University of Wisconsin-Whitewater led to a population shift of 1,699 between Jefferson and Walworth Counties. The Crossroads Correctional Center in Cameron, Missouri—a 1,500-capacity prison completed in March 1997—was omitted from the city’s population (as well as that of DeKalb County); correcting the error raised the city’s total from 8,312 to 9,788. The Oxford Federal Correctional Institution is physically located in Grand Marsh, Wisconsin, in New Chester town of Adams County. However, the count at this prison was assigned to the town of Packwaukee in Marquette County; Packwaukee and New Chester towns do not have a common boundary. In a rare case of a complaint by local officials of an overcount in their area, officials in the town of Fredonia, Wisconsin, argued that the census count of 886 in one census block in the northeastern section of the town was not credible. The block in question was home to the 150-year-old dairy farm owned by the town chairman, a 6–7 house subdivision, a restaurant, and a campground—perhaps close to 52 people, as was the case in the 1990 census, but nowhere near 886. Ultimately, the Count Question Resolution Program certified the block’s true population as 2 The U.S. Government Accountability Office (2005:34) notes that—as part of the Count Question Resolution process in the North Carolina case—the Bureau concluded that “a similar issue was not problematic elsewhere in the country” on the strength of a search for the text string “dorm” in the address field of the decennial census MAF. 3 Using the final population figures from the Count Question Resolution Program, North Carolina would retain the 435th seat, but by a margin of only 87 residents (U.S. Government Accountability Office, 2005).

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census 43; the other 843 people were dormitory residents at Concordia University in the city of Mequon—in the same county, but tallied some 18 miles away from the proper location (Cole, 2001a,b). Poor Levels of Full, Self-Report Data A higher-than-expected proportion of group quarters enumerations were obtained from administrative or facility records; the means by which group quarters questionnaires were completed in 2000 are described in Table 7-1. Of the 83 percent of enumerations for which enumerators indicated the source of data, 59 percent were filled out from administrative data, 30 percent were filled out by the resident, and 12 percent were filled out by an enumerator interviewing the resident. Types of group quarters with high percentages of enumerations obtained from administrative data included nursing homes, hospitals, group homes, and prisons. These types of group quarters had especially high rates of missing data for long-form items. Lack of Coverage Measurement and Use of Reported “Usual Home Elsewhere” Addresses The Census Bureau opted not to include group quarters residents in the 2000 Accuracy and Coverage Evaluation program, principally because of low match rates in the 1990 Post-Enumeration Survey for group quarters residents in comparison with the rest of the household population. That is, the Bureau found that it was difficult to match group quarters residents interviewed in a follow-up survey (conducted a few months after the census) to their census records. This low match rate was attributed to high short-term mobility in the group quarters population (e.g., college students returning from their studies and shelter residents moving from one locale to another). Low match rates complicate the creation of population-adjusted census estimates using dual-systems estimation (the core methodology of the Accuracy and Coverage Evaluation Program), and so group quarters were dropped from consideration for the coverage evaluation program in 2000. Because of the difficulty in matching follow-up survey records with samples from the census, dual-systems estimation is most likely not the best means of assessing group quarters coverage. However, in the absence of estimates of omissions and erroneous enumerations in group quarters settings, the 2000 census plan did not include any systematic or comprehensive review of the coverage in group quarters, whether through rigorous comparison with facility or administrative records or through structured observational studies. Design choices in the Bureau’s Non-ID Process—the procedure by which the Bureau processed all census forms without a MAF identification number, including “Be Counted” forms and all group quarters forms indicating a “usual home elsewhere”—also led to a major lost opportunity for understanding residence reporting problems. As shown in Box 3-1, the Individual

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Table 7-1 Mode of Completion, Group Quarters Individual Census Reports, 2000 Census   Percent   Group Quarters Type Self-Response Administrative Records Enumerator Interview Unknowna Number Colleges and universities 57.5 30.2 5.5 6.7 2,028,150 Correctional institutions 15.3 56.3 4.4 24.0 1,930,233 Nursing homes 5.0 72.8 15.1 7.1 1,707,039 Service-based facilities and other group quartersb 25.3 41.3 23.3 10.1 669,702 Group homes 9.4 59.5 16.0 15.1 415,205 Hospitals 8.8 65.8 9.8 15.6 216,403 Juvenile institutions 23.8 48.8 9.9 17.5 122,291 Military facilities 36.9 37.6 5.7 19.7 279 Total 25.8 51.7 10.0 12.5 7,089,302 aEnumerators were supposed toenter a 1, 2, or 3 in a box on the back of group quarters forms to indicate that they were self-responses, enumerator interviews, oradministrative record entries, respectively. This “unknown” category represents those forms on which the box was left blank or an invalid entry was made. bService-based facilities include shelters and soup kitchens; “other group quarters” include work dormitories, religious group quarters, and crews of maritime vessels. SOURCE: Jonas (2003:Table 6.1b).

