台灣老年人口年齡結構快速高齡化將面臨人口結構變遷所帶來的衝擊，人口遷移的分布更顯得是需要注重的議題。而人口的遷徙不僅可以調節區域間的資源配置，也可以看出產業的發展變化以及城市的興衰。本研究將採用近三次包含2002、2007、2012年共三年度由行政院主計處調查的「國內遷徙調查」資料，關注在個體分析不同個人屬性及家戶型態對遷徙機率的影響。 本文研究模型使用logistic迴歸模型，探討家戶的戶長個人特徵，以及家庭的型態探討各項特性對遷移機率的關聯性；由於遷移次數資料因為零值過多而呈現過度離散及零膨脹的疑慮，因此本文除了使用logistic模式以外，也加入零膨脹卜瓦松迴歸及零膨脹負二項迴歸模型加以比較。 實證結果顯示，家戶的總收入越高對遷移機率及遷移次數有顯著正相關；戶長居住在六都和居住型態屬於配住對遷移機率有顯著正相關；戶長的年齡、有在工作、擁有自宅對遷移機率及次數有顯著負相關。在過往研究指出男性、未婚、高教育程度對遷移的機率較高，但研究結果顯示上述因素對於戶長的遷移機率及次數並無顯著影響。另外，考量戶長的職業屬於白領工作人員及藍領工作人員，以及家中小孩人口數及老年人口數是否對遷移機率及次數有影響，其估計結果顯示不顯著。 ;While the rapidly aging population in Taiwan would be affected by the demographic change, population migration would be a crucial issue that deserved concerned. Population migration helps allocate resources among areas and reflects the transition of industrial and urban development. The study will apply the “Internal Migration Survey” conducted by DGBAS (The Directorate General of Budget, Accounting, and Statistics) in 2002, 2007, and 2012 to understand how personal and household attributes affect migration. The study applied a logistic regression model to identify the correlation among personal attributes for head of household, family attributes, and possibilities of migration. To consider the problems of too many zero and over-dispersion, zero-inflated Poisson regression & zero-inflated negative binomial regression are employed to estimate migration likelihood and frequency. Results have indicated that high household income has significant positive correlation to possibility and frequency of migration, while household head’s residency in six special municipalities and dormitory allocation have significant possibility correlation to possibility of migration. Age of household head, employed, household ownership have significant negative correlation to possibility and frequency of migration. Historical research has shown a higher possibility of migration for males, single and the highly-educated; however, we do not find a significant impact of the above factors on possibility and frequency of migration in our research. In addition, the results also suggested that whether the household head is blue or while-collar worker, number of children, and number of elders in the household, have no significant correlation to the possibility and frequency of migration.
