The CSDA is committed to continuing its long-standing practice of contributing to the collection and dissemination of demographic data the advancement of methods in population research. Past projects and the new projects planned demonstrate the CSDA’s innovative research in this area.

1) Data collections.

The Rochester Youth Development Study (RYDS) is an ongoing longitudinal investigation of the development of antisocial behavior, including delinquency, drug use and gang membership. The study began in 1988 when 1000 adolescents and their parents from Rochester, New York, were interviewed about topics including their family relationships, peers, gang membership, delinquency, drug use, and education. The panel members have since been interviewed 11 more times. In addition, data were collected from official records such as police, schools, and social services. The oldest biological children of the adolescents of the original sample are the subjects of a new phase of the study, initially funded for 5 annual assessments. This important data set has provided the foundation for intensive analyses of the implications of family context and processes for the generation of various behavioral outcomes. Its unusual strengths include its representative design, in contrast to many studies of gangs that are selective samples of gang members only. The RYDS project depends crucially on the infrastructure of CSDA for the maintenance of the electronic data set. Because CSDA has technical expertise in dealing with confidential data, it maintains the complete version of the data and takes responsibility for integrating new waves into the main data set and archival activities. Lizotte, CSDA associate and Dean of Criminal Justice has requested that CSDA continue managing and archiving the RYDS.

The Upstate New York Infant Development Screening Program (Upstate KIDS) is an ongoing prospective cohort study of growth and development in children born in New York State and is a collaborative effort with the University at Albany School of Public Health, the New York State Health Department and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (Bell, Health Policy, is University at Albany PI). A sample of 5,147 families were enrolled when their infant (or infants) was 4 months of age and will be followed to age 3. Information on parental medical history, maternal pregnancy history, parental behaviors and exposures (e.g. dietary patterns, smoking status, alcohol consumption), and residential, occupational and daycare environments are topics covered on this annual questionnaire. Child development data are collected every 4 months during the first year and every 6 months thereafter. Developmental data includes infant growth, medical histories and the Ages & Stages Questionnaires® (a standardized screening instrument that addresses their child’s communication, gross motor, fine motor, problem solving and personal-social skills development). Infants or children who screen positive for developmental delays are referred for clinical assessment through the New York State Early Intervention Program (EIP). Finally, an M-CHAT screen for Autism Spectrum Disorders (ASDs) is completed at 18 and 24 months of age. The 5147 families enrolled include 1113 sets of twins and 49 higher order multiple births. Families were randomly selected using a population based sampling paradigm, with an oversample of children conceived with fertility medications and/or artificial reproductive technologies to ensure sufficient power to assess whether development and growth varied for children conceived with these technologies and treatments. Finally, at 8 months of age, parental consent to sample the newborn bloodspot for biologic markers, (e.g. cytokines), and environmental contaminants, (e.g. BPA), was obtained for infants in the cohort.

The New York State National Birth Defects Prevention Study (NBDPS) is funded by the Centers for Disease Control and Prevention. The University at Albany PI is Bell. This is an ongoing study designed to investigate genetic and environmental risk factors for major structural birth defects. The New York Center has participated in this study since it started in 1997. Each year approximately 11,000 children with major malformations are reported to the New York State Congenital Malformations Registry. In New York State, congenital anomalies are the second leading cause of death for infants and many who live require significant long-term care. The New York Center, which is a collaborative project with the New York State Department of Health and the University at Albany School of Public Health analyzes the associations among maternal medications, environmental and occupational exposures and birth defects. Using a CATI protocol, mothers are asked to provide information regarding medical history, pregnancy history and exposures during pregnancy such as diet, occupational exposures, medication use and economic indicators. For participants who agree, buccal cells are collected for genetic analyses. Investigators are currently obtaining consent to use the newborn bloodspots to measure additional biologic markers and environmental contaminants.

2) Methodological contributions.