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Census Reports (ICRs) to be filled in by group quarters respondents asked for address information on the place where they live most of the time, regardless of whether the Bureau considered their group quarters type eligible for “usual home elsewhere” consideration. In principle, these group quarters questionnaires were supposed to be twice-filtered before Non-ID Process clerks geocoded the records and attempted to find matches on the MAF—first to determine if the Bureau considered the particular group quarters type in question as eligible for a “usual home elsewhere” and second on the basis of the response to a screening question on the form (“Do you have a place where you live or stay MOST OF THE TIME?”). In practice, though, the filtering by the screening question was not performed at all. More significantly, the filtering by group quarters type was only done after the reported address data had been read and matched to the MAF by census clerks (though before field verification would have been used to resolve the true location). Hence, as shown in Table 7-2, only 659,000 out of 2.9 million group quarters forms should have gone through the Non-ID Process; however, all 2.9 million went through the first clerical geocoding stage. Only then, when group quarters type was considered, were almost 2 million of them dropped from the process (and tabulated at the group quarters locations). Another 389,000 should have been excluded by the screening questions, but were not. From the standpoint of conducting the operation as planned, the failure to filter the group quarters questionnaires before Non-ID geocoding and matching was a significant mistake and a burden; the workload of the operation was multiplied. But from the perspective of research and evaluation, a geocoded set of claimed “usual home elsewhere” responses—which could be compared with the actual group quarters locations—would have been a trove of information on the nature of potential census duplicates and could have identified particular problem areas. That this type of file was (albeit inadvertently) constructed, then discarded when the filter was applied late, was a highly regrettable lost research opportunity. Ineffective Processing of Group Quarters Questionnaires Instead of a bar code tracking system for residents of particular group quarters sites, the Bureau relied on a total count of questionnaires from the group quarters site logged onto a control sheet. These control sheets were easily separated from their questionnaires: a special review of approximately 700,000 questionnaires had to be initiated in May 2000 when they became separated from their control sheets and hence could not be identified with their proper location. Worse, a small but undetermined number of group quarters questionnaires were never returned to the appropriate local census office and never included in the census.

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Table 7-2 Group Quarters Questionnaire Records in the Non-ID Process by Form Type, 2000 Census Record Type ICQs ICRs MCRs SCRs Total GQ records entering the Non-ID Process 8,551 2,232,674 630,252 69,801 2,941,278 Records filtered out due to invalid GQ type 222 1,862,295 30,225 0 1,892,742 Records remaining in Non-ID Process 8,329 370,379 600,027 69,801 1,048,536 Records that would have been filtered based on screening questions 1,432 345,524 39,251 2,763 388,970 Records that actually belonged in Non-ID Process, as planned 6,897 24,855 560,776 67,038 659,566 NOTES: Individual Census Questionnaires (ICQs) were used only for enumeration at soup kitchens and mobile food vans, and were intended to be filled out by enumerators (either through an interview or by observation). Individual Census Report (ICRs) are the basic group quarters data collection form, along with Military Census Reports (MCRs) and Shipboard Census Reports (SCRs); see Box 3-1. SOURCE: Census Bureau tabulation of Non-ID File; Jonas (2003:Table 8.2a).

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census High Levels of Imputation In July 2000 two special telephone operations were implemented to follow up group quarters with no recorded population (presumably refusals) and group quarters for which the reported resident count on census questionnaires fell far below the approximate population count (obtained in advance visits to special places conducted in February–March 2000). Population counts were obtained for these group quarters, and the results used to impute group quarters residents. More than 200,000 group quarters records (almost 3 percent of the group quarters population) were wholly imputed as a consequence of the telephone follow-up and the reconciliation of multiple population counts on the group quarters control sheets. Failure to Unduplicate Within the Group Quarters Population Some group quarters residents were mailed a housing unit questionnaire. If they returned it and the address was matched to a group quarters address, they were added to the appropriate group quarters count, but there was no provision to unduplicate such enumerations with enumerations obtained through the group quarters enumeration procedure. From a clerical review of a sample of cases in selected types of group quarters (excluding prisons, military bases, and service-based facilities such as soup kitchens), an estimated 56,000 group quarters enumerations were duplicates of other enumerations within the same group quarters. Our predecessor panel (National Research Council, 2004c:155) concluded that the 2000 census “procedures for enumerating group quarters residents and processing the information collected from them were not well controlled or carefully executed” and resulted in poor data quality. The National Research Council (2004b) concurred that group quarters enumeration needs complete revision for the 2010 census, as do we. Finding 7.1: As implemented in the 2000 and recent censuses, group quarters enumeration is unacceptably bad. Failure to reconcile the group quarters roster with the MAF contributed to a host of census errors. Group quarters frames were constructed without sufficient standardization and awareness of diversity in housing unit and group quarters stock, and data from the 2000 census long-form sample were particularly marred by extremely high levels of item nonresponse. It should be reiterated that group quarters represent only a very small portion of the overall population, but that the decennial census remains the sole source of comprehensive data on the size of all segments of this diverse group (which includes some especially policy-relevant subgroups). The problematic manner in which group quarters data were collected in 2000 raises the question

