夫妻互动行为是影响婚姻质量最近、最直接的一条路径。然而对于积极和消极互动行为对婚姻质量的预测作用，研究并未得出一致结论；积极和消极行为之间的相互作用机制对婚姻质量的影响还尚不清晰；积极和消极行为是如何通过即时效应，扩散至对总体婚姻质量评价的路径也不明确。本研究采用“问题解决”和“社会支持”的观察研究范式，对106对城市新婚夫妻进行录像观察，采用修订后的爱荷华家庭互动编码体系（IFIRS），对夫妻互动行为进行编码，并采用问卷法调查夫妻互动的即时满意度和对婚姻质量的总体评价。通过相互关联的三个研究，对IFIRS编码体系在中国文化下的适用性，夫妻的互动特点及互动模式、夫妻互动行为的相互作用机制对婚姻质量的影响、夫妻互动行为的即时效应和一般效应路径机制进行探索，得出以下结论：（1）修订后的IFIRS具有良好的编码者一致性信度、内容效度和结构效度。编码可以聚为三个因素：积极情绪、消极行为和积极卷入。（2）总体上看，妻子在任何情境任何话题中都比丈夫有更多敌意、质问等消极行为，有更多倾听反应、有效沟通、问题解决等积极卷入行为；夫妻双方在积极情绪上没有差别。在问题解决中，大部分丈夫以“附和”的方式与妻子互动，而大部分妻子以指责和质问的方式与丈夫互动。在社会支持互动中，“温暖支持型”的丈夫最多，而妻子更多以“理智”的方式提供支持。“温暖支持型”的夫妻婚姻质量最高，“敌意-回避型”夫妻的婚姻质量最低。（3） 丈夫的积极卷入在各个互动中均能够稳定地正向预测丈夫的婚姻质量，且在丈夫问题解决和妻子社会支持话题中还具有客体效应。丈夫的消极行为和积极情绪只在妻子的问题解决中具有客体效应；妻子的积极情绪只在丈夫的问题解决情境中具有主体效应；妻子的消极行为和积极卷入在任何情境下对婚姻质量均无预测力。（4）丈夫的积极卷入在问题解决中均可以缓解丈夫的消极行为对夫妻婚姻质量的破坏力。当积极卷入达到一定临界值时，丈夫消极行为的破坏作用消失，甚至还可以产生积极影响。在妻子的社会支持中，如果丈夫的积极卷入较低，妻子的积极情绪反而负向预测丈夫的婚姻质量，妻子的消极行为负向预测妻子的婚姻质量。然而，随着丈夫积极卷入的增多，这种破坏力消失甚至被转化。（5）消极行为对于婚姻关系仅具有即时效应，它的破坏力不会扩散至夫妻对婚姻质量的总体评价；而丈夫的积极卷入，不仅能通过即时效应对婚姻质量产生影响，且在控制了即时效应后，依然可以直接产生一般效应。“丈夫的积极卷入”对夫妻的婚姻质量具有直接效应和间接过滤器作用。

Amongst a multitude of factors, couples’ interactive behavior most directly significantly affects marital quality. Despite its’ importance however, researchers have found inconsistent results regarding how positive and negative behaviors affect marital quality, particularly concerning which is the more powerful predictor. It is unclear how the interaction between positive and negative behavior affects marital quality, as well as how the immediate effects of positive and negative behavior affect marital quality in general. Furthermore, most observational studies of couples’ interactive behavior have been conducted in Western contexts. This study explored the important question of how Chinese couples resolve marital conflicts and support each other in daily life, the different effects of positive and negative behaviors in these two contexts, and the mechanisms through which these behaviors affected both their own and their partner’s marital quality. Under laboratory conditions, 106 newlywed couples’ behavior was observed in both problem solving and social support contexts. All couples were also asked to report their satisfaction with the interaction immediately after conversing as well as their general marital quality. The Revised Chinese Version of Iowa Family Interaction Ration Scale (IFIRS) was used to rate couples’ interactive behavior. The results were as follows: 1) The Revised Chinese Version of Iowa Family Interaction Ration Scale (IFIRS) demonstrated good inter-coder reliability and good content validity and construct validity. From IFIRS results, this study deduced three factors: positive emotion, negative behavior, and positive engagement.2) In general, wives were more hostile and critical, as well as more responsive listeners and communicators exhibiting more effective problem solving and less avoidance behaviors than their husbands. This study found no difference in positive emotions exhibited by husbands and wives. In problem solving contexts, husbands were more likely to passively follow their wives’ lead, whereas wives were more likely to be critical and interrogative towards their husbands. In social support contexts, most husbands were supportive and warm with positive emotions, whereas wives were more likely focus on rational problem solving with neutral emotions and moods.3) Among all behaviors, husbands’ positive engagement was the most powerful predictor, resulting in actor effects in all interactions and partner effects in husband problem solving and wife social support contexts. Husbands’ negative behavior and positive emotion only resulted in partner effects in wife problem solving contexts. Wives’ positive emotion was only predictive of wives’ own marital quality in husband problem solving contexts. Neither negative behavior nor positive emotion had any actor or partner effects.4) Concerning interactive mechanisms, husbands’ positive engagement buffered the effects of negative behavior on marital quality in problem solving contexts. When a husband’s positive engagement reached a certain level, the negative effects of his negative behavior on his own and partner’s marital quality disappeared and even transformed into positive effects. In social support contexts, if the husband’s positive engagement was low, not only did the wife’s negative behavior negatively affect her marital quality, even the wife’s positive emotions had negative effects on her husbands’ marital quality. However, if the husband’s positive engagement in the conversation reached a medium to high level, these negative effects diminished and could even be transformed into positive effects.5) Negative behaviors only functioned on the immediate level, namely interaction satisfaction immediately after conversing, whereas husbands’ positive engagement functioned at both the immediate and general level. In other words, husbands’ positive engagement can influence marital quality by first affecting interaction satisfaction immediately after conversing, and, after controlling for specific effects, marital quality in general. Overall, husbands’ positive engagement was the most important predictor for couples’ marital quality and an important filter for the negative effects of other behaviors. KEY WORDS: couple interaction behavior, problem solving, social support, marital quality, observational paradigm.