Abstract: Default positive rating rules automatically record unsubmitted consumer reviews as positive. While this practice significantly increases the overall positive rating rate, the resulting information bias diminishes the reference value of online review systems for building trust between buyers and sellers. To address this issue, this paper develops a latent variable model based on online review systems of Chinese e-commerce platforms. The model effectively utilizes actual transaction data to identify and reveal the relationship between consumers’ true satisfaction and biased online reviews. Combining survey data on consumers’ review habits, it employs maximum likelihood estimation to infer consumers’ true satisfaction and proposes corresponding bias correction methods. The proposed latent variable model quantifies the degree of bias caused by default positive rating rules in current online review systems and reveals the true evaluation tendencies of silent consumers. These findings provide theoretical guidance for optimizing e-commerce platform review systems and supporting consumer purchase decisions.

Key words: default positive rating policies, online reviews, online review system, bias

摘要： 默认好评规则将消费者未及时撰写的评论直接计为好评，这种做法虽然显著提高商品好评率，但其造成的信息偏差削弱了在线评论系统在买卖双方信任过程中的参考价值。针对该问题，本研究基于中国电商平台的在线评论系统，构建了一个潜变量模型。该模型能有效利用实际交易数据，识别并揭示消费者真实满意度与平台上存在偏差的在线评论之间的关系。研究结合问卷调研分析了消费者在线评论反馈习惯，通过最大似然估计推断出消费者的真实满意度，并提出了相应的偏差修正方法。所构建的潜变量模型量化了现有在线评论系统中默认好评规则的偏差程度，揭示了未主动评价的“沉默”消费者的真实评价倾向，为电商平台优化在线评论系统设计与辅助消费者做出更明智的购买决策提供了理论参考。

关键词: 默认好评规则, 在线评论, 在线评论系统, 偏差