id "HEADER" 的內容放在這裡

　　在台灣，與子女同居是老年人的主要居住型態，60%以上的老年人與子女同居安養晚年。老年人的居住型態直接關聯著其生平抉擇與境遇，本研究的目的即在於深入解析老年人與子女同居的動態過程，探討老年人的個人特質及其與子女同居機率間的關連。研究使用行政院主計處與內政部合作舉辦的九次老人狀況調查資料，使用年齡分組迴歸探討老年人的個人特質變項對於其與子女同居機率的影響，企圖說明老年人與子女同居機率隨著老年人的年齡、性別、健康狀況、是否有偶變動而有不同的情形，進一步建構老年人與子女同居機率的最適模型。我們認為能清楚說明老年人與子女同居動態過程的迫切模型為年齡、性別、是否有偶的主效果，配合健康狀況與年齡、性別及是否有偶的交互作用，共58個參數的模型。
　　研究結果顯示隨著年齡增加，老年人與子女同居的機率呈拋物線的形狀，且為統計上顯著的效果；結果顯示老年女性與子女同居的機率大於老年男性，即使在控制其他變項後，台灣老年女性依舊傾向於與子女同居就養；進一步考量年齡與其他變項的交互作用，性別差異顯得複雜且因年齡而不同，老年男性較女性不傾向與子女同居的特性隨著年事增高而減退，而有偶男性較有偶女性或無偶男性更傾向與子女同居的特性也隨著年事增高而減退；我們發現是否有偶為影響老年人與子女同居的關鍵因素，愈是老年無偶與子女同居的機率則愈高。文獻對於健康狀況如何影響老年人居住安排，意見相當分歧。在控制其他變項的條件下，本研究健康狀況的年齡別參數均不顯著，顯然係透過與是否有偶在年齡軸上的交互作用來影響老年人與子女同居的機率。

　　More than 60% of the elderly in Taiwan live with adult children after the retirement age. The living arrangement is contingent on the choices they made and the circumstances they faced during the life time. This paper aims at exploring the dynamics of the elderly living arrangement, with a focus on the characteristics of the elderly living with adult children. Returns to a total of 9 batches of "Survey on the Elderly Conditions in Taiwan" during the period of 1986-2002 were analyzed and the effects of sex, age, health condition, and marital status of the elderly on the probability examined, while the changes along the axis of age emphatically noted.
　　We hypothesized that probability of living with adult children is a function of the sex, age, health condition, and marital status of the elderly. While the health condition and marital status change with age of the elderly, it brings about a change in the probability. Our results indicate that sex, age, and marital status are three key factors to the dynamics of living with adult children at old age. Among the elderly in Taiwan, men with spouse present are more inclined to live with adult children than those without and conversely women without spouse present are more inclined to live with adult children than those with spouse present. In general, due to a higher mortality for men, women show a higher tendency to live with adult children at old age. The sex differential tends to diminish as the elderly progress to higher age, however.
　　The effect of health condition on the probability appears to be less defined. Statistically controlling the influence of age, sex, and marital status, the effect of health condition on the probability becomes insignificant. Health condition can assert an effect on the probability only through its interaction with marital status at very old age.