The design of a multilevel survey of children, families, and communities: The Los Angeles Family and Neighborhood Survey☆
Introduction
In the last fifteen years, there has been a growing interest in the role of neighborhoods in shaping a variety of outcomes for children, adults, and families.1 The broad set of outcomes that have been studied includes violent crime (Sampson et al., 1997), educational attainment (Garner and Raudenbush, 1991), domestic violence (O’Campo et al., 1995), fertility (Billy and Moore, 1992), residential mobility (Lee et al., 1994), children’s development (Duncan et al., 1994), teenage pregnancy (South and Baumer, 2000), and health status (Robert, 1999). Studies examining neighborhood effects have considered not only the US, but also countries overseas (e.g., Pebley et al., 1996, Sampson and Groves, 1989, Sastry, 1996). Understanding the effects of neighborhoods is potentially quite important, from both a research and a policy perspective. For example, neighborhoods may provide a more cost-effective focus for interventions than individuals or families. Research in this area has, however, failed to produce persuasive and consistent results about the nature and strength of neighborhood effects, especially regarding children’s outcomes (Duncan and Raudenbush, 2001, Furstenberg and Hughes, 1997, Gephard, 1997, Jencks and Mayer, 1990). Although theoretical perspectives are well advanced and statistical methods for modeling neighborhood effects are in place, a major shortcoming concerns the limitations of existing datasets. The Los Angeles Family and Neighborhood Survey (L.A.FANS) was designed to address many of these limitations.
Very few existing datasets have been designed for the purpose of studying neighborhood effects on child and family outcomes.2 Rather, the existing neighborhood effects literature has largely used standard household surveys—such as the Panel Study of Income Dynamics—linked with contextual data obtained primarily from the decennial census. The near-universal practice of employing multistage, clustered sampling schemes in the design of standard household surveys has supported the widespread use of multilevel models in this research. These sampling schemes are designed to achieve a balance between efficiency of fieldwork operations, which is facilitated by concentrating the sample in a relatively small number of compact areas, and the magnitude of design effects, which increase with the number of respondents per cluster to reduce the effective sample size.
Existing datasets for studying neighborhood effects on child and family outcomes have a number of limitations. Contextual data are generally restricted to published sources. Extremely few surveys have collected neighborhood measures as part of the fieldwork, either from the respondents themselves (for example, on their perceptions of neighborhood characteristics and conditions) or by interviewers (for example, based on their observation of neighborhood conditions during visits to conduct interviews). Furthermore, sample sizes per cluster are generally too small to estimate community-level explanatory variables with sufficient reliability and precision. The small cluster sample sizes in many of these datasets also make it difficult to distinguish family effects from neighborhood effects.3 Existing datasets lack detailed measures on several topics that are important for studying neighborhood effects. These topics include past residential mobility, neighborhood choice, neighborhood definitions, neighborhood characteristics and conditions, and the salience of neighborhoods for daily life. Finally, although longitudinal information on individuals is available from several existing datasets, prospective information on sampled neighborhoods has not been collected. In particular, people moving into sampled neighborhoods are not interviewed. Therefore, the surveys do not capture a representative cross-section of residents at each wave, which is essential for understanding patterns of residential mobility and tracking neighborhood change.
These limitations in the designs of existing datasets make it difficult to address the most pressing unresolved research issue in understanding the effects of neighborhoods on children’s outcomes, which is to uncover the extent to which these effects are causal. The prime process leading to endogenous neighborhood characteristics is that families choose the neighborhoods in which they live—and this choice may be related to the outcome of interest, such as children’s development or well-being. It is necessary to understand this process to understand the effects of neighborhood characteristics on outcomes. Another implication is that few analyses to date have examined the effects of a richer set of contextual characteristics, particularly those relating to neighborhood social processes. Finally, little is known about how residential mobility and neighborhood choice work to influence how neighborhoods affect children and families. We return to these issues in the conclusions, where we discuss methods for studying neighborhood effects while addressing the issue that neighborhood characteristics may be endogenous.
In this paper, we describe the sampling design of the Los Angeles Family and Neighborhood Study (L.A.FANS), a survey of children, families, and neighborhoods in Los Angeles County. This survey was designed to support multilevel studies on a number of topics relating to child and family well-being, with a focus on children’s cognitive, behavioral, and social development, anti-social and self-destructive behaviors, schooling, child care, and health. The baseline wave was completed in 2000–2001. We first outline the theoretical roots of the study and how they affected design choices. We highlight the main design and analytical considerations that shaped the baseline wave and describe the results of an in-depth statistical investigation of the survey’s ability to support multilevel analyses that was carried out as part of the study design.
Section snippets
Theoretical background
L.A.FANS is based on a conceptual model drawn from several theoretical traditions in sociology, psychology, economics, and other disciplines. The model assumes that social environments (e.g., families, work and school environments, social circles and networks, formal and informal social institutions, etc.) influence individuals’ lives, including their physical safety, health, attitudes and behaviors, social ties and interactions, and economic and social opportunity structures. For example,
Sample design for multilevel models
A central issue in sample design for standard social surveys is to maximize the efficiency of fieldwork operations by concentrating the sample in the smallest number of primary sampling units while minimizing design effects by limiting the concentration of the sample within primary sampling units. Given sufficient funding, the optimal design would be for the sample to be chosen completely at random from the population with no spatial clustering. In surveys designed to support multilevel
The Los Angeles Family and Neighborhood Survey
The Los Angeles Family and Neighborhood Survey is a study of families in Los Angeles County and of the neighborhoods in which they live. As described above, the survey was designed to support research in three main areas: the effects of neighborhoods and families on children’s development and well-being; the effects of welfare reform at the neighborhood level; and the process of residential mobility and neighborhood change.
