近年來霸凌事件逐漸受到重視，然而另一種較隱微以操弄人際關係為手段的攻擊方式（即“關係攻擊”），近期才逐漸納入攻擊行為的討論。目前多數研究以性別來比較兩性間的攻擊行為差異，但鮮少說明其原因，而本研究則嘗試以社會文化脈絡的角度來探討造成兩性攻擊行為差異的可能原因。由於社會中會傳遞性別角色規範，使得個體在社會化過程中習得、遵守相對應的性別特質與行為，因此本研究將探討性別、性別角色以及典型性別特質與攻擊行為間的關係。本研究在前置研究中證實受試者對於攻擊行為確實存有性別角色刻板印象，並針對各研究工具的信效度進行檢核與修訂後，方進行正式研究。正式研究一，以問卷調查法收集台灣四個地區（北、中、南、東部地區）的國中八年級生與大學生，國中生共回收251份有效問卷（男性113人；有效率72.5%）；大學生則有249份有效問卷（男性92人；有效率82.2%），所有研究資料以SPSS14.0進行單因子變異數分析、迴歸分析與中介變項分析。結果發現：1.不論國中或大學階段，男性外顯攻擊比女性來得高（all ps<.043），而關係攻擊則無性別差異（all ps>.440）；2.典型男性特質可以預測國中男性與女性的外顯攻擊，亦可預測大學女性的外顯攻擊（all ps<.029）；3.典型男性特質可以預測國中女性的關係攻擊行為（p<.01）；4.典型女性特質無法預測任何攻擊行為；5.經中介效果檢定，典型男性特質不論在國中或大學階段，皆是性別預測外顯攻擊的部分中介變項。正式研究二，為探討性別角色對攻擊行為的影響，以國中生113人（男性55人）與大學生120人（男性62人），採二因子完全受試者間設計：2實驗操弄（激發自我性別意識vs.未激發自我性別意識）× 2性別（男性vs.女性）分別探討國中生與大學生樣本。結果發現：1.國中生部分（1）在外顯攻擊上，實驗操弄與性別的交互作用未達顯著差異（p=.974），然而，兩者皆有主要效果（all ps<.034）。實驗組（M=23.38）分數較控制組（M=29.76）低；男性分數（M=32.37）較女性（M=21.23）高。（2）在關係攻擊上，實驗操弄與性別無交互作用（p=.673），且皆無主要效果（all ps>.236），但值得注意的是，實驗組（M=43.71）與控制組（M=46.98）間雖未達顯著差異，但以平均數來看有下降的趨勢。2.大學生部分（1）在外顯攻擊上，實驗操弄與性別無交互作用（p=.466），但皆有主要效果（all ps<.028）。實驗組（M=22.72）分數較控制組（M=19.11）高；男性分數（M=23.55）較女性（M=17.98）高。（2）在關係攻擊上，實驗操弄與性別無交互作用（p=.849），且皆無主要效果（all ps>.076），但值得注意的是，實驗組（M=48.25）與控制組（M=43.21）間雖未達顯著差異，但以平均數來看有升高的趨勢。總結來說，男性外顯攻擊穩定地比女性高，但本研究也發現典型男性特質是影響攻擊行為的重要關鍵，表示女性攻擊性不一定比男性低。這樣的結果也說明了極有可能因社會化過程中男性性別角色被要求符合典型男性特質，故使得男性更展現外顯攻擊行為。然而，以實驗法操弄的結果並未符合預期，針對這樣的結果表示可能操弄過程中激發了自我性別意識之外的概念，這也顯示攻擊行為的複雜性。更多詳細資料與討論請詳閱本文。
In recent years, people have become more concerned about overt bullying behavior. However, a more covert form of aggressive behavior (e.g. relational aggression) was often overlooked. In addition, most previous research simply pointed out the gender differences in aggressive behavior without providing clear explanation. This study aims at exploring the causes of gender differences in aggressive behavior from a social contextual perspective. There are genders role that the norms may expect individuals to follow. Thus, individuals learn to compliance with the corresponding gender characteristice and behavior. This study will investigate the relationship among gender, gender roles, gender charateristic, and aggressive behavior in different age level. In pilot study, this study confirmed that participants did have perception of gender role stereotype, and the reliability and validity of study materials were also checked. In study 1, this study 8th grade students and undergraduates; 251 valid data from junior high school students were collected (113 males; efficiency of 72.5%); 249 valid data were obtained from undergraduates (92 males; efficiency 82.2%), all data analyzed by one-way ANOVA, regression analysis and mediator analysis. The results showed that: (1) Regardless of the age level, male overt aggression was higher than female (all ps < .043), whereas no gender differences in relational aggression (all ps> .440). (2) The typical male characteristics could predict the junior high school student and undergraduate female overt aggression (all ps <.029). (3) The typical male characteristics could predict the junior female relational aggression (p<.01). (4) The typical female characteristics couldn't predict overt and relational aggression. (5)The regardless age level, typical male characteristics as a mediator of the relationship between gender and overt aggression. In study 2, this study used 2 (gender: male vs. female) x 2 (conditions: control vs. gender consciousness activated) independent design to investigate the effect of gender roles on aggressive behavior in each age level. There were 113 junior high school students (55 males) and 120 undergraduates (62amles) partienpating in this study. The results showed that: (1) Overt aggression in junior high school sample, gender has no interaction with condition (p=.974), but both of them had main effect (all ps <.034), experimental group (M =23.38) scores was lower than control group (M=29.76); male (M=32.37) was higher than female (M=21.23). (2) Relational aggression in junior high school sample, gender was no interaction with condition (p=.673), and both of them had no main effect (all ps>.236). (3) Overt aggression in undergraduate sample, gender has no interaction with condition (p=.466), but both of them had main effect (all ps<.028), experimental group (M=22.72) scores was higher than control group (M= M=19.11); male (M=23.55) was higher than female(M=17.98). (4) Relational aggression in undergraduate sample, gender has no interaction with condition (p=.849), and both of them had no main effect (all ps>.076). In conclusion, male take overt aggression stably more than female; however, we found that typical male characteristics is an important factor to influence aggressive behavior, meaning that female are not always less aggression than male. This result also explained that male were requested to comply with a standard (gender role) in the process of socialization, then conform to typical male characteristices, which make male takes more overt aggression. More details and discussion is in this article.