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census of whether the effort is given the right allocation of resources in the Bureau’s census plans. 7–B RETHINKING THE CONCEPT The Census Bureau has taken some positive steps with regard to group quarters enumeration, the most significant being an internal effort to revise and clarify its definitions of group quarters. These revised definitions were crafted for use in the 2006 census test and the 2008 dress rehearsal, and are described in Box 7-1. The redefinition effort reflects serious work to provide more meaningful categories and to be more consistent with terms used by practitioners. When the panel reviewed the definitions, they were still incomplete, with some major categories—among them military and seaborne quarters—yet to be fully developed. The new definitions were not ready for full testing in the 2004 field test, but were expected to be given a full airing in the 2006 census test and the 2008 dress rehearsal. We encourage the redefinition efforts and look forward to continued refinement. We concur with Drabek (2005:2) that it is particularly useful that the Bureau continue to work with experienced staff of all types of group quarters facilities in order to have a working vocabulary of terms that are relevant to communities and to data users. Particularly in elderly and health care, the Census Bureau needs to make sure that its definitions are based on the services that the facilities are licensed to provide, rather than rely on the name of an institution to provide a categorization. Though we support the group quarters redefinition effort in the short term, we strongly encourage the Census Bureau to broaden its focus. The lines between what the Bureau has traditionally walled off as “group quarters” as distinct from the “household” population are becoming increasingly blurred, and it is useful for the Bureau to reexamine the very concept of what it means to be a “group quarters.” Finding 7.2: There is sufficient diversity in what the Census Bureau has treated as the “group quarters” population that the term “group quarters” no longer makes conceptual sense. Its compartmentalization as a separate list and a separate operation—trying to force this entire segment of the population to respond to the census using a single form—is fundamentally flawed. To a very limited degree, two recent censuses attempted (or at least signaled an intent) to provide a more substantive split in the usual “group quarters” population. In 1980, enumerators assigned to group quarters were instructed to transcribe information from ICRs onto census short or long forms as appropriate, using “CENSUS USE ONLY” boxes on the form to indicate

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Box 7-1 2006 Census Test Group Quarters Definitions Draft definitions provided by the Census Bureau as of March 2005. The list of definitions is incomplete; military quarters, domestic violence shelters, and crews of maritime vessels are among the group quarters types left to be specified before the 2008 dress rehearsal. Group Quarters A group quarters is a place where people live or stay that is normally owned or managed by an entity or organization providing housing and/or services for the residents. These services may include custodial or medical care as well as other types of assistance, and residency is commonly restricted to those receiving these services. People living in group quarters are usually not related to each other. Group quarters include such places as college residence halls, residential treatment centers, skilled nursing facilities, group homes, military barracks, correctional facilities, workers’ dormitories, and facilities for people experiencing homelessness. Correctional Facilities for Adults Correctional Residential Facilities: These are community-based facilities operated for correctional purposes. The facility residents may be allowed extensive contact with the community, such as for employment or attending school, but are obligated to occupy the premises at night. Examples are halfway houses, restitution centers, and prerelease, work release, and study centers. Federal Detention Centers: Standalone, generally multilevel, federally operated correctional facilities that provide “short-term” confinement or custody of adults pending adjudication or sentencing. These facilities may hold pretrial detainees, holdovers, sentenced offenders, and Immigration and Customs Enforcement (ICE) inmates, formerly called Immigration and Naturalization Service inmates. These facilities include: Metropolitan Correctional Centers, Metropolitan Detention Centers, Federal Detention Centers, Bureau of Indian Affairs Detention Centers, ICE Service Processing Centers, and ICE contract detention facilities. Federal and State Prisons: Adult correctional facilities where people convicted of crimes serve their sentences. Common names include: prison, penitentiary, correctional institution, federal or state correctional facility, and conservation camp. The prisons are classified by two types of control: (1) “federal” (operated by or for the Bureau of Prisons of the Department of Justice) and (2) “state.” Residents who are forensic patients or criminally insane are classified on the basis of where they resided at the time of enumeration. Patients in hospitals (units, wings, or floors) operated by or for federal or state correctional authorities are counted in the prison population. Other forensic patients will be enumerated in psychiatric hospital units and floors for long-term nonacute patients. This category may include privately operated correctional facilities. Local Jails and Other Municipal Confinement Facilities: Correctional facilities operated by or for counties, cities, and American Indian and Alaska Native tribal governments. These facilities hold adults detained pending adjudication and/or people committed after adjudication. This category also includes work farms and camps used to hold people awaiting trial or serving time on relatively short sentences. Residents who are forensic patients or criminally insane are classified on the basis of where they resided at the time of enumeration. Patients in hospitals (units, wings, or floors) operated by or for local correctional authorities are counted in the jail population. Other forensic patients will be enumerated in psychiatric hospital units and floors for long-term nonacute patients. This category may include privately operated correctional facilities.