Gage with support from DiRienzo and Stratton have developed the Covariate Density Defined mixture ofregressions (CDDmr) method to account for unmeasured heterogeneity. The method is innovative because it derives information on heterogeneity (latent group membership) from the density of a covariate, using a finite parametric mixture model. The subpopulations are typically assumed normal distributions, but other distributions are possible, as well. Simultaneously, a separate general linear model is fit for each subpopulation, weighted by the latent membership determined by the mixing submodel. In the case of infant mortality, where the general linear models take the logistic form, the mixture submodel is composed of 2 or 3 normal distributions fitted to the density of birth weight. If covariates such as maternal age are included, the model can decompose the direct and indirect (through birth weight) effects of maternal age on infant mortality. The innovation fully operationalizes the proximate determinants model of infant mortality. The investigators are now applying this methodology to the fetal programing hypothesis.

DiRienzo is also developing a statistical methodology for selecting the parsimonious covariates in non¬parametric and semi-parametric statistical models with longitudinal data, missing data and censoring. This work investigates the statistical theory underpinning such methods, along with developing computational algorithms to execute the methodology. The development of computational algorithms is especially paramount for datasets with a great many potential covariates. He is active in applying such methodology in areas such as infant mortality, fetal programming, investigating the genetic basis for HIV-1 drug resistance, and estimation and prediction of HCV incidence and prevalence.

Spatial data analysis has been a longstanding area of interest to CSDA associate Deane because of its reliance on formal specifications of the structure of neighbors and its emphasis on identifying spatial outliers. Efficient use of spatially rendered data with varying spatial resolution motivated work produced with colleagues at CIESIN-Columbia. An NSF-funded project with political scientist Ron Mitchell investigated the link between environmental treaty protocols (LRTAP) and structural breaks in cross-national time series of pollution emissions. Collaboration with biological anthropologist Lawrence Schell, education psychologist Joan Newman, and their associates, has examined the effects of environmental toxicants on human growth and cognitive functioning. Deane and his collaborators used methods of exploratory spatial data analysis (ESDA), substantive applications of spatial regression, and diagnostic assessments of spatial regressions to yield insights into omitted sources of spatial dependence to better understand spatial variability in crime. In several papers addressing offense specialization Deane exploits the utility of marginal logit models to expose the underlying structure of generalized offending. Elsewhere Deane and Denton propose a related use of generalized estimating equations (GEE) in survey data analysis to allow multiracial individuals to contribute multiple observations, each with an unique race identification but with redundant information elsewhere. He has other projects linking environment to health and health behavior includes work with child and family development specialist Davison to examine environmental influences on physical activity and healthy body weight among children and, with community health scientist Jurkowski, the design and evaluation of the effectiveness of a community-based child obesity prevention program. Funding is currently under review, with research scientists brought together by the Cedar Grove Institute for Sustainable Communities, to calculate local (neighborhood/community) indices of cumulative health risk from inadequate or failed sanitation systems, contaminated water, and related factors from point, line, and polygon features of local government infrastructure in residential communities, census block-group attributes using network-based measures of spatial structure.

Yucel’s (Epidemiology and Biostatistics) recent research has focused on the problem of missing data in surveys, experiments, or observational studies. Motivated by substantive issues in health sciences as well as social sciences, Yucel has sought to address these issues by extending missing-data methods. He has focused his research efforts in four areas: 1) analysis of incomplete data; 2) software development; 3) response error and combining information using multiple data sources; and 4) novel applications of these methods in public health, genetic epidemiology, social sciences and biodemography. Yucel has participated in several of the working groups, Biodemography, and the Adolescent Health Data, and provided formal workshops on missing values for CSDA.

The recent hiring of Dreby (Sociology) has added expertise in the use of ethnographic methods for population level studies to CSDAs associates. Her in-depth interviews of Mexican children, their parents, and grandparents explore the issues when different generations of the same family are in different countries. As immigration continues, this practice will increase, andDreby’s work is at the cutting edge. We expect that she will provide a new and relatively little used tool to CSDA’s methods.

In the future we will continue our support for the three data collection efforts described above as they are all relevant to DBSB’s mission of child and adolescent health. We expect our methodological contributions to increase, especially as we invest in nurturing what we plan to be a fully-fledged Statistical Core (explained in more detail in the Developmental Infrastructure Core) by the time of our next submission. CSDA’s initiatives in this theme contribute to almost all of NICHD’s thematic areas, including pregnancy, early origins of health, disease and growth, and development behavior and cognition, plasticity and rehabilitation, and finally, population dynamics.