The survey research project proposed here will develop a common data resource for the study of family decisionmaking in a context of rapid demographic and socioeconomic change. This project will complement NICHD's other initiatives to advance population research with carefully designed and executed national surveys placed in the public domain. This proposal requests support for: (1) the design and implementation of a Second Malaysian Family Life Survey (MFLS- 2); (2) selected methodological and descriptive studies that will benefit future users of the data; (3) analyses of MFLS-2 data on infant feeding, child survival, fertility, child schooling, and intergenerational transfer; and (4) preparation and dissemination of a public-use data tape. The survey, based on the successful 1976-77 Malaysian Family Life Survey (MFLS-1), will be administered to three samples in Peninsular Malaysia: (1) the original respondents to the MFLS-1 and their husbands; (2) their children now aged 15 and over and their spouses; and (3) a new random sample of women aged 15-49 and their husbands. These are expected to total 4,000 main respondents. Preliminary studies suggest that adequate proportions of the first two samples can be located and interviewed. Data will be collected on respondents' marriage, fertility, infant feeding, child mortality, education, employment, and migration histories and those of their spouses; on household income, wealth, and expenditures; and on the 52 communities in which the MFLS-1 respondents resided and the 200 communities from which the new random sample is selected, covering health, family planning, and education services. Descriptive and data quality analyses, including analyzes of attrition bias and nonsampling error in the retrospective data will be performed, enabling preparation and dissemination of a fully documented public use data set. Intensive analysis of the MFLS-2 data, linked to the MFLS-1 data, will deal with infant feeding determinants and consequences for child survival; the implications of child mortality for fertility and investments in surviving children; and fertility and intergenerational transfers. The empirical analyses will use multivariate statistical techniques, including hazard models with time-varying covariates.