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Group Homes and Residential Treatment Centers for Adults Group Homes Intended for Adults: Group homes are community-based group living arrangements in residential settings usually consisting of three or more clients of a service provider. The group home provides room and board and supportive services, such as assistance with daily living skills, and social, psychological, or behavioral programs. Clients are generally not related to the caregiver or to each other. Group homes do not include residential treatment centers or facilities operated by or for correctional authorities. Residential Treatment Centers for Adults: Residential facilities that provide treatment on-site in a highly structured live-in environment for the treatment of drug/alcohol abuse, mental illness, and emotional/behavioral disorders. They are staffed 24 hours a day. The focus of a residential treatment center is on the treatment program. Residential treatment centers do not include facilities operated by or for correctional authorities. Juvenile Facilities Correctional Facilities Intended for Juveniles: Includes specialized facilities that provide strict confinement for its residents and detain juveniles awaiting adjudication, commitment or placement, and/or those being held for diagnosis or classification. Also included are correctional facilities where residents are permitted contact with the community, for purposes such as attending school or holding a job. Examples are residential training schools and farms, reception and diagnostic centers, group homes operated by or for correctional authorities, detention centers, and boot camps for juvenile delinquents. Group Homes for Juveniles: Includes community-based group living arrangements for youth in residential settings usually consisting of three or more clients of a service provider. The group home provides room and board and supportive services, such as assistance with daily living skills, and social, psychological, or behavioral programs. Clients are generally not related to the caregiver or to each other. Examples are maternity homes for unwed mothers, orphanages, and homes for abused and neglected children in need of services. Group homes for juveniles do not include residential treatment centers for juveniles or group homes operated by or for correctional authorities. Residential Treatment Centers for Juveniles: Includes facilities that primarily serve youth that provide services on-site in a highly structured live-in environment for the treatment of drug/alcohol abuse, mental illness, and emotional/behavioral disorders. These facilities are staffed 24 hours a day. The focus of a residential treatment center is on the treatment program. Residential School-Related Facilities College/University Housing: College/university housing includes residence halls and dormitories owned, leased, or managed by a college, university, or seminary. Fraternity and sorority housing identified by the college or university are included as college housing. Students attending the U.S. Naval Academy, the U.S. Military Academy (West Point), the U.S. Coast Guard Academy, and the U.S. Air Force Academy are counted in military group quarters.

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census individual, and an enumeration based on a length-of-stay distinction may be more relevant for policy needs. 7–C ALLOW “ANY RESIDENCE ELSEWHERE” We strongly support a system in which all group quarters and nonhousehold residents are approached and enumerated in the same manner as the general household population, to the greatest extent practicable. Accordingly, Recommendation 6.2 applies to all census respondents: the panel recommends that “any residence elsewhere” (ARE) address information be collected for all group quarters and nonhousehold residents, just as we advocate its collection in the main household census form. As with the main household population, the physical collection of ARE data from nonhousehold respondents in 2010 is certainly feasible; indeed, “usual home elsewhere” (UHE) was asked on all group quarters ICRs in 2000, but only deemed valid for specific group quarters types. The Census Bureau’s failure to analyze the UHE data it collected on group quarters forms in 2000, particularly having progressed through the geocoding of reported addresses, looms large as a lost opportunity. That failure is a principal reason that we encourage the analysis of ARE data as a large-scale experiment in the 2010 census (and a future direction for 2020), rather than as part of the 2010 count. Unlike the 2000 experience, ARE information gathered in 2010 should be collected from all group quarters residents, and those data should be analyzed and evaluated extensively, to establish—for example—the degree to which college students’ reported address information matches (or fails to match) reports from their parental households. 7–D CONDUCTING THE COUNT Consistent with our support for treating household and nonhousehold populations alike, the direct enumeration of people in nonhousehold settings—questionnaires distributed to and filled out by respondents or administered by enumerator interview—is preferable to other means of data collection. But two important caveats are in order: they concern the use of facility and administrative records and the need for different forms for different purposes. 7–D.1 Facility and Administrative Records The first of these arises from the sobering, underlying message of Table 7-1. The 2000 census mounted a vigorous and highly visible partnership with community organizations; even with these strong efforts to boost awareness of and cooperation with the census, only about one-half of census records

