Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Integrating Federal Statistics on Children Interest in monitoring and understanding the lives of children has grown rapidly in recent years. Fueled in part by growing pressures to hold public programs accountable for outcomes, as well as by mounting concerns about the instability and apparently worsening problems that characterize the lives of many children, those who shape our nationâs child policies are increas- ingly looking to the federal statistical system for answers to complex ques- tions about the development of children in todayâs society. At the same time, the nation is contemplating a major shift in responsibility for several major childrenâs programs from the federal to state governments. Such a shift will place even greater demands on the capacity of national data to track and release in a timely fashion information on the effects on childrenâs well-being of this major redirection of public resources. It is in this context that the Committee on National Statistics and the Board on Children and Families of the National Research Council and the Institute of Medicine convened a workshop to examine the adequacy of federal statistics on children and families. Through a series of background papers, discussantsâ remarks, and participant discussions, the workshop pro- vided a forum for a preliminary assessment of the strengths and shortcom- ings of existing and proposed federal statistical data sources, particularly with respect to their capacity to fill the most pressing information needs of those who formulate, implement, and analyze policies for children. The papers covered several types of data collection initiatives: (1) those designed to track the effects of major policy developments: we examined 1
2 INTEGRATING FEDERAL STATISTICS ON CHILDREN data needs related to health care reorganization (Newacheck and Starfield), but similar issues also pertain to policies in the area of public assistance and child welfare; (2) those designed primarily to explain the development and well-being of children, particularly across developmental transitions, such as school and work force entry, and with regard to the context of family, community, and government resources (Brooks-Gunn et al., Hofferth, Pallas); and (3) those designed to access the incidence, sequelae, and consequences of relatively rare but developmentally salient events, such as child abuse and other forms of violence involving children (Loftin and Mercy). Over the course of the workshop, the discussion converged on several cross-cutting themes, and the participants made numerous suggestionsâ some highly detailed and others more generalâfor improving the existing statistical system. This introductory chapter highlights the common themes that emerged at the workshop, none of which should be viewed as formal consensus opinions, and concludes with a few ideas about next steps. The next chapter provides a summary of the workshop presentations and discus- sion. The five workshop papers conclude the body of the report. CROSS-CUTTING THEMES The title of the workshop itself reflects an assumption, amply con- firmed by the participants, that federal statistical data on children and fami- lies are highly fragmented. Just as policy for children in the United States is best portrayed as an accumulation of responses to problems that are only rarely viewed as interrelated, federal statistics on children are far from being a coherent, logically developed system. Data collection efforts are highly fragmented, administered by different agencies, each with its own substantive interests and each with distinctive histories and constituencies. As noted in the paper by Aaron Pallas, this context must be taken into account in trying to understand why the âsystemâ looks the way it does, and in attempting to address the key redundancies and gaps that characterize the existing collection of data resources. In effect, each of the participants grappled with the conclusion that we have both too many and too few statistical data on children. Furthermore, the participants were cognizant of the limited, if not shrinking, funds for federal data collection efforts. The net effect of budgetary limita- tions on heightening pressures for greater efficiency and coordination across the federal statistical agencies and on constraining the range of possible responses to improving the existing system was a topic of substantial dis- cussion at the workshop. With these shared concerns about fragmentation and resource constraints as a point of departure, the following common themes regarding problems and suggestions for improving federal statistics on children emerged from the workshop discussions.
