我國人口資料的蒐集是採用戶籍登記制度為主,定期發佈各級人口資料,此外亦固定進行每十年一次的戶口及住宅普查(簡稱戶口普查),另輔以各式的抽樣調查來符合特定需求。然而,戶口普查的對象主要為常住人口,往往與戶籍人口有一定程度的差異;戶籍資料未隨著人口流動及時更新,因而有著涵蓋率與準確性的問題,常造成「人籍不一致」的情形。另方面,常住人口具有反映人口分佈現狀的特性,但卻經常缺乏具時效性的客觀數據可比對。當利用大型調查資料來估算人口分佈時,經常因為戶籍人口與常住人口的差異,而使得推估出的數值有所差異。本研究利用「105年度全國客家人口暨語言基礎資料調查研究」的六萬五千多筆電話訪談資料,進行臺灣地區的常住人口推估,並比較不同地區層級(縣市、鄉鎮市區)所推估的人口分佈狀況與戶籍人口的差異。此外,我們也進一步檢視戶籍人口與常住人口的推估結果,在各地區層級所呈現的人口結構,同時也以資料所推估出的客家人口為例,討論其政策與調查實務上的意涵,以做為未來執行大型抽樣調查與常住人口推估之參考;然而在缺乏及時更新與較細層級之常住人口進行資料檢視的條件下,以調查資料推估母體的作法仍須謹慎為之。
Population statistics in Taiwan are derived from the husehold registration system and released regularly. The household registration, however, suffers from coverage errors because of incomplete or inaccurate information, residential mobility, and updating delays. The inconsistency between registered and actual, current residential addresses is an important issue in population statistics. On the other hand, census and large-scale sample surveys are conducted to estimate the de jure population as a whole or for certain demographic groups. De jure population has the advantage of reflecting the most up-to-date population distribution, while residential mobility also leads to the lack of a gold standard. Using more than 65 thousand cases from the 2016 National Hakka Population and Language Survey (NHPLS), this study aims to estimate the registered and de jure populations at the county or city and borough or township levels. The inconsistency between the two population estimates is reported. As the purpose of the 2016 NHPLS is to estimate the distribution of Hakka population in Taiwan, results using registered and current residential address are reported to illustrate the inconsistency. Researchers need to be aware of the limitation when using survey results to estimate population when a gold standard is lacking. Based on our findings, implications for policy making and survey practices are discussed.