隨著電子口碑（eWOM）成為消費者購買決策之重要資訊來源，現有研究多聚焦 於負面評論之直接說服效果，對其背後心理歷程之探討仍有不足。本研究整合情緒傳染理論（Hatfield et al.， 1993）、結構平衡理論（Heider， 1946）與認知失調理論（Festinger， 1957；Elliot & Devine， 1994），以「心理失衡」（Psychological Disequilibrium）為核心中介構念，建構「負評比例→心理失衡→品牌態度後測」之因果模型，並引入產業類別（餐飲業 vs. 電子產品業）作為調節變數。 本研究採2（產業類別）×3（負評比例：20%、50%、80%）between-subjects實驗設計，以虛擬品牌（Nerra／Zevon）為刺激材料，對160名受試者進行問卷施測，並以SmartPLS 4 進行二階PLS-SEM分析（拔靴法，N = 10，000）。 假說驗證結果：H1獲得支持，負評比例對品牌態度後測具有顯著負向直接影響 （β = −0.369，p < 0.001，直接效果模型R² = 0.327）； H2 獲得支持，負評比例正向影響心理失衡（β = 0.531，p < 0.001，R² = 0.282）； H3 獲得支持，心理失衡負向影響品牌態度後測（β = −0.240，p = 0.004）； H4獲得支持，心理失衡具有部分中介效果（間接效果β = −0.127，p = 0.006），整體R² = 0.541；H5 獲得支持，產業類別之調節效果達統計顯著（β = −0.318，p = 0.011），搜尋型商品（電子產品業）受負評衝擊顯著強於體驗型商品（餐飲業）。 本研究之貢獻在於：（一）首次整合三大心理理論建構心理失衡構念，豐富eWOM 心理機制研究；（二）實證確認心理失衡之部分中介角色，並提供二階PLS-SEM操作化範例；（三）H5之方向性發現為eWOM產品類別調節效果研究提供重要理論反思。管理上建議企業積極監測負評比例，並優先強化電子產品類別之負評因應策略。;As electronic word-of-mouth (eWOM) has become a critical information source in consumer decision-making， existing research has predominantly focused on its direct persuasive effects while leaving the underlying psychological processes theoretically underexplored. Drawing on Emotional Contagion Theory (Hatfield et al.， 1993)， Structural Balance Theory (Heider， 1946)， and Cognitive Dissonance Theory (Festinger， 1957; Elliot & Devine， 1994)， this study constructs a causal model of "negative review ratio → psychological disequilibrium → brand attitude (post)，" incorporating industry type (food service vs. consumer electronics) as a moderating variable. A 2 (industry) × 3 (negative review ratio: 20%， 50%， 80%) between-subjects experimental design was employed， using fictitious brands (Nerra and Zevon) as stimuli， with 160 participants randomly assigned to one of six conditions. Data were analyzed using second-order PLS-SEM in SmartPLS 4 with bootstrapping (N = 10，000). Hypothesis testing yielded the following results. H1 was supported: negative review ratio exerted a significant negative direct effect on brand attitude (β = −.369， p < .001; direct-effect model R² = .327). H2 was supported: negative review ratio positively predicted psychological disequilibrium (β = .531， p < .001， R² = .282). H3 was supported: psychological disequilibrium negatively predicted brand attitude (β = −.240， p = .004). H4 was supported: psychological disequilibrium served as a partial mediator (indirect effect β = −.127， p = .006; overall R² = .541). H5 was supported: industry type significantly moderated the relationship (β = −.318， p = .011)， with negative reviews exerting a stronger impact on brand attitude in the search-goods condition (consumer electronics) than in the experience-goods condition (food service). This study contributes by (1) integrating three psychological theories into a novel psychological disequilibrium framework， enriching the eWOM mechanism literature; (2) empirically establishing psychological disequilibrium as a partial mediator and providing a methodological template for second-order PLS-SEM operationalization; and (3) offering important theoretical nuance on product-type moderation in eWOM research through the unexpected direction of the H5 moderation effect. Managerially， firms are advised to actively monitor negative review ratios and prioritize eWOM management for high-involvement categories such as consumer electronics.
