我国大型抽样调查一般以省级为单位展开,造成市县级的样本量不足甚至为零的情况,难以满足政府多级决策需求。本文围绕多级需求难题,介绍了基于设计推断的方法和基于模型估计的方法,认为基于模型的小区域估计方法摆脱了对样本量的依赖,一定程度上解决了多级需求难题,具有重要的理论和实践意义。 Large-scale sample surveys in our country generally start at the provincial level, resulting in a sample size of less than or even zero at city and county level, which makes it difficult to meet the government’s multi-level decision-making needs. This article focuses on the problem of multi-level requirements, introduces the method based on the inference of the design and the method based on the model estimation, and considers the model-based small region estimation method to get rid of the dependence on the sample size, to some extent to solve the multi- Theoretical and practical significance.