研究目的：本研究欲比較關係特質取向(有指定對象，包括父、母、愛情)與個人特質取向(無特定對象)等兩種測量成人依附風格的差異性，並依據成人依附光譜假說與依附轉換模式，探討所得測量結果之間一維光譜的關連性，以及愛情依附與個人特質依附風格與親子依附風格之相關。研究方法：研究樣本共計185名大學生，男性59人，女性125人，平均年齡約19.95歲。以Bartholomew成人依附短文描述句為研究工具，請受試者依不同對象進行歸類，並以卡方檢定與MDS分析。研究結果：1.有無特定對象的依附風格歸類結果具差異性；2.光譜假說未獲得支持，但愛情依附風格與個人特質依附風格兩者分別與親子依附風格的分類結果有關連。研究結論：若依附測量需針對特定對象卻以無特定對象方式進行測量，會有高估不安全依附歸類的現象；個人特質上的不安全依附歸類結果與親子的不安全依附有關，愛情關係上的安全依附歸類結果則與親子的安全依附有關。

Purpose: Bartholomew has integrated Bowlby's theory of internal working models in a four category classification scheme. Four prototypical adult attachment styles were measured in this study, including secure, preoccupied, fearful, and dismissing attachment styles. However, in the past, most of adult attachment researches ignored the difference between with-specific-object and without-specific-object measures. Therefore, the past results of adult attachment category had been criticized their over-evaluated or underevaluated trend. This study attempted to compare the difference between the two ways of measuring the adult attachment styles: relationship character approach (with specific objects, e.g., father, mother, or a dating partner) and individual character approach (without specific objects). Based on the hypothetical continuum of adult attachment (Bartholomew & Shaver, 1998) and the model of attachment transfer processes (Hazan & Shaver, 1994), this study also tried to investigate the one-way continuum correlation between the results of the different measurements, and the correlation among the results from dating relationship attachment styles, individual attachment styles, and parental attachment styles. The figure of hypothetical continuum of adult attachment indicated that when appropriate comparisons were drawn, one would find considerable evidence for convergence across various measures of adult attachment. This study supposed that the results of the measurements of adult attachment category with specific objects would be different from the measurements of adult attachment category without specific objects. Besides, according to the model of attachment transfer processes, the results of category in parental attachment styles would be positive correlated with the results of category in dating relationship attachment styles, and the results of category in parental attachment styles would be also positive correlated with the results of category in personal attachment styles. Methods: The study sample consisted of 185 college students, including 59 males and 125 females, and their average age was 19.95 years. None of them were from single family, and 61.6% students were living with their parents. Bartholomew's paragraph-long description form was used to categorize adult attachment styles. In addition, we conducted chi-square test and MDS to analyze the data. Results: First, there were differences between the results of the measurements of adult attachment category with specific objects and the measurements of adult attachment without specific objects. Second, the hypothetical continuum of adult attachment measures was not supported by MDS. Finally, there were positive correlations between dating relationship attachment styles and parental attachment styles, and between personal attachment styles and parental attachment styles. Conclusions: The article concluded with a discussion of unclear questions about adult attachment that rose from various measurement perspectives. First, this study supported that insecure parental attachment would be over-evaluated if without specific-object measurement of attachment styles was used to measure with-specific-object attachment styles. There is no 'best choice' between the two measured approaches. If one adult attachment research investigates specific situation (e.g., love relationship, peer relationship, mentor relationship, or parental relationship), the research will suggest that the way of measuring the adult attachment styles should use relationship character approach (with specific objects). Second, according to the model of attachment transfer processes, attachment style in personal character and in dating relationship were both positive correlate with attachment styles in parental relationships. More specifically, the insecure attachment style in personal character was positive related to the insecure attachment style in parental attachment, and the secure style in dating relationship was positive related to the secure style in parental attachment. In addition, implication for adult attachment measures in adult relationships was discussed. Research limitation was also explicated.