基於回歸模型探討臺中市婚姻人口及空間效應—以 2011 年及 2020 年為例

Abstract

現存國內外婚姻研究多數傾向探討婚姻與社會結構的關聯性，並僅限於「群體」類別，而忽略對「空間」的論述，例如家庭收入高低、夫婦教育程度、初婚年齡、育兒數量、宗教信仰與種族等等。本研究立足於過往婚姻文獻研究結果，利用空間統計方式探討婚姻人口、社會與空間的交互關係，以臺中市為研究區域，探討升格為直轄市的第一年（2011年）與第十年（2020年）的婚姻人口與空間變化，並投入相關變數進行探討。研究發現兩年度台中市婚姻人口冷區集中於原台中舊市區，熱區集中在郊區至沿海一帶；透過逐步法篩選投入變數，兩年度中的扶幼比、自然增加率與所得中位數皆與婚姻人口呈現正相關；相對負相關的部分為女性變數，包含女性高等教育比、女性離婚比與女性喪偶比；變數投入SLM後，亦發現婚姻人口存在空間鄰近效應，並透過Breusch-Pagan檢定亦再進行GWR，將其係數執行K-means演算法後，發現海線一帶女性變數相較市區影響甚多，皆為負向關；至於扶幼比、所得中位數與自然增加率變數，無論區域皆呈現正相關。升格為直轄市後的臺中市，人口增加的同時各區戶量卻呈現逐年降低，面臨少子化與老年化現象，即使經濟所得中位數在2011年有強烈正向影響，但在2020卻明顯減少，表明經濟與婚姻間的關聯性逐漸示弱。整體而言，沿海地區相對都市地區婚姻現象更明顯，近期主要婚姻人口集中在西屯至沙鹿一帶，推測與周圍工業及科學園區帶來人口群聚有關；本研究將臺中市婚姻人口分為四大群集，證實整體婚姻人口亦有地域之分。Existing domestic and international marriage studies mostly tend to explore the relationship between marriage and social structure, focusing only on"group" categories while neglecting the discussion of "spatial" aspects, such as household income, couple's educational level, age at first marriage, number of children, religious beliefs, and ethnicity. This study builds upon the results of previous marriage literature and utilizes spatial statistical methods to investigate the interactions between the married population, society, and space, using Taichung City as the research area. It explores the changes in the married population and spatialdistribution in the first year (2011) and the tenth year (2020) after Taichung was upgraded to a special municipality, and analyzes related variables.The study finds that in both years, the cold spots of the married population in Taichung are concentrated in the old urban areas of Taichung, while the hot spots are concentrated in suburban and coastal areas. Through stepwise selection of input variables, it is found that the child dependency ratio, natural increase rate, and median income are positively correlated with the married population in both years. On the contrary, female variables, including the ratio of women with higher education, female divorce rate, and female widowhood rate, are negatively correlated. After inputting variables into the Spatial Lag Model (SLM), it is also found that the married population exhibits spatial proximity effects. Through the Breusch-Pagan test and subsequent Geographically Weighted Regression (GWR), the coefficients were analyzed using the K-means algorithm. It was found that female variables in the coastal areas have a greater negative impact compared to urban areas, while the child dependency ratio, median income, and natural increase rate variables are positively correlated regardless of the region.After being upgraded to a special municipality, the population of Taichung City increased, but the number of households in each district decreased annually, facing low birth rates and an aging population. Even though the median income had a strong positive impact in 2011, it significantly decreased by 2020, indicating that the relationship between the economy and marriage is gradually weakening. Overall, the marriage phenomena are more evident in coastal areas compared to urban areas. Recently, the main married population has been concentrated from Xitun to Shalu, speculated to be related to population clustering brought by surrounding industrial and science parks. This study divides the married population of Taichung City into four major clusters, confirming that there are regional differences in the overall married population