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census for the nonhousehold population were obtained from direct enumeration. Instead, the dominant source of reported data (among those questionnaires where enumerators coded the source) was reference to facility and administrative records. Though direct enumeration is absolutely preferable to other means of other data collection, it is likewise absolutely unrealistic to assume that questionnaire delivery or enumerator access to all parts of the nonhousehold universe will be granted. Instead, participation will vary from place to place. It is admittedly anecdotal evidence, but an observation report filed by a Census Bureau employee (Jones, 2000) during the conduct of the 2000 census usefully highlights these varying levels of participation, commenting on the group quarters enumeration at three Seattle-area universities. One university participated fully, providing comprehensive lists of university housing (including fraternity and sorority houses and student apartments) and directing its own staff to complete the basic roster of all persons living in those units.4 That university also permitted enumerators (with student escorts) to knock on dormitory room doors to collect questionnaires and to set up tables in common areas. At the other extreme, another university cited privacy concerns: the university blocked access to its housing records and denied enumerators permission to enter dormitories, though it did distribute ICRs to current students. The third university fell in between these positions: it permitted access to housing records but denied access to the dorms, instead only permitting enumerators to sit outside campus cafeterias at lunchtime. The Census Bureau needs to try to achieve direct enumeration whenever and wherever possible—in prisons, colleges, and hospitals alike. To do so, it should continue to aggressively pursue partnerships with relevant authorities and to streamline processes by which institution staff may be sworn in as census agents for purposes of administering questionnaires. However, the Bureau also needs to confront the reality that it will have to rely on facility or administrative data in many (if not most) nonhousehold settings. Albeit on a much smaller scale than the group quarters portion of the full decennial census, both of these models are not foreign to the Census Bureau’s current repertoire of surveys. For example, the biennial Census of Juveniles in Residential Placement (CJRP), conducted on behalf of the Office of Juvenile Justice and Delinquency Prevention (discussed in Section 3–D.3), is an example of a data collection effort aimed at collecting administrative data from facilities. The questionnaire is oriented around a roster and provides a mechanism by which facilities can provide data to the Census Bureau in electronic 4 The enumeration of the student apartments hit a snag, though, in that the university-completed roster included only current students. The apartment complexes also housed spouses and children of current students as well as nonstudent roommates.

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census (spreadsheet) form if that is more convenient for them.5 The specific implementation of the CJRP instrument is not necessarily something we endorse—indeed, the list of “include” and “exclude” instructions on the instrument is particularly elaborate and potentially confusing, but the general approach of tailoring the collection method to the anticipated means by which facilities can respond is a useful one. Because administrative and facility records have been, and will almost certainly continue to be, a major source of data on the nonhousehold population, it is imperative that the Census Bureau undertake a continuing research effort to assess the accessibility of facility records at group quarters facilities and to determine whether the existing data systems meet census data collection needs. In a sense, what we envision is analogous to the development of the Census Bureau’s MAF. Following the 1990 census, the Bureau elected to take the 1990 address list as a base and to continue to update and edit it over subsequent years, rather than follow past practice and rebuild the list from scratch 10 years later. The 1990 and 2000 censuses relied on pre-census facility visit operations to establish contact with group quarters sites and generate preliminary population estimates for workload planning purposes. The research effort we envision would maintain facility listings as a continuous resource, much like the MAF. Through queries to the facilities, it should be determined whether the facility’s records can provide the data of interest—the short-form census items like name and race, plus whether alternate address information (ARE) is known for each person in the facility and whether stays can be characterized as short or long term. 7–D.2 Different Forms for Different Settings The second caveat to our general preference for direct enumeration whenever possible is that too much homogeneity in the style and substance of the questionnaire used for the nonhousehold population may be harmful. For instance, soldiers and sailors have a different vocabulary related to the nature of their residence and their length of stay at a location than do college students, who in turn probably respond differently to questions and probes than people incarcerated in a county jail. In 2000, the Military and Shipboard Census Reports used to collect information from soldiers and sailors differed slightly (with terminological differences) from the ICRs used for the rest of the nonhousehold population. Though it is inadvisable to push this guidance to extremes and try to develop a form for every population type—just as it is self-defeating to attempt to delineate a specific residence rule for ev- 5  The questionnaire is available at http://ojjdp.ncjrs.org/ojstatbb/cjrp/asp/methods.asp [6/1/06].

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census ery group—the Census Bureau should consider further opportunities to tailor forms to best reach large segments of the nonhousehold population. Recommendation 7.1: The Census Bureau should produce a small number of alternative census forms that collect a common core of information for different types of residence settings, such as those that are known to have long lengths of stay rather than short term stays. The Census Bureau should also develop a spreadsheet-type ledger form that reflects the reality that some “responses” will have to be obtained from facility administrative records or a central “gatekeeper.” These custom nonhousehold forms should include ARE queries and other relevant data items, such as time spent at the location and expected length of stay. The precise tailoring of census forms for the nonhousehold population is arguably more of an issue for the ACS than the short-form-only 2010 census. Subjecting group quarters persons to the full battery of long-form items leads to such incongruities as asking prisoners about their employment status last year or long-term-care residents about their commute to work. We discuss other aspects of the ACS in Section 8–C. 7–E COUNTING PRISONERS IN THE CENSUS As we describe in Section 3–D, a particular issue involving the nonhousehold population that has drawn considerable attention in the buildup to the 2010 census is whether prisoners should be counted at the prison location or at some other place. A provision in the Census Bureau’s 2006 appropriations “direct[ed] the Bureau to undertake a study on using prisoner’s permanent homes of record, as opposed to their incarceration sites, when determining their residences.” The Bureau was obligated to report back to the congressional appropriations committees within 90 days, and it did so in a report on February 21, 2006 (U.S. Census Bureau, 2006b). Prior to the release of the Bureau’s report, two reports from proponents of a change in the Bureau’s stance on tabulating prisoners were transmitted to the Bureau (Wagner et al., 2006; Levingston and Muller, 2006). The debate over counting prisoners has drawn the attention of editorial pages, most prominently the New York Times. In September 2005, an editorial described the Census Bureau’s practice of “[counting] incarcerated people as ‘residents’ of the prisons where most are held for only a short time, instead of counting them in the towns and cities where they actually live” as “a longstanding quirk” and “a troubling flaw.” “Instead of waiting until the next census in 2010,” the Times concluded, “the Census Bureau should simply change its procedures now” (New York Times, 2005). In April 2006, the

