This paper discusses problems researchers confront when intending to clarify the influences of relevant policies on child health in developed countries. It also elaborates how comparison of siblings within a family can contribute to c1arification of causal effects between policies and child health through controlling heterogeneity among observations and increasing the variations of variables with respect to policies. Utilizing a survey conducted by the N ational Health Research Institutes (NHRI) in year 2000 as the example, this paper further introduces a method to take advantage of household registration computer datasets to estimate the expected numbers of children of certain ages in a family and construct samples with certain age structures for conducting within-family comparisons. The age structure of the sample generated by this NHRI survey shows that this method is easy to operate and very helpful for predicting the age structure of the sample and estimating the budget for the survey.
