不同於以往採用數學測驗的性別刻板印象威脅研究，本研究採用男性不擅長的剪紙作業為實驗材料，並以男性為烙印團體成員。本研究同時將過去研究未探討的團體作業情境納入實驗考量，以探討在團體作業情境中，同一「刻板印象」可使參與者擔心自己表現不佳而降低表現（刻板印象威脅），但又可使參與者預期伙伴表現不佳而傾向付出更多心力（社會補償），則參與者是否因為付出更多心力而提升了原本受到刻板印象威脅影響所降低的表現。 本研究為2(刻板印象威脅：有 vs. 無)× 2（伙伴性別：男 vs. 女）× 2（伙伴表現：好 vs. 差）的受試者間設計。其中，在刻板印象威脅部分，採用直接告知「女性比男性擅長剪紙作業」的方式引發刻板印象威脅效果。在社會補償方面，分別以「伙伴男性」與「伙伴表現差」二種訊息來引發參與者社會補償。本研究之主要依變項為完成剪紙作業之作業速度與作業品質。本研究有效樣本為121位男性大學生。 研究結果發現，「伙伴表現」與「伙伴性別」二變項之效果皆未達顯著，顯示伙伴訊息未能引發社會補償效果；而針對社會補償是否能減除刻板印象威脅效果之考驗，亦未達顯著水準，因此以社會補償作為減除刻板印象威脅效果的可能性在本研究中未獲支持。然而，卻發現「伙伴表現」與「伙伴性別」的交互作用顯著，即當操弄的伙伴訊息與刻板印象「不一致」時，參與者表現比「一致」時更好，顯示「伙伴訊息」可能另有影響。本文將針對上述結果進行探討，並對研究限制與對未來研究之建議加以說明。 / Different from previous gender stereotype threat studies that use mathematics tests as tools, this study uses the manual task—paper cutting generally not familiar to males as experimental material and take them as stigmatized group numbers. This research take into consideration the group task (cooperation situation) to investigate whether participants’ additional efforts will compensate the degraded performance resulting from gender stereotype threat, in the situation that a stereotype can cause people to lower their effort as a result of their worry about their perceived poor performance (stereotype threat) and can also motivate them to make more effort thanks to their prediction of their partner’s poor performance (social compensation). The participants were randomly assigned to one of the eight conditions in a factorial design: 2 (stereotype threat: salience, not salience) × 2（partner’s gender: male, female）× 2（partner’s performance: good, bad）. To evoke stereotype threat effect, the experimenters directly told the participants that “Females are better than males in the manual task.” To evoke social compensation effect, researchers arranged “males” and “poor performance” information. Valid samples were 121 male undergraduates, and their reaction time and quality of the manual task were measured. Not as expected, the results do not support hypotheses. However, it has found out that “the interaction of partner’s gender and partner’s performance” was significant. That is, the participant whose partner’s information is consistent with stereotype performs better than the participant whose partner’s information was inconsistent with stereotype. The result suggests that partner’s information may have unexpected influence on participants’ performance. This article also discusses the results, explains the limitations and offers suggestions for future research.