L.A.FANS sample design
L.A.FANS was designed as a multilevel survey, first sampling neighborhoods, then sampling blocks within these neighborhoods, next selecting families within these blocks, and finally sampling children and adults within these families. As discussed below in greater detail, the multilevel sampling scheme has several strengths, as well as certain limitations. Two strengths are worth mentioning here. First, it provides an efficient and cost-effective method for collecting detailed information about
Discussion and future directions
The design of innovative new surveys is a crucial methodological ingredient for advancing research in many areas of social science, but especially in understanding the effects of neighborhood factors on the health, development, and well-being of children, adolescents, and families. The goal of this paper was to outline the central substantive and methodological issues that shaped the sampling design of the Los Angeles Family and Neighborhood Survey, a new survey to support multilevel analyses
References (87)
- et al.
Segregation, the concentration of poverty, and the life chances of individuals
Social Science Research
(1991) Sibling estimates of neighborhood effects
Using sibling data to estimate the impact of neighborhoods on children’s educational outcomes
Journal of Human Resources
(1998)- et al.
Development in context: implications for studying neighborhood effects
- Afshartous, D., 1995. Determination of sample size for multilevel model design. Paper presented at the AERA meeting in...
- et al.
The neighborhood context of adolescent mental health
Journal of Health and Social Behavior
(1996) - et al.
The neighborhood history calendar: a data collection method designed for dynamic multilevel modeling
Sociological Methodology
(1997) Social cognitive theory: an agentic perspective
Annual Review of Psychology
(2001)- Bassiri, D., 1988. Large and Small Sample Properties of Maximum Likelihood Estimates for the Hierarchical Linear Model....
- et al.
A multilevel analysis of marital and nonmarital fertility in the United States
Social Forces
(1992)
Job control, personal characteristics, and health disease
Journal of Occupational Health Psychology
Ecology of the family as context for human development
Developmental Psychology
The value of sociological community studies for the work of social agencies
Social Forces
Analysis of covariance with qualitative data
Review of Economic Studies
Mothers’ transitions from welfare to work and the well-being of preschoolers and adolescents
Science
Sampling Techniques
Determining sample sizes for surveys with data analyzed by hierarchical linear models
Journal of Official Statistics
Social capital in the creation of human capital
American Journal of Sociology
Mapping residents’ perceptions of neighborhood boundaries: a methodological note
American Journal of Community Psychology
Design and Analysis of Cluster Randomization Trials in Health Research
Neighborhoods and adolescent development: How can we determine the links
Conceptual and methodological issues in estimating causal effects of neighborhoods and family conditions on individual development
Teenage underclass behavior and subsequent poverty: Have the rules changed?
Economic deprivation and early childhood development
Child Development
Measuring peer group effects: a study of teenage behavior
Journal of Political Economy
Opportunity and Change
Black students’ school success: coping with the ‘burden of acting white’
The Urban Review
An illustration of the use of instrumental variables: Do neighborhood conditions affect a young person’s chance of finishing high school?
Psychological Methods
The influence of neighborhoods on children’s development: a theoretical perspective and a research agenda
Neighborhood effects on educational attainment: a multilevel analysis
Sociology of Education
Neighborhoods and communities as contexts for development
Neighborhood attributes as determinants of children’s outcomes: How robust are the relationships?
Journal of Human Resources
The strength of weak ties
American Journal of Sociology
How urbanites define their neighborhoods
Population and Environment
Counterfactual models of neighborhood effects: the effect of neighborhood poverty on dropping out and teenage pregnancy
American Journal of Sociology
Succeeding Generations: On the Effects of Investments in Children
Human Ecology: A Theory of Community Structure
Sample size estimation for longitudinal designs with attrition: comparing time-related contrasts between two groups
Journal of Educational and Behavioral Statistics
Multilevel modeling: When and how
Sample sizes for multilevel modeling
Cited by (111)
Building towards an adolescent neural urbanome: Expanding environmental measures using linked external data (LED) in the ABCD study
2024, Developmental Cognitive NeuroscienceRevealing racial-ethnic segregation with individual experienced segregation indices based on social media data: A case study in Los Angeles-Long Beach-Anaheim
2023, Computers, Environment and Urban SystemsPredicting the slump of industrially produced concrete using machine learning: A multiclass classification approach
2022, Journal of Building EngineeringUtilizing geo-referenced imagery for systematic social observation of neighborhood disorder
2021, Computers, Environment and Urban SystemsShifting neighborhoods, shifting health: A longitudinal analysis of gentrification and health in Los Angeles County
2021, Social Science ResearchThe spaces between: Parents' perceptions of neighborhood cohesion and child well-being
2021, Journal of Applied Developmental Psychology
- ☆
The Los Angeles Family and Neighborhood Survey was funded by NICHD Grant R01HD35944. Additional funding was provided by the Office of the Assistant Secretary for Planning and Evaluation (OASPE) of the US Department of Health and Human Services, Los Angeles County, the National Institutes of Health Office of Behavioral and Social Science Research (OBSSR), and the Russell Sage Foundation. We thank our colleagues at UCLA and RAND for comments and suggestions.