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Times returned to the topic, contending that the Bureau’s report to the appropriations committee was “stonewalling,” “an obtuse and evasive report that supports the bad old status quo.” “The report sets up a straw man by suggesting that the desired change might require the costly and invasive procedure of interviewing every inmate.” However, the Times contends, “all that is really necessary” to implement “the common-sense idea of counting inmates at their home rather than at prison” is “to treat inmates like everyone else. That means giving them questionnaires that ask, among other things, for their home addresses and interviewing them only when a form is not returned or when some other problem occurs” (New York Times, 2006). Others perceive the matter differently. The town of Jackson, Michigan, and the surrounding Blackman Township is home to the Southern Michigan Correctional Center as well as several other state correctional prisons (most on the site of the former State Prison of Southern Michigan). Thus, about 32 percent of the township’s total population of 22,500 are listed as being in institutionalized group quarters. “As we peruse Blackman [Township]’s census data,” editorialized the Jackson Citizen Patriot (2006), “we can’t be sure any of those results have not been skewed in one way or another by the inmate populace.” One possible approach for “the Census Bureau is to count inmates, not where they are incarcerated, but where they lived before.” However, the paper urged, “we do not favor that approach because the inmates are, well, not living there and not using the services offered by their hometowns”—not using the police, fire, and other municipal services that might be required of a prison-hosting community. That the problem of counting prisoners is complex is undeniable. Including thousands of people in legislative districts even though they are legally prohibited from voting can create distortions in representation, and both the urban areas from which many prisoners originate and the rural areas which house prisoners can make strident (and valid) arguments for state and federal funds to compensate for current and future services. Finding 7.3: Major growth in the prison population, accompanied by expansion in the number of correctional facilities maintained by the federal government and the states, has prompted challenges to the Census Bureau’s “usual residence” standard regarding the counting of the incarcerated population. Our guidance on how the prisoner population should be counted in the census is consistent with both our recommendation of a core set of residence principles and our proposal to broadly change group quarters enumeration. Under the panel’s recommended principles for determining residence (see Chapter 6), federal and state prisoners would be counted at the prison location because that location is the place where the prisoner lives and sleeps

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census more than any other place. This is consistent with current Census Bureau practice. We do not rule out a principle being added to our suggested listing to count prisoners at a location other than a prison; however, the information necessary for such a decision does not now exist. We differ from proponents of an immediate change in Census Bureau policy on counting prisoners because our focus is broader: our desired objective is the collection of secondary (“any residence elsewhere”) information for all census respondents. This broader scope leads us to recommend a major experiment as part of the 2010 census as the research base, rather than trying to force massive changes in the design and operations of the 2008 dress rehearsal. Accordingly, and consistent with our other recommendations on changes to nonhousehold enumeration, we recommend: Recommendation 7.2: A research and testing program, including experimentation as part of the 2010 census, should be initiated by the Census Bureau to evaluate the feasibility and cost of assigning incarcerated and institutionalized individuals, who have another address, to the other location. Some elaborations on our approach, and on the information needed to inform a change in the principle applying to the counting of prisoners, are in order. Our principles hold that determination of usual residence should be made at the level of the individual; this would mean that persons in prison need not have their residency fixed solely by virtue of their location in a structure identified as a prison. Using the panel’s recommended question-based approach and revised nonhousehold enumeration operations, the census could obtain individual-level information on time spent in prison and expected date of release. The Bureau would then have the information needed to make individual-level judgments on the most appropriate counting location. Wagner and Lotke (2004), Wagner et al. (2006), and Levingston and Muller (2006) all favor direct enumeration—individual questionnaire distribution and interviewing—as a preferred means for gathering information from prisoners. We agree, as we stated above. Wagner et al. (2006) cite press accounts of two notable success stories—direct enumeration, even in “supermax” security areas, of Folsom State Prison (California) and the Utah State Prison.6 However, particularly in the case of prisons, complete direct enumeration of inmates is an ideal but almost certainly unattainable goal. In 2000, less than 20 percent of 6 A “supermax” facility or unit “segregates inmates, restricts their movement, and limits their direct access to other inmates or prison staff.” As of 1997, 57 pure-“supermax” prisons were open in the United States, 16 of them in Texas alone (Cullen and Sundt, 2000:498). In the enumeration of a supermax unit in Utah, a prison official commented that “rather than pass forms back and forth … we had [census takers] ask them the questions through the cell” (Wagner et al., 2006:25).

