本文采用计算器随机数字法,进行二阶段整群随机抽样,比较山东省地区及省级5％及10％人口比例抽样的代表性,全省范围内以县为单位和以乡(原公社)为单位抽样的代表性,结果表明:以乡为单位5％与10％比例抽样各种癌亡率与总体的死因顺位均无显著差异,且以乡为单位10％抽样效果明显优于以县为单位抽样。 This paper uses the random number method of the calculator to carry out two-stage cluster random sampling to compare the representativeness of the sample of 5% and 10% of the population in Shandong Province and the province. The county is the unit and the county (formerly the commune) For the sample representative of the unit, the results show that there are no significant differences in the rate of sampling all cancer rates between the 5% and 10% of township units and the overall cause of death, and the sampling effect of 10% in the township unit is significantly better than The county samples for the unit.