INTEGRATING FEDERAL STATISTICS ON CHILDREN 3 â¢ Improvements in data are needed to understand the connections be- tween resources and child outcomes, as well as the family and com- munity processes that translate resources into outcomes. Many of the most pressing policy questions now center on identifying promising targets of intervention, whether the topic is preventing childhood illness, ensuring for children a successful start in school, or reducing ineq- uitable patterns of school dropout and completion. Supplementing family income, changing parentsâ behavior through parent education or family planning interventions, investing in social services, and improving the neighborhoods and other settings in which children develop offer competing policy re- sponses to a range of issues. Answers to these questions emphasize the importance of collecting data that can be put to more than purely descriptive purposes. This, in turn, requires that careful assessment of resources and other inputs to childrenâs development be accompanied by assessments of child outcomes in the same survey. Assessments of family functioning (e.g., communication patterns, supervision and discipline, the learning environment) and of the peer, schooling, and neighborhood-level settings and processes that can explain how re- sources affect outcomes would further enhance the capacity of childhood data to inform the efficient allocation of public resources toward the most potent sources of child well-being. Although several major datasets (i.e., the Panel Study of Income Dy- namics, the Survey of Income and Program Participation) provide relatively rich data on inputs to development and others focus on family functioning (i.e., the National Survey of Families and Households), data on child out- comes are substantially more limited, particularly during the pre-high-school years, and no single survey presently encompasses a good complement of measures of resources, processes, and outcomes. With respect to outcomes, the dearth of âpositiveâ measures that will enable us to learn from those who fare well about the ingredients of successful development was specifi- cally noted as a critical need to complement our current capacity to learn from those who fare poorly about the factors that compromise development. Among the many types of resources that were discussed (e.g., mon- etary, psychological, human capital, community resources and networks, and resources that arise from government policies), the resource of parental time inputs surfaced as a major gap in current data collection efforts. Par- ticipants called for the addition of question sequences focused on parental time use to existing surveys, as well as for a separate national time use survey of children and their parents. Beyond the simple accounting of resources, data collection on the orga- nization and distribution of resourcesânotably health care, early education and schooling experiencesâacross subgroups of children was recommended.
4 INTEGRATING FEDERAL STATISTICS ON CHILDREN This information is essential to assessing equity of access to public re- sources, as well as differential outcomes for children who might otherwise appear to grow up under similar circumstances. More careful measurement of income, including items that distinguish among various sources of in- come, capture what families do with their income, and assess the allocation of income (and other resources) across family members and continuing through the time when children move out of the parental home, was also highlighted as a major information need. â¢ It is important to collect information on family relationships that goes beyond each memberâs relation to a single reference person, distinguishes among biological, step-, and adoptive parents, and in- cludes noncustodial parents. Several presenters and discussants called attention to the value of estab- lishing relations among all pairs of individuals in a household and of identi- fying the nature of the parent-child relationship (biological, stepparent, adoptive) for all child-parent dyads living in and out of the household. This is the only way in which relations among subgroups of family members, as well as issues that bear on differing degrees of genetic relatedness among care- takers and children, can be examined. Grandparent-child-grandchild rela- tions, comparisons of developmental trajectories for siblings, and the role of noncustodial parents and nonresidential family members in childrenâs lives were salient themes in the discussion. A related need concerned the inclusion of identifiers that enable the linkage of family members and the subsequent creation of family-level records in addition to individual and person records aimed at encouraging family-level data analysis of datasets ranging from the National Health Interview Survey to the Public Use MicroSample (PUMS) microdata files derived from the 1990 census. Other presenters noted the limits of household-based surveys for ad- dressing particular policy issues, notably those concerning violence against children and educational attainment and the transition to work. In the first case, exclusive reliance on a household-based sampling frame misses vital information on episodes of violence perpetrated by individuals who are weakly attached to the childâs household. It also militates against important efforts to integrate information about child victimization at home, in school, and in other settings. In the second case, household surveys typically are not designed to ensure a sufficient sample of individuals within a particular organizational context, such as a school or workplace. These examples raise the methodological challenge of experimenting with unconventional sampling plans. â¢ Given the current move toward block grants, the state and local locus of many policies and programs for children, and growing in- terest in the effects of state variation in benefit levels, service struc-
