本研究的主要目的是探討台灣夫妻間的情感內涵，並以利翠珊於1999年所發展的「婚姻親密情感量表」為基礎，透過二組夫妻配對樣本進行一系列的對偶資料分析，包括驗證性因素分析、複核效度及效標關聯效度分析，減少量表的題項，並建構出一個具有信度與效度的簡短測量模式；此外，本研究運用actor-partner interdependence model (APIM)模式進一步檢視夫妻情感對彼此婚姻品質的影響。研究結果顯示，以理論與模式簡效性來考量，二階單因素模式優於其他模式，亦即婚姻情感的內涵可以分為感激、欣賞、親近及契合四個面向，而這四個面向又可統合成一個高階的潛在變項。經過信效度的評鑑，各個觀察變項及潛在變項具有良好的信度與效度，而複核效度分析亦顯示此測量模式不僅可以在一個樣本中成立，而且可以複製在其他樣本中，具有預測效度。最後，婚姻情感的各面向與本研究所選取的效標（婚姻滿意度、婚姻穩定度）皆有顯著的正相關，表示此婚姻情感量表具有效標關聯效度；且APIM模式顯示，情感對於自己的婚姻品質影響較大，對配偶的婚姻品質影響較弱。本研究未能證實的是婚姻情感的二階段式構念組成，亦即婚姻中的四種情感並無法再被歸類為「恩情」與「親密」兩個構念，一方面可能因為簡短量表較難完整地反映多層次構念所致，另一方面也指出了夫妻情感的不同面向與測量均仍有再精進的空間。總之，本土婚姻情感的內涵，可以用一個簡短而信效度良好的測量工具來捕捉，對於促進婚姻情感的後續研究頗有助益。

The purpose of the study was to develop an abbreviated form of the Chinese Marital Affection Scale (MAS) originally designed by Li in 1999. Two sets of samples with a total of 349 married couples living in northern Taiwan were used. The first sample set was used to confirm the factor structure of marital affection through confirmative factor analysis; this resulted in a scale of 4 factors: gratitude, appreciation, togetherness, and tacit understanding. Using dyadic analyses, a series of competing models was established and evaluated. Based on the principle of parsimony and previous theories, one second-order factor was retained as the model of best fit. The internal-component fit evaluation showed good reliability, convergent validity, and discriminate validity in both observed and latent variables. The second sample set was used to cross-validate and to confirm stability and predictive power for the hierarchical model with one second-order factor. The final procedure was to test the criterion-related validity of the brief form of the MAS. The results showed strong associations with 2 validity-related measures (marital satisfaction and marital stability). Moreover, actor-partner interdependence model showed that a person’s affection towards his or her spouse was more associated with his or her own perception of the quality of the marriage, and less related to the marital quality of the spouse. The study failed to confirm the hierarchical model with 2 second-order factors including en-qing and intimacy. Future research should verify the nature of the MAS in more diverse samples. In sum, the brief version is less time consuming and constitutes a reliable and valid alternative to the original MAS for clinicians and researchers.