Namekata T, Levy P S, O'Rourke T W
Public Health Rep. 1975 Nov-Dec;90(6):532-8.
A method of synthetic estimation of health characteristics for local areas, devised by the National Center of Health Statistics, was evaluated in a recent study. In the method, local data on population are combined with national data on a given health characteristic to produce an indirect estimate of that characteristic. The health characteristic selected in the study was that of complete and partial work loss disability. Therefore, synthetic estimates of complete and partial work loss disability were calculated for each State by combining the estimated rates of such disability for the United States, specific to a set of demographic domains, with the data relating to the distribution of each State into this set of demographic domains. The synthetic estimates of complete and partial work loss disability for each State were then compared with the direct estimates available from the 1970 decennial census. For partial work loss disability, agreement between the synthetic and the direct estimates, as judged by the median percentage absolute difference, was fairly good; for complete work loss disability, agreement was rather poor. On the other hand, the correlation between the synthetic and the direct estimates was higher for complete work loss disability than for partial, partly because the synthetic estimates had a greater spread for complete work disability than for partial. The skewness of the distributions of the squared differences indicates that the evaluation based on median percentage absolute differences was more descriptive than the one based on mean square errors.
美国国家卫生统计中心设计的一种用于对局部地区健康特征进行综合估计的方法，在最近的一项研究中得到了评估。在该方法中，将局部地区的人口数据与关于特定健康特征的全国数据相结合，以得出该特征的间接估计值。该研究中选取的健康特征是完全和部分工作丧失性残疾。因此，通过将美国特定于一组人口统计学领域的此类残疾估计率与每个州在这组人口统计学领域中的分布数据相结合，计算出每个州完全和部分工作丧失性残疾的综合估计值。然后，将每个州完全和部分工作丧失性残疾的综合估计值与1970年十年一次人口普查可得的直接估计值进行比较。对于部分工作丧失性残疾，根据中位数绝对差异百分比判断，综合估计值与直接估计值之间的一致性相当好；对于完全工作丧失性残疾，一致性则相当差。另一方面，完全工作丧失性残疾的综合估计值与直接估计值之间的相关性高于部分工作丧失性残疾，部分原因是完全工作残疾的综合估计值比部分工作残疾的综合估计值分布范围更广。平方差分布的偏度表明，基于中位数绝对差异百分比的评估比基于均方误差的评估更具描述性。
