本篇文章主要在驗證不同情況下的自我肯定偏誤是否存在，討論其強韌度與偏誤的方向，並探討自我肯定偏誤的心理機制—維持自尊的重要性。本篇文章分成兩個部分，第一部分包含三個不同的實驗驗證自我肯定偏誤的心理機制。 實驗一的結果發現當降低癌症發生的自我控制力並提供受測者癌症發生的參考值時，即能消除自我肯定偏誤的情形。實驗二將實驗一的情境延伸至四個不同事件—包含正、負面事件與高低控制力事件。實驗二結果發現，參考值的提供只有在低控事件會影響對自我的衡量，並降低偏誤的情形，但在高控事件中，參考值的提供卻會增加並惡化偏誤的情形。實驗三，加入了個人樂觀/悲觀的個性差異變數，結果顯示，樂觀者在提供了參考值之後，只有在低控事件願意依據參考值去修改對自我的衡量，而在高控事件，參考值對自我衡量不會產生影響；而悲觀者無論在高控或低控的事件，都願意依據參考值去修正對自我的衡量。整體而言，三個實驗告訴我們，造成自我肯定偏誤的原因乃是為了要維護自尊的理由。第二部分--實驗4則是以不同的方法驗證自我肯定偏誤的心理機制。和低控事件相比，高控事件代表和自尊攸關的事件，因此會引發出較大的自我肯定偏誤的情形。然而，即使是較不代表自尊的低控事件，自我肯定偏誤卻會因為衡量順序的不同，而產生不同的變化。先衡量”自我”時，自我肯定偏誤的情形並不明顯，但若先衡量”他人”時，則會產生自我肯定偏誤的情形。實驗4的結果證實了前三個實驗結果的強韌度，並再次反映了自我肯定偏誤乃是為提升並維持自尊而產生。 This paper studies the presence, resilience and direction of the self-positivity bias under various conditions to examine the role of self-esteem maintenance as an important antecedent for the bias. There are two parts in this paper. First part includs three studies which examining the reason underlining the self-positivity bias. Study 1 shows that perceptions of the controllability of cancer as well as presence of base-rate information can be contextually manipulated, and together eliminate the self-positivity bias in perceptions of the risk of cancer. Study 2 shows the same effects using four life events that differ in terms of valence and perceived controllability: base-rate information affects self-estimates for uncontrollable life events, reducing the self-positivity bias, but does not affect self-estimates for controllable events. In Study 3, we show that this effect only applies for optimistic individuals who fail to incorporate base-rate information to update self-perceptions for controllable events. Pessimists, on the other hand, use base-rates to update self-estimates irrespective of the controllability of the event. Overall, the pattern suggests that self-positivity is attenuated in conditions that implicate self-esteem. Another supplement study examines the underlying reason behind the self-positivity bias. In the second part of this paper, a single study --study 4 replicates the findings of previous three studies. As events perceived to be controllable implicate self-esteem more so than less controllable ones, they are more prone to self-positivity effects. On the other hand, as less controllable events do not implicate self-esteem, only when the order-of-elicitation cues comparative (versus absolute) judgments about the self, does the self-positivity effect emerge. When information about “self” is asked first, the bias is attenuated, but when others’ estimates are elicited prior to self-estimates, the bias re-emerges even for uncontrollable events. Implications for health marketing are offered.