INTEGRATING FEDERAL STATISTICS ON CHILDREN 5 tures, and other key features of social policy programs, there is a critical need for data that will allow for reliable state- and local- level estimates. This recommendation surfaced with respect to every policy area dis- cussed at the workshop. For example, health care reform is taking different forms in different states, yet none of the major federal health surveys has the capacity to assess the effects of state-level reform efforts on children and families. Pressing needs include the selection of primary sampling units that are consistent with state-level estimation, inclusion of a sufficient number of cases at the state level to permit accurate estimation, and making better use of claims data and other administrative records. The value of linking qualitative and local data to national health surveys in order to interpret local health care trends, processes, and effects on children and families was also noted. Similarly, efforts to track and prevent violence are implemented at the local level and require data that can produce small-area estimates. Presenters also called for adding state-of-residence identifiers to national data as a routine matter in order to encourage analysis of state-to- state variation in program benefits and structures. The significant contribution to available information that would be made by appending neighborhood-based data to existing data files was also noted by several participants, particularly with respect to questions of local re- sources, economic opportunities, aggregate poverty levels, and other fea- tures of the surrounding population. Suggestions for enhancing capacity at state and local levels to conduct surveys and to collect useful administrative data were also made. The broader issue in which this discussion was embedded concerned the importance of being able to link data across geographic levels (federal, state, local) for the purpose of examining the effects of how policies devel- oped at different levels of government interact with each other and, in turn, create differing patterns of intended or unintended consequences for chil- dren. Coordination of data across other levels of analysis unique to particu- lar issues was also discussed. Understanding the effects of health care reforms requires outcome data that can be aggregated at the patient-pro- vider, plan or system of care, and community or population levels. Ex- planatory variables in datasets aimed at clarifying child development would ideally be measured at the household, community, institutional (i.e., schools), and state levels. â¢ The changing demographics of the childhood population, as well as shifting policy concerns, require new strategies for oversampling currently under- and unrepresented subgroups of children. Each of the presenters noted that the most significant effects of policy developments are often experienced by groups of children that are seriously
6 INTEGRATING FEDERAL STATISTICS ON CHILDREN underrepresented or excluded from federal statistical data on children. Fur- thermore, rapid growth in immigrant families and the fact that many chil- dren experience shifting family structures call for a reassessment of the sampling strategies that are typically employed in federally supported sur- veys. Populations of particular concern vary somewhat by substantive area. For example, chronically ill children were identified as a critical subgroup to capture in health surveys; youth in prisons or jails, in the military, and those who were early school dropouts were identified with respect to sur- veys of the transition from school to work; and children of non-English- speaking parents were identified as a critical subgroup when studying the preschool and early school years. Common across topics were calls to sample sufficient numbers of His- panic and Asian children (and to disaggregate these broad categories into distinct subgroups, such as Mexican-Americans and Puerto Ricans), institu- tionalized and homeless children, children in out-of-home placements, and children with disabilities and other special needs. In light of immigration trends, calls were also made for routinely including information about length of time in the United States and country of origin for all family members. â¢ Improved longitudinal data on children and families would facilitate efforts to address several critical policy issues pertaining to changes in family resources, predictors of successful development across key transition points, and the identification of early precursors of seri- ous problems in middle childhood and adolescence. Among the pressing questions that require longitudinal data are: (1) How do changes in resources and family arrangements affect children? (2) What factors predict successful transitions into school and from schooling to work? (3) What are the long-term consequences of assaultive violence on children? (4) What is known about the sequencing of health events, as well as behavioral problems, over the course of development? It was uniformly perceived that available data are seriously inadequate to address these ques- tions. Numerous longitudinal surveysâboth active and planned (e.g., the Na- tional Longitudinal Survey of Youth and its Child-Mother Supplement, the Survey of Income and Program Participation, the National Survey of Family Growth, the Early Childhood Longitudinal Survey)âwere discussed. As a collection, however, they were portrayed as flawed by a wide range of problems. These include limited and unrepresentative target populations; problems with length (age range followed) and periodicity of data collec- tion; specialized foci that militate against efforts to examine how behavior in one domain (i.e., health) interrelates with behavior in another domain (i.e., achievement); a failure to follow children across critical transition points, such as from the preschool to the school years; and sampling strate-