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census the population in correctional facilities was enumerated through self-response to a questionnaire or through enumerator interview (see Table 7-1). Even with more prominent efforts to facilitate direct interviews in prisons, it is realistic to assume that a major share of the prison population in 2010 will have to be counted through reference to administrative and prison records.7 Hence, any prospect for counting prisoners at locations other than the presence depends vitally on the completeness, consistency, and accessibility of records maintained by individual prisons or by state and federal departments of corrections. The quality of these data resources is not well known, and this creates the critical research need—determining how well corrections department data sources match the information that can be gathered on a Census Bureau questionnaire. The necessity for reliance on administrative and facility records accentuates a critical flaw in the argument for an immediate change in prisoner counting policy: the alternative to counting prisoners at a place other than prison is not well defined. Both New York Times editorials noted above give the desired counting location as (an unqualified) “home” (New York Times, 2005, 2006), a word that has numerous interpretations as we have described throughout this report. Levingston and Muller (2006) refer to state corrections departments’ coding of “home of record,” while the 2006 appropriations language mandating a Census Bureau study directed them to look at “permanent homes of record.” Other words abound: an expert cited by Levingston and Muller (2006) considers how prisoners can be mapped to a “return address” (a street address to which they will return after release); the September 2005 New York Times editorial briefly mentions counting prisoners at their “preprison residences” (New York Times, 2005); and Rep. Jose Serrano (D-New York) fused several of these concepts in arguing that prisoners should be counted at “their last known permanent residenc[e] where they are most likely to return to upon release” (East Oregonian and Associated Press, 2006). Finally, Wagner et al. (2006) characterize the alternative as counting prisoners “where they live, not where they are temporarily confined.” These are more than mere semantic differences, in that they blend elements of past residential history (“permanent,” or the last-known “preprison residence”), prospective intent that is perhaps years in the future (“return”), and enduring ties to family or community (generally, “home” and “where they live”). A clear idea of what residence information is or is not coded in administrative records is essential to defining questions to be used in direct interviews when those are possible. In addition to the basic meaning of address record information that is accessible through facility records, it is also necessary to consider what level of geographic resolution is available. The interviews conducted by Levingston 7 An earlier Brennan Center for Justice (Allard et al., 2004:7) recommended adopting a “uniform prison enumeration process” rather than mixing self-enumeration and administrative fill-in.

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census and Muller (2006:9, 11) suggest different geographic resolutions in the prison files: some may yield street addresses (and individual housing units), but other records might be limited to the “sentencing district” or the “county of conviction.” An analysis described in Levingston and Muller (2006:9) compared on-file residence locations with the addresses to which prisoners moved on parole; this work suggested agreement at the level of the “neighborhood.” Though an “enduring ties” argument is frequently invoked to argue for changes to the Bureau’s prisoner counting policy, the strength of those ties merits empirical assessment. For example: Has the property (to which a prisoner is connected) changed ownership? Do respondents at the address have any contact with or relation to the prisoner? With what frequency would assigning prisoners to specific addresses add to the “household” count people whose very crimes have severed any “enduring ties” that may have existed (e.g., domestic or child abuse)? Levingston and Muller (2006) conjecture that about 20 percent of the prison population would not have addresses that are “meaningful, accurate, available, and verifiable,” and that procedures for counting the other 80 percent should not be unduly impeded by these “exceptional” populations.8 Assessment of these ideas requires gathering actual data such as through an experiment in 2010, as we recommend. The base residence information contained in state corrections department databases—and even the format of the data—is not well known. The last comprehensive survey of the information systems maintained by state corrections departments was conducted by the Bureau of Justice Statistics in 1998 (Bureau of Justice Statistics, 1998). More recently, Wagner and Lotke (2004) estimate that about 30 states maintain some form of electronic record of home address information. Collection of these data is likely done “by asking prisoners to self-report a last address during intake.” However, it is not clear exactly how that address is defined (e.g., last known address, specified family address, etc.) and queried in individual states’ questionnaires or how accurate the address data may be. Wagner et al. (2006:16) add that “often the addresses [on file] are not updated until the time of release.” Among those states that do maintain address data, the accuracy and geographic resolution is unclear: Clement and Keough (2004:10) analyzed Rhode Island Department of Corrections data to study patterns created by the state’s laws that deny convicted felons the right to vote (even when they are on parole, rather than in prison). The Rhode Island data include “last known self-reported address.” Excluding cases that were not Rhode Island residents, 4.4 percent of these self-reported addresses (688 records) were 8 The four “exceptional” populations specifically referenced by Levingston and Muller (2006:14) are someone whose family has moved to another address following his or her incarceration; someone who will never leave prison (e.g., serving a life sentence); someone who had no address at the time of incarceration; and a noncitizen whose official home is in another country.

