有鑑於長期以來「台灣社會變遷基本調查」於調查執行時，所據以進行抽樣分層之羅啟宏(1992)七個鄉鎮發展類型，已無法實質反映出近幾年來因社會與經濟變遷所造成的各鄉鎮市區差距程度，本研究於2005年利用最新的人文區位資料重新進行抽樣分層的分類，將台灣的鄉鎮市區分爲七個集群，此新抽樣分層己應用在2005年執行的「台灣社會變遷基本調查」第五期第一次調查訪問中。爲求實務調查之便利性，在調查執行時合併了其中的兩個集群（屬於低度開發的第六層與第七層），而以六個集群作爲抽樣分層架構。爲了瞭解這六個集群的層級效果是否顯著，本研究也進一步整併了「台灣社會變遷基本調查」第五期第一次與第二次的四份調查資料，利用與鄉鎮市區發展差異具有高度相關的人口基本變項，如年齡層、教育程度、產業結構、職業層級及個人月收入等五項進行檢測。從卡方檢定與變異數分析結果顯示，雖然本研究的分層檢測礙於調查資料的限制，僅進行六個集群分層，然前述之人口基本變項在各集群間確實達到統計上的顯著差異，意味著這六個集群分層具有高度的區辨力，相當適合作爲國內大型調查在抽樣設計上之應用參考。

Lou's (1992) typology of seven strata from 309 townships has been adopted in the "Taiwan Social Change Survey" for a long time. In response to dramatic changes in socio-demographics in the past decade, however, such a typology no longer represents the differences among boroughs and townships nowadays. This paper employs updated socio-economic data derived from both the 2000 census and 2004 population statistics for stratification. A total of 358 boroughs and townships were regrouped into seven strata. The "Taiwan Social Change Survey" has first adopted this new typology as the stratification for sampling design in 2005. For the management of field work, the sixth and the seventh stratum were finally combined, resulting in six strata as the sampling scheme. To investigate the efficiency of this stratification, this paper combined and analyzed four datasets from the "Taiwan Social Change Survey" conducted in 2005 and 2006. Socio-demographic variables, including age, education, industrial structure, occupation, and personal income, were highly related to various levels of development among boroughs and townships and were, therefore, used to test the differences among strata. The results of chi-square test and ANOVA show that all socio-demographic variables are statistically significant, indicating that the differences among these six embedded strata are substantial and as expected as well.