INTEGRATING FEDERAL STATISTICS ON CHILDREN 7 gies that follow households or adults and therefore lose children as they shift from one household to another or move out of a family setting. Although cognizant of the vast expense of mounting a new panel study of children and familiesâwhether initiated with a new, independently sampled cohort of children or by augmenting an existing surveyâseveral of the presenters felt that, if substantial new resources were devoted to federal statistics on children, this would be the wisest option to pursue. Other presenters noted that no single survey would ever address all of the infor- mation needs of policy makers and researchers and favored a more diversi- fied approach. Each of the presenters and discussants made suggestions for lower-cost supplements to samples and items as well as other changes to existing surveys as a short-term strategy and one that would complement the longer-term strategy of developing a new national survey of children. â¢ The critical need for improved cross-agency planning and coordina- tion, as well as coordination across private and public data sources, emerged as a priority across the topical issues under discussion. As federal statistical data on children have accumulated, the databases from which these data derive have become increasingly specialized and categorical in nature. Furthermore, even within agencies, supplements, topical modules, and new surveys tend to proliferate with minimal attention to opportunities for linkages with existing data collection efforts. In the area of health care, the need to integrate data across public and private data collection organizations was noted as an additional challenge facing those who seek useful data about the course and effects of health care reforms. The urgency of developing a cross-agency data coordinating mecha- nism that can direct efforts to reduce redundancies, cover gaps, standardize definitions and item formats, and, in general, establish a more efficient system, is compounded by current fiscal pressures. Furthermore, absent coordination, opportunities to share protocols and instruments, to consider complementary sampling strategies, to develop compatible data coding sys- tems, and to collaborate on analyses and reporting of results will be missed. Calls were also made for a coordinated methodological research program aimed at such issues as techniques for improving the validity of responses to sensitive questions, controversial tensions that surround the need to pro- tect respondent confidentiality, methods of linking data across surveys, and innovative approaches to collecting data from multiple informants and in new settings. The model of the Federal Interagency Forum on Aging-Related Statis- tics, which was established to encourage cooperation among federal agen- cies in the development, collection, analysis, and dissemination of data on the older population, was noted as an effective mechanism for interagency communication. A promising development, since the workshop, is the for-
8 INTEGRATING FEDERAL STATISTICS ON CHILDREN mation of a Federal Interagency Forum on Child and Family Statistics. The forum, which is a voluntary mechanism to improve coordination of child and family data on behalf of the federal statistical agencies, offers the op- portunity to engage in precisely the types of cross-agency planning sug- gested by the workshop participants. NEXT STEPS Over the course of the workshop, an ambitious set of suggestions for improving existing data and developing new data was made. Initial steps toward ensuring thoughtful and ongoing deliberations about the issues raised at the workshop were also suggested. These included: â¢ Promoting collaborative meetings. Participants noted an immediate opportunity to convene key federal statistical and research agencies (e.g., the National Institute on Mental Health, the National Institute on Child Health and Human Development, the Department of Education, the Depart- ment of Health and Human Services, the Bureau of the Census, the National Institute of Justice) for the purpose of sharing notes about surveys currently being planned. Along these lines, the value of developing an integrated consortium of sampling frames and topical modules for new surveys was noted. â¢ Coordinating efforts aimed at supplementing existing datasets. Nu- merous detailed suggestions were made by the workshop participants for improving the utility and quality of existing datasets. A forum for continu- ing this type of discussion and effective mechanisms for implementing agreed- on recommendations were viewed as pressing needs. Such a forum could also provide for consideration of the pros and cons of launching a new national survey of children. â¢ Continuing to examine the fit between the information needs of deci- sion makers and the terrain of available data. A sustained debate among data developers and data users, including policy makers, centered on the degree to which available data match information needs, was viewed as a very promising avenue for guiding the future of the federal statistical sys- tem. Key players would include staff from the statistical agencies, academ- ics, contractors, and policy experts from a variety of vantage points. â¢ Addressing the underutilization of existing data. Efforts to develop and disseminate means of encouraging greater access to existing data on behalf of evaluators and researchers not involved in the original data design and collection were also identified as warranting immediate attention. Spe- cific suggestions included involving data users early on in the development of new surveys, ascertaining effective approaches to training new users, and developing more accessible, user-friendly data files.