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census unable to be geocoded to the municipality level, and an additional 7.4 percent were unable to be associated with specific tracts or neighborhoods in urban Providence. Wagner et al. (2006:16) observe that the North Carolina Department of Corrections “appears to have a virtually complete dataset of home residences for its population.” However—at least in the publicly accessible data—the information seems to be limited to county of conviction. For the remaining 20 states, it is not clear what information would be available from administrative records save perhaps for court records from time of sentencing, which would suffer as a measure of the “current” prison population. The evidence of political inequities in redistricting that can arise due to the counting of prisoners at the prison location is compelling. Short of counting prisoners at some location other than the prison—for which there is currently insufficient information as well as the lack of any principled way to do so—a partial remedy might be to provide tract- or block-level counts of prisoner populations as part of the Bureau’s data products for redistricting, a compromise position noted by both Wagner et al. (2006) and Levingston and Muller (2006). In spirit, this approach follows that already taken by the state of Kansas, which disagrees with the Census Bureau’s default placement of college students and military personnel (see Box 7-2). Similar to the Kansas model, at least two recent state legislative initiatives mandating “adjustment” similar to the Kansas model—but including prisoners as a target group—have recently been mounted. The “Prisoner Census Adjustment Act” has been introduced in both the 2004 and 2005 sessions of the Illinois state legislature. The act would require that “each State and local governmental entity in this State that operates a facility for the incarceration of persons convicted of a criminal offense, including a mental health institution for those persons, or that places any person convicted of a criminal offense in a private facility to be incarcerated on behalf of the governmental entity,” transmit data on prisoners already counted in the federal decennial census to the Illinois secretary of state. These data include prisoner name, age, gender, race, and “the last address at which the person resided before the person’s current incarceration.” Federal prisons in Illinois would be requested, though not obligated, to provide the same data. The secretary of state would then be responsible for producing tract-level data “adjust[ing] all relevant population counts reported in the census, including populations by age, gender, and race, as if the person resided at [the preincarceration] address on the day for which the census reports population” and eliminating them from the facility count. These adjusted data would then be the official series to be used for legislative redistricting. Identical legislative initiatives have been proposed in the New York state assembly, and they were proposed (but not enacted) in Texas in 2001.

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census Box 7-2 Kansas Census Adjustment Article 10, Section 1 of the Kansas state constitution dictates that state senatorial and representative districts be apportioned using population totals that count “military personnel stationed within the state” and “students attending college and universities within the state” at their permanent residence. Military personnel and college students currently stationed in the state who are not Kansas residents are to be excluded from the counts. [Enabling legislation would later define “permanent residence” as “a fixed place of abode or fixed domicile which a person intends to be such person’s residence and to which such person presently intends to return” and “resident” as “a person who declares that he or she is a resident of the state of Kansas and has a present intent to remain in the state.”] Added to the constitution through a popular referendum in 1988, the constitutional provision changed the apportionment date from the seventh year to the second year of each decade (e.g., 1992, 2002, and so forth), and indicated that the apportionment is to be done by modifying figures from the U.S. decennial census. Kansas’ secretary of state reported the adjusted 2000 census totals to the state legislature in July 2001 (Thornburgh, 2001). Respondents were advised that “your answers should reflect your residence as of census day: April 1, 2000.” Respondents were asked for address information of both their “current college/military address’ and their “permanent residence.” Notes in Thornburgh (2001) indicate that “many adjustment questionnaires were returned with duplicate responses to the current and permanent address questions,” suggesting some confusion with the questionnaire instructions. The questionnaire also included a question on whether the respondent would be 18 years or older by April 1, 2000, and a question on race and Hispanic origin (the latter questions asked using the same categories and one-or-more options as the 2000 census). Questionnaires were distributed by staff at colleges and military installations; some schools preferred to provide administrative data electronically rather than distribute questionnaires, and this was permitted if the provided information was deemed to satisfy the needed criteria. Questionnaires and data were to be returned by June 1, 2000. All Kansas colleges and military installations participated, although Fort Leavenworth initially refused; subsequently, Fort Leavenworth refused to actively participate in distributing and collecting questionnaires, but did permit state officers to do that work. Addresses (both “permanent” and “current”) reported on the questionnaires were geocoded and assigned to census blocks; counts aggregated by census block, age (17 and younger, 18 and older), and race were then added to or subtracted to block-level counts from the 2000 census P.L. 94-171 redistricting data file. Thornburgh (2001) notes particular difficulty in the 2000 processing, since the Kansas questionnaire results were tabulated by 1990 census block and the P.L. 94-171 data were in the new 2000 census tabulation geography; the new boundaries were not available early enough for state processing. The state also encountered 259 census blocks where—post-adding or subtracting students and prisoners—the population count was negative; 818 blocks had negative counts when the block count was further subdivided by race. The state was able to resolve some of the largest such deviations and issued a letter to the Census Bureau regarding the others. The Kansas adjustment in 1990 yielded a total net decrease of 32,194 (1.30%) for the state apportionment population; the 2000 adjustment yielded a net decrease of 16,161 (0.60%).

OCR for page 225
Once, Only Once, and in the Right Place: Residence Rules in the Decennial Census With prisoner counts at the time of redistricting, state redistricting bodies would then have the capacity to decide whether to include or exclude prisoners from proposed districts. The states’ interest in having such a separate prisoner count should be assessed by the Bureau as part of its work with state officials to determine the layout of the standard P.L. 94-171 redistricting